Merge branch 'AUTOMATIC1111:master' into force-push-patch-13

javascript/extensions.js

@@ -0,0 +1,35 @@
+
+function extensions_apply(_, _){
+    disable = []
+    update = []
+    gradioApp().querySelectorAll('#extensions input[type="checkbox"]').forEach(function(x){
+        if(x.name.startsWith("enable_") && ! x.checked)
+            disable.push(x.name.substr(7))
+
+        if(x.name.startsWith("update_") && x.checked)
+            update.push(x.name.substr(7))
+    })
+
+    restart_reload()
+
+    return [JSON.stringify(disable), JSON.stringify(update)]
+}
+
+function extensions_check(){
+    gradioApp().querySelectorAll('#extensions .extension_status').forEach(function(x){
+        x.innerHTML = "Loading..."
+    })
+
+    return []
+}
+
+function install_extension_from_index(button, url){
+    button.disabled = "disabled"
+    button.value = "Installing..."
+
+    textarea = gradioApp().querySelector('#extension_to_install textarea')
+    textarea.value = url
+	textarea.dispatchEvent(new Event("input", { bubbles: true }))
+
+    gradioApp().querySelector('#install_extension_button').click()
+}

javascript/progressbar.js

@@ -3,8 +3,21 @@ global_progressbars = {}
 galleries = {}
 galleryObservers = {}
 
+// this tracks laumnches of window.setTimeout for progressbar to prevent starting a new timeout when the previous is still running
+timeoutIds = {}
+
 function check_progressbar(id_part, id_progressbar, id_progressbar_span, id_skip, id_interrupt, id_preview, id_gallery){
-    var progressbar = gradioApp().getElementById(id_progressbar)
+    // gradio 3.8's enlightened approach allows them to create two nested div elements inside each other with same id
+    // every time you use gr.HTML(elem_id='xxx'), so we handle this here
+    var progressbar = gradioApp().querySelector("#"+id_progressbar+" #"+id_progressbar)
+    var progressbarParent
+    if(progressbar){
+        progressbarParent = gradioApp().querySelector("#"+id_progressbar)
+    } else{
+        progressbar = gradioApp().getElementById(id_progressbar)
+        progressbarParent = null
+    }
+
     var skip = id_skip ? gradioApp().getElementById(id_skip) : null
     var interrupt = gradioApp().getElementById(id_interrupt)
     
@@ -26,18 +39,26 @@ function check_progressbar(id_part, id_progressbar, id_progressbar_span, id_skip
 	    global_progressbars[id_progressbar] = progressbar
 
         var mutationObserver = new MutationObserver(function(m){
+            if(timeoutIds[id_part]) return;
+
             preview = gradioApp().getElementById(id_preview)
             gallery = gradioApp().getElementById(id_gallery)
 
             if(preview != null && gallery != null){
                 preview.style.width = gallery.clientWidth + "px"
                 preview.style.height = gallery.clientHeight + "px"
+                if(progressbarParent) progressbar.style.width = progressbarParent.clientWidth + "px"
 
 				//only watch gallery if there is a generation process going on
                 check_gallery(id_gallery);
 
                 var progressDiv = gradioApp().querySelectorAll('#' + id_progressbar_span).length > 0;
-                if(!progressDiv){
+                if(progressDiv){
+                    timeoutIds[id_part] = window.setTimeout(function() {
+                        timeoutIds[id_part] = null
+                        requestMoreProgress(id_part, id_progressbar_span, id_skip, id_interrupt)
+                    }, 500)
+                } else{
                     if (skip) {
                         skip.style.display = "none"
                     }
@@ -47,13 +68,10 @@ function check_progressbar(id_part, id_progressbar, id_progressbar_span, id_skip
                     if (galleryObservers[id_gallery]) {
                         galleryObservers[id_gallery].disconnect();
                         galleries[id_gallery] = null;
-                    }    
+                    }
                 }
-
-
             }
 
-            window.setTimeout(function() { requestMoreProgress(id_part, id_progressbar_span, id_skip, id_interrupt) }, 500)
         });
         mutationObserver.observe( progressbar, { childList:true, subtree:true })
 	}

launch.py

@@ -7,6 +7,7 @@ import shlex
 import platform
 
 dir_repos = "repositories"
+dir_extensions = "extensions"
 python = sys.executable
 git = os.environ.get('GIT', "git")
 index_url = os.environ.get('INDEX_URL', "")
@@ -16,11 +17,11 @@ def extract_arg(args, name):
     return [x for x in args if x != name], name in args
 
 
-def run(command, desc=None, errdesc=None):
+def run(command, desc=None, errdesc=None, custom_env=None):
     if desc is not None:
         print(desc)
 
-    result = subprocess.run(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True)
+    result = subprocess.run(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True, env=os.environ if custom_env is None else custom_env)
 
     if result.returncode != 0:
 
@@ -101,9 +102,27 @@ def version_check(commit):
         else:
             print("Not a git clone, can't perform version check.")
     except Exception as e:
-        print("versipm check failed",e)
+        print("version check failed", e)
+
+
+def run_extensions_installers():
+    if not os.path.isdir(dir_extensions):
+        return
+
+    for dirname_extension in os.listdir(dir_extensions):
+        path_installer = os.path.join(dir_extensions, dirname_extension, "install.py")
+        if not os.path.isfile(path_installer):
+            continue
+
+        try:
+            env = os.environ.copy()
+            env['PYTHONPATH'] = os.path.abspath(".")
+
+            print(run(f'"{python}" "{path_installer}"', errdesc=f"Error running install.py for extension {dirname_extension}", custom_env=env))
+        except Exception as e:
+            print(e, file=sys.stderr)
+
 
-        
 def prepare_enviroment():
     torch_command = os.environ.get('TORCH_COMMAND', "pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 --extra-index-url https://download.pytorch.org/whl/cu113")
     requirements_file = os.environ.get('REQS_FILE', "requirements_versions.txt")
@@ -189,6 +208,8 @@ def prepare_enviroment():
 
     run_pip(f"install -r {requirements_file}", "requirements for Web UI")
 
+    run_extensions_installers()
+
     if update_check:
         version_check(commit)
     

localizations/de_DE.json

@@ -70,7 +70,7 @@
     "None": "Nichts",
     "Prompt matrix": "Promptmatrix",
     "Prompts from file or textbox": "Prompts aus Datei oder Textfeld",
-    "X/Y plot": "X/Y Graf",
+    "X/Y plot": "X/Y Graph",
     "Put variable parts at start of prompt": "Variable teile am start des Prompt setzen",
     "Iterate seed every line": "Iterate seed every line",
     "List of prompt inputs": "List of prompt inputs",
@@ -455,4 +455,4 @@
     "Only applies to inpainting models. Determines how strongly to mask off the original image for inpainting and img2img. 1.0 means fully masked, which is the default behaviour. 0.0 means a fully unmasked conditioning. Lower values will help preserve the overall composition of the image, but will struggle with large changes.": "Gilt nur für Inpainting-Modelle. Legt fest, wie stark das Originalbild für Inpainting und img2img maskiert werden soll. 1.0 bedeutet vollständig maskiert, was das Standardverhalten ist. 0.0 bedeutet eine vollständig unmaskierte Konditionierung. Niedrigere Werte tragen dazu bei, die Gesamtkomposition des Bildes zu erhalten, sind aber bei großen Änderungen problematisch.",
     "List of setting names, separated by commas, for settings that should go to the quick access bar at the top, rather than the usual setting tab. See modules/shared.py for setting names. Requires restarting to apply.": "Liste von Einstellungsnamen, getrennt durch Kommas, für Einstellungen, die in der Schnellzugriffsleiste oben erscheinen sollen, anstatt in dem üblichen Einstellungs-Tab. Siehe modules/shared.py für Einstellungsnamen. Erfordert einen Neustart zur Anwendung.",
     "If this values is non-zero, it will be added to seed and used to initialize RNG for noises when using samplers with Eta. You can use this to produce even more variation of images, or you can use this to match images of other software if you know what you are doing.": "Wenn dieser Wert ungleich Null ist, wird er zum Seed addiert und zur Initialisierung des RNG für Noise bei der Verwendung von Samplern mit Eta verwendet. Dies kann verwendet werden, um noch mehr Variationen von Bildern zu erzeugen, oder um Bilder von anderer Software zu erzeugen, wenn Sie wissen, was Sie tun."
-}
+}

localizations/ko_KR.json

@@ -9,11 +9,13 @@
   " images in this directory. Loaded ": "개의 이미지가 이 경로에 존재합니다. ",
   " pages": "페이지로 나뉘어 표시합니다.",
   ", divided into ": "입니다. ",
+  ". Use Installed tab to restart.": "에 성공적으로 설치하였습니다. 설치된 확장기능 탭에서 UI를 재시작해주세요.",
   "1st and last digit must be 1. ex:'1, 2, 1'": "1st and last digit must be 1. ex:'1, 2, 1'",
   "[wiki]": " [위키] 참조",
   "A directory on the same machine where the server is running.": "WebUI 서버가 돌아가고 있는 디바이스에 존재하는 디렉토리를 선택해 주세요.",
   "A merger of the two checkpoints will be generated in your": "체크포인트들이 병합된 결과물이 당신의",
   "A value that determines the output of random number generator - if you create an image with same parameters and seed as another image, you'll get the same result": "난수 생성기의 결과물을 지정하는 값 - 동일한 설정값과 동일한 시드를 적용 시, 완전히 똑같은 결과물을 얻게 됩니다.",
+  "Action": "작업",
   "Add a random artist to the prompt.": "프롬프트에 랜덤한 작가 추가",
   "Add a second progress bar to the console that shows progress for an entire job.": "콘솔에 전체 작업의 진행도를 보여주는 2번째 프로그레스 바 추가하기",
   "Add difference": "차이점 추가",
@@ -22,6 +24,8 @@
   "Add model hash to generation information": "생성 정보에 모델 해시 추가",
   "Add model name to generation information": "생성 정보에 모델 이름 추가",
   "Add number to filename when saving": "이미지를 저장할 때 파일명에 숫자 추가하기",
+  "Aesthetic Gradients": "스타일 그라디언트",
+  "Aesthetic Image Scorer": "스타일 이미지 스코어러",
   "Aesthetic imgs embedding": "스타일 이미지 임베딩",
   "Aesthetic learning rate": "스타일 학습 수",
   "Aesthetic steps": "스타일 스텝 수",
@@ -33,22 +37,32 @@
   "Always save all generated images": "생성된 이미지 항상 저장하기",
   "api": "",
   "append": "뒤에 삽입",
+  "Append commas": "쉼표 삽입",
+  "Apply and restart UI": "적용 후 UI 재시작",
   "Apply color correction to img2img results to match original colors.": "이미지→이미지 결과물이 기존 색상과 일치하도록 색상 보정 적용하기",
   "Apply selected styles to current prompt": "현재 프롬프트에 선택된 스타일 적용",
   "Apply settings": "설정 적용하기",
+  "Artists to study": "연구할만한 작가들",
+  "auto": "자동",
   "Auto focal point crop": "초점 기준 크롭(자동 감지)",
+  "Autocomplete options": "자동완성 설정",
+  "Available": "지원되는 확장기능 목록",
   "Batch count": "배치 수",
   "Batch from Directory": "저장 경로로부터 여러장 처리",
   "Batch img2img": "이미지→이미지 배치",
   "Batch Process": "이미지 여러장 처리",
   "Batch size": "배치 크기",
+  "behind": "최신 아님",
   "BSRGAN 4x": "BSRGAN 4x",
   "built with gradio": "gradio로 제작되었습니다",
+  "Calculates aesthetic score for generated images using CLIP+MLP Aesthetic Score Predictor based on Chad Scorer": "Chad 스코어러를 기반으로 한 CLIP+MLP 스타일 점수 예측기를 이용해 생성된 이미지의 스타일 점수를 계산합니다.",
   "Cancel generate forever": "반복 생성 취소",
+  "cfg cnt": "CFG 변화 횟수",
   "cfg count": "CFG 변화 횟수",
   "CFG Scale": "CFG 스케일",
   "cfg1 min/max": "CFG1 최소/최대",
   "cfg2 min/max": "CFG2 최소/최대",
+  "Check for updates": "업데이트 확인",
   "Check progress": "진행도 체크",
   "Check progress (first)": "진행도 체크 (처음)",
   "checkpoint": " 체크포인트 ",
@@ -64,10 +78,14 @@
   "CodeFormer weight parameter; 0 = maximum effect; 1 = minimum effect": "CodeFormer 가중치 설정값 (0 = 최대 효과, 1 = 최소 효과)",
   "Collect": "즐겨찾기",
   "Color variation": "색깔 다양성",
+  "Combinations": "조합",
+  "Combinatorial batches": "조합 배치 수",
+  "Combinatorial generation": "조합 생성",
   "copy": "복사",
   "Create a grid where images will have different parameters. Use inputs below to specify which parameters will be shared by columns and rows": "서로 다른 설정값으로 생성된 이미지의 그리드를 만듭니다. 아래의 설정으로 가로/세로에 어떤 설정값을 적용할지 선택하세요.",
   "Create a text file next to every image with generation parameters.": "생성된 이미지마다 생성 설정값을 담은 텍스트 파일 생성하기",
   "Create aesthetic images embedding": "스타일 이미지 임베딩 생성하기",
+  "Create an embedding from one or few pictures and use it to apply their style to generated images.": "하나 혹은 그 이상의 이미지들로부터 임베딩을 생성해, 그 이미지들의 스타일을 다른 이미지 생성 시 적용할 수 있게 해줍니다.",
   "Create debug image": "디버그 이미지 생성",
   "Create embedding": "임베딩 생성",
   "Create flipped copies": "좌우로 뒤집은 복사본 생성",
@@ -78,14 +96,18 @@
   "custom fold": "커스텀 경로",
   "Custom Name (Optional)": "병합 모델 이름 (선택사항)",
   "Dataset directory": "데이터셋 경로",
+  "Dataset Tag Editor": "데이터셋 태그 편집기",
+  "date": "생성 일자",
   "DDIM": "DDIM",
   "Decode CFG scale": "디코딩 CFG 스케일",
   "Decode steps": "디코딩 스텝 수",
   "Delete": "삭제",
+  "delete next": "선택한 이미지부터 시작해서 삭제할 이미지 갯수",
   "Denoising": "디노이징",
   "Denoising Diffusion Implicit Models - best at inpainting": "Denoising Diffusion Implicit Models - 인페이팅에 뛰어남",
   "Denoising strength": "디노이즈 강도",
   "Denoising strength change factor": "디노이즈 강도 변경 배수",
+  "Description": "설명",
   "Destination directory": "결과물 저장 경로",
   "Determines how little respect the algorithm should have for image's content. At 0, nothing will change, and at 1 you'll get an unrelated image. With values below 1.0, processing will take less steps than the Sampling Steps slider specifies.": "알고리즘이 얼마나 원본 이미지를 반영할지를 결정하는 수치입니다. 0일 경우 아무것도 바뀌지 않고, 1일 경우 원본 이미지와 전혀 관련없는 결과물을 얻게 됩니다. 1.0 아래의 값일 경우, 설정된 샘플링 스텝 수보다 적은 스텝 수를 거치게 됩니다.",
   "Directory for saving images using the Save button": "저장 버튼을 이용해 저장하는 이미지들의 저장 경로",
@@ -108,6 +130,8 @@
   "Draw mask": "마스크 직접 그리기",
   "Drop File Here": "파일을 끌어 놓으세요",
   "Drop Image Here": "이미지를 끌어 놓으세요",
+  "Dropdown": "드롭다운",
+  "Dynamic Prompts": "다이나믹 프롬프트",
   "Embedding": "임베딩",
   "Embedding Learning rate": "임베딩 학습률",
   "Emphasis: use (text) to make model pay more attention to text and [text] to make it pay less attention": "강조 : (텍스트)를 이용해 모델의 텍스트에 대한 가중치를 더 강하게 주고 [텍스트]를 이용해 더 약하게 줍니다.",
@@ -127,6 +151,9 @@
   "Euler a": "Euler a",
   "Euler Ancestral - very creative, each can get a completely different picture depending on step count, setting steps to higher than 30-40 does not help": "Euler Ancestral - 매우 창의적, 스텝 수에 따라 완전히 다른 결과물이 나올 수 있음. 30~40보다 높은 스텝 수는 효과가 미미함",
   "Existing Caption txt Action": "이미 존재하는 캡션 텍스트 처리",
+  "Extension": "확장기능",
+  "Extension index URL": "확장기능 목록 URL",
+  "Extensions": "확장기능",
   "Extra": "고급",
   "Extras": "부가기능",
   "extras": "부가기능",
@@ -134,7 +161,7 @@
   "Face restoration": "얼굴 보정",
   "Face restoration model": "얼굴 보정 모델",
   "Fall-off exponent (lower=higher detail)": "감쇠 지수 (낮을수록 디테일이 올라감)",
-  "favorites": "즐겨찾기",
+  "Favorites": "즐겨찾기",
   "File": "파일",
   "File format for grids": "그리드 이미지 파일 형식",
   "File format for images": "이미지 파일 형식",
@@ -150,6 +177,7 @@
   "First Page": "처음 페이지",
   "Firstpass height": "초기 세로길이",
   "Firstpass width": "초기 가로길이",
+  "Fixed seed": "시드 고정",
   "Focal point edges weight": "경계면 가중치",
   "Focal point entropy weight": "엔트로피 가중치",
   "Focal point face weight": "얼굴 가중치",
@@ -184,8 +212,10 @@
   "ignore": "무시",
   "Image": "이미지",
   "Image Browser": "이미지 브라우저",
+  "Image browser": "이미지 브라우저",
   "Image for img2img": "Image for img2img",
   "Image for inpainting with mask": "마스크로 인페인팅할 이미지",
+  "Image not found (may have been already moved)": "이미지를 찾을 수 없습니다 (이미 옮겨졌을 수 있음)",
   "Images Browser": "이미지 브라우저",
   "Images directory": "이미지 경로",
   "Images filename pattern": "이미지 파일명 패턴",
@@ -193,6 +223,7 @@
   "img2img alternative test": "이미지→이미지 대체버전 테스트",
   "img2img DDIM discretize": "이미지→이미지 DDIM 이산화",
   "img2img history": "이미지→이미지 기록",
+  "Implements an expressive template language for random or combinatorial prompt generation along with features to support deep wildcard directory structures.": "무작위/조합 프롬프트 생성을 위한 문법과 복잡한 와일드카드 구조를 지원합니다.",
   "In loopback mode, on each loop the denoising strength is multiplied by this value. <1 means decreasing variety so your sequence will converge on a fixed picture. >1 means increasing variety so your sequence will become more and more chaotic.": "루프백 모드에서는 매 루프마다 디노이즈 강도에 이 값이 곱해집니다. 1보다 작을 경우 다양성이 낮아져 결과 이미지들이 고정된 형태로 모일 겁니다. 1보다 클 경우 다양성이 높아져 결과 이미지들이 갈수록 혼란스러워지겠죠.",
   "Include Separate Images": "분리된 이미지 포함하기",
   "Increase coherency by padding from the last comma within n tokens when using more than 75 tokens": "75개보다 많은 토큰을 사용시 마지막 쉼표로부터 N개의 토큰 이내에 패딩을 추가해 통일성 증가시키기",
@@ -205,6 +236,11 @@
   "Inpainting conditioning mask strength": "인페인팅 조절 마스크 강도",
   "Input directory": "인풋 이미지 경로",
   "Input images directory": "이미지 경로 입력",
+  "Inspiration": "\"영감\"",
+  "Install": "설치",
+  "Install from URL": "URL로부터 확장기능 설치",
+  "Installed": "설치된 확장기능",
+  "Installed into ": "확장기능을 ",
   "Interpolation Method": "보간 방법",
   "Interrogate\nCLIP": "CLIP\n분석",
   "Interrogate\nDeepBooru": "DeepBooru\n분석",
@@ -223,6 +259,7 @@
   "Just resize": "리사이징",
   "Keep -1 for seeds": "시드값 -1로 유지",
   "keep whatever was there originally": "이미지 원본 유지",
+  "keyword": "프롬프트",
   "Label": "라벨",
   "Lanczos": "Lanczos",
   "Last prompt:": "마지막 프롬프트 : ",
@@ -230,23 +267,29 @@
   "Last saved image:": "마지막으로 저장된 이미지 : ",
   "latent noise": "잠재 노이즈",
   "latent nothing": "잠재 공백",
+  "latest": "최신 버전",
   "LDSR": "LDSR",
   "LDSR processing steps. Lower = faster": "LDSR 스텝 수. 낮은 값 = 빠른 속도",
   "leakyrelu": "leakyrelu",
   "Leave blank to save images to the default path.": "기존 저장 경로에 이미지들을 저장하려면 비워두세요.",
+  "Leave empty for auto": "자동 설정하려면 비워두십시오",
   "left": "왼쪽",
+  "Lets you edit captions in training datasets.": "훈련에 사용되는 데이터셋의 캡션을 수정할 수 있게 해줍니다.",
   "linear": "linear",
   "List of prompt inputs": "프롬프트 입력 리스트",
   "List of setting names, separated by commas, for settings that should go to the quick access bar at the top, rather than the usual setting tab. See modules/shared.py for setting names. Requires restarting to apply.": "설정 탭이 아니라 상단의 빠른 설정 바에 위치시킬 설정 이름을 쉼표로 분리해서 입력하십시오. 설정 이름은 modules/shared.py에서 찾을 수 있습니다. 재시작이 필요합니다.",
   "LMS": "LMS",
   "LMS Karras": "LMS Karras",
   "Load": "불러오기",
+  "Load from:": "URL로부터 불러오기",
   "Loading...": "로딩 중...",
+  "Local directory name": "로컬 경로 이름",
   "Localization (requires restart)": "현지화 (재시작 필요)",
   "Log directory": "로그 경로",
   "Loopback": "루프백",
   "Loops": "루프 수",
   "Loss:": "손실(Loss) : ",
+  "Magic prompt": "매직 프롬프트",
   "Make an attempt to produce a picture similar to what would have been produced with same seed at specified resolution": "동일한 시드 값으로 생성되었을 이미지를 주어진 해상도로 최대한 유사하게 재현합니다.",
   "Make K-diffusion samplers produce same images in a batch as when making a single image": "K-diffusion 샘플러들이 단일 이미지를 생성하는 것처럼 배치에서도 동일한 이미지를 생성하게 하기",
   "Make Zip when Save?": "저장 시 Zip 생성하기",
@@ -260,7 +303,9 @@
   "Minimum number of pages per load": "한번 불러올 때마다 불러올 최소 페이지 수",
   "Modules": "모듈",
   "Move face restoration model from VRAM into RAM after processing": "처리가 완료되면 얼굴 보정 모델을 VRAM에서 RAM으로 옮기기",
-  "Move VAE and CLIP to RAM when training hypernetwork. Saves VRAM.": "하이퍼네트워크 훈련 진행 시 VAE와 CLIP을 RAM으로 옮기기. VRAM이 절약됩니다.",
+  "Move to favorites": "즐겨찾기로 옮기기",
+  "Move VAE and CLIP to RAM when training if possible. Saves VRAM.": "훈련 진행 시 가능하면 VAE와 CLIP을 RAM으로 옮기기. VRAM이 절약됩니다.",
+  "Moved to favorites": "즐겨찾기로 옮겨짐",
   "Multiplier (M) - set to 0 to get model A": "배율 (M) - 0으로 적용하면 모델 A를 얻게 됩니다",
   "Name": "이름",
   "Negative prompt": "네거티브 프롬프트",
@@ -285,6 +330,7 @@
   "original": "원본 유지",
   "Original negative prompt": "기존 네거티브 프롬프트",
   "Original prompt": "기존 프롬프트",
+  "Others": "기타",
   "Outpainting direction": "아웃페인팅 방향",
   "Outpainting mk2": "아웃페인팅 마크 2",
   "Output directory": "이미지 저장 경로",
@@ -303,6 +349,7 @@
   "Overwrite Old Hypernetwork": "기존 하이퍼네트워크 덮어쓰기",
   "Page Index": "페이지 인덱스",
   "parameters": "설정값",
+  "path name": "경로 이름",
   "Path to directory where to write outputs": "결과물을 출력할 경로",
   "Path to directory with input images": "인풋 이미지가 있는 경로",
   "Paths for saving": "저장 경로",
@@ -330,6 +377,7 @@
   "Prompt template file": "프롬프트 템플릿 파일 경로",
   "Prompts": "프롬프트",
   "Prompts from file or textbox": "파일이나 텍스트박스로부터 프롬프트 불러오기",
+  "Provides an interface to browse created images in the web browser.": "생성된 이미지를 브라우저 내에서 볼 수 있는 인터페이스를 추가합니다.",
   "Put variable parts at start of prompt": "변경되는 프롬프트를 앞에 위치시키기",
   "quad": "quad",
   "Quality for saved jpeg images": "저장된 jpeg 이미지들의 품질",
@@ -337,11 +385,13 @@
   "R-ESRGAN 4x+ Anime6B": "R-ESRGAN 4x+ Anime6B",
   "Random": "랜덤",
   "Random grid": "랜덤 그리드",
+  "Randomly display the pictures of the artist's or artistic genres typical style, more pictures of this artist or genre is displayed after selecting. So you don't have to worry about how hard it is to choose the right style of art when you create.": "특정 작가 또는 스타일의 이미지들 중 하나를 무작위로 보여줍니다. 선택 후 선택한 작가 또는 스타일의 이미지들이 더 나타나게 됩니다. 고르기 어려워도 걱정하실 필요 없어요!",
   "Randomness": "랜덤성",
   "Read generation parameters from prompt or last generation if prompt is empty into user interface.": "클립보드에 복사된 정보로부터 설정값 읽어오기/프롬프트창이 비어있을경우 제일 최근 설정값 불러오기",
   "Read parameters (prompt, etc...) from txt2img tab when making previews": "프리뷰 이미지 생성 시 텍스트→이미지 탭에서 설정값(프롬프트 등) 읽어오기",
   "Recommended settings: Sampling Steps: 80-100, Sampler: Euler a, Denoising strength: 0.8": "추천 설정값 - 샘플링 스텝 수 : 80-100 , 샘플러 : Euler a, 디노이즈 강도 : 0.8",
   "Reload custom script bodies (No ui updates, No restart)": "커스텀 스크립트 리로드하기(UI 업데이트 없음, 재시작 없음)",
+  "Reloading...": "재시작 중...",
   "relu": "relu",
   "Renew Page": "Renew Page",
   "Request browser notifications": "브라우저 알림 권한 요청",
@@ -361,6 +411,7 @@
   "Reuse seed from last generation, mostly useful if it was randomed": "이전 생성에서 사용된 시드를 불러옵니다. 랜덤하게 생성했을 시 도움됨",
   "right": "오른쪽",
   "Run": "가동",
+  "Sample extension. Allows you to use __name__ syntax in your prompt to get a random line from a file named name.txt in the wildcards directory. Also see Dynamic Prompts for similar functionality.": "샘플 확장기능입니다. __이름__형식의 문법을 사용해 와일드카드 경로 내의 이름.txt파일로부터 무작위 프롬프트를 적용할 수 있게 해줍니다. 유사한 확장기능으로 다이나믹 프롬프트가 있습니다.",
   "Sampler": "샘플러",
   "Sampler parameters": "샘플러 설정값",
   "Sampling method": "샘플링 방법",
@@ -368,6 +419,7 @@
   "Save": "저장",
   "Save a copy of embedding to log directory every N steps, 0 to disable": "N스텝마다 로그 경로에 임베딩을 저장합니다, 비활성화하려면 0으로 설정하십시오.",
   "Save a copy of image before applying color correction to img2img results": "이미지→이미지 결과물에 색상 보정을 진행하기 전 이미지의 복사본을 저장하기",
+  "Save a copy of image before applying highres fix.": "고해상도 보정을 진행하기 전 이미지의 복사본을 저장하기",
   "Save a copy of image before doing face restoration.": "얼굴 보정을 진행하기 전 이미지의 복사본을 저장하기",
   "Save an csv containing the loss to log directory every N steps, 0 to disable": "N스텝마다 로그 경로에 손실(Loss)을 포함하는 csv 파일을 저장합니다, 비활성화하려면 0으로 설정하십시오.",
   "Save an image to log directory every N steps, 0 to disable": "N스텝마다 로그 경로에 이미지를 저장합니다, 비활성화하려면 0으로 설정하십시오.",
@@ -412,6 +464,7 @@
   "Show progressbar": "프로그레스 바 보이기",
   "Show result images": "이미지 결과 보이기",
   "Show Textbox": "텍스트박스 보이기",
+  "Shows a gallery of generated pictures by artists separated into categories.": "생성된 이미지들을 작가별로 분류해 보여줍니다. 원본 - https://artiststostudy.pages.dev",
   "Sigma adjustment for finding noise for image": "이미지 노이즈를 찾기 위해 시그마 조정",
   "Sigma Churn": "시그마 섞기",
   "sigma churn": "시그마 섞기",
@@ -424,6 +477,7 @@
   "Skip": "건너뛰기",
   "Slerp angle": "구면 선형 보간 각도",
   "Slerp interpolation": "구면 선형 보간",
+  "sort by": "정렬 기준",
   "Source": "원본",
   "Source directory": "원본 경로",
   "Split image overlap ratio": "이미지 분할 겹침 비율",
@@ -431,6 +485,7 @@
   "Split oversized images": "사이즈가 큰 이미지 분할하기",
   "Stable Diffusion": "Stable Diffusion",
   "Stable Diffusion checkpoint": "Stable Diffusion 체크포인트",
+  "step cnt": "스텝 변화 횟수",
   "step count": "스텝 변화 횟수",
   "step1 min/max": "스텝1 최소/최대",
   "step2 min/max": "스텝2 최소/최대",
@@ -447,6 +502,7 @@
   "System": "시스템",
   "Tertiary model (C)": "3차 모델 (C)",
   "Textbox": "텍스트박스",
+  "The official port of Deforum, an extensive script for 2D and 3D animations, supporting keyframable sequences, dynamic math parameters (even inside the prompts), dynamic masking, depth estimation and warping.": "Deforum의 공식 포팅 버전입니다. 2D와 3D 애니메이션, 키프레임 시퀀스, 수학적 매개변수, 다이나믹 마스킹 등을 지원합니다.",
   "This regular expression will be used extract words from filename, and they will be joined using the option below into label text used for training. Leave empty to keep filename text as it is.": "이 정규표현식은 파일명으로부터 단어를 추출하는 데 사용됩니다. 추출된 단어들은 하단의 설정을 이용해 라벨 텍스트로 변환되어 훈련에 사용됩니다. 파일명 텍스트를 유지하려면 비워두십시오.",
   "This string will be used to join split words into a single line if the option above is enabled.": "이 문자열은 상단 설정이 활성화되어있을 때 분리된 단어들을 한 줄로 합치는 데 사용됩니다.",
   "This text is used to rotate the feature space of the imgs embs": "이 텍스트는 이미지 임베딩의 특징 공간을 회전하는 데 사용됩니다.",
@@ -467,7 +523,9 @@
   "txt2img": "텍스트→이미지",
   "txt2img history": "텍스트→이미지 기록",
   "uniform": "uniform",
+  "unknown": "알수 없음",
   "up": "위쪽",
+  "Update": "업데이트",
   "Upload mask": "마스크 업로드하기",
   "Upload prompt inputs": "입력할 프롬프트를 업로드하십시오",
   "Upscale Before Restoring Faces": "얼굴 보정을 진행하기 전에 업스케일링 먼저 진행하기",
@@ -479,15 +537,20 @@
   "Upscaler 2 visibility": "업스케일러 2 가시성",
   "Upscaler for img2img": "이미지→이미지 업스케일러",
   "Upscaling": "업스케일링",
+  "URL for extension's git repository": "확장기능의 git 레포 URL",
   "Use a two step process to partially create an image at smaller resolution, upscale, and then improve details in it without changing composition": "저해상도 이미지를 1차적으로 생성 후 업스케일을 진행하여, 이미지의 전체적인 구성을 바꾸지 않고 세부적인 디테일을 향상시킵니다.",
   "Use an empty output directory to save pictures normally instead of writing to the output directory.": "저장 경로를 비워두면 기본 저장 폴더에 이미지들이 저장됩니다.",
   "Use BLIP for caption": "캡션에 BLIP 사용",
+  "Use checkbox to enable the extension; it will be enabled or disabled when you click apply button": "체크박스를 이용해 적용할 확장기능을 선택하세요. 변경사항은 적용 후 UI 재시작 버튼을 눌러야 적용됩니다.",
+  "Use checkbox to mark the extension for update; it will be updated when you click apply button": "체크박스를 이용해 업데이트할 확장기능을 선택하세요. 업데이트는 적용 후 UI 재시작 버튼을 눌러야 적용됩니다.",
+  "Use cross attention optimizations while training": "훈련 진행 시 크로스 어텐션 최적화 사용",
   "Use deepbooru for caption": "캡션에 deepbooru 사용",
   "Use dropout": "드롭아웃 사용",
   "Use following tags to define how filenames for images are chosen: [steps], [cfg], [prompt], [prompt_no_styles], [prompt_spaces], [width], [height], [styles], [sampler], [seed], [model_hash], [prompt_words], [date], [datetime], [datetime<Format>], [datetime<Format><Time Zone>], [job_timestamp]; leave empty for default.": "다음 태그들을 사용해 이미지 파일명 형식을 결정하세요 : [steps], [cfg], [prompt], [prompt_no_styles], [prompt_spaces], [width], [height], [styles], [sampler], [seed], [model_hash], [prompt_words], [date], [datetime], [datetime<Format>], [datetime<Format><Time Zone>], [job_timestamp]. 비워두면 기본값으로 설정됩니다.",
   "Use following tags to define how subdirectories for images and grids are chosen: [steps], [cfg], [prompt], [prompt_no_styles], [prompt_spaces], [width], [height], [styles], [sampler], [seed], [model_hash], [prompt_words], [date], [datetime], [datetime<Format>], [datetime<Format><Time Zone>], [job_timestamp]; leave empty for default.": "다음 태그들을 사용해 이미지와 그리드의 하위 디렉토리명의 형식을 결정하세요 : [steps], [cfg], [prompt], [prompt_no_styles], [prompt_spaces], [width], [height], [styles], [sampler], [seed], [model_hash], [prompt_words], [date], [datetime], [datetime<Format>], [datetime<Format><Time Zone>], [job_timestamp]. 비워두면 기본값으로 설정됩니다.",
   "Use old emphasis implementation. Can be useful to reproduce old seeds.": "옛 방식의 강조 구현을 사용합니다. 옛 시드를 재현하는 데 효과적일 수 있습니다.",
   "Use original name for output filename during batch process in extras tab": "부가기능 탭에서 이미지를 여러장 처리 시 결과물 파일명에 기존 파일명 사용하기",
+  "Use same random seed for all lines": "모든 줄에 동일한 시드 사용",
   "Use same seed for each image": "각 이미지에 동일한 시드 사용",
   "use spaces for tags in deepbooru": "deepbooru에서 태그에 공백 사용",
   "User interface": "사용자 인터페이스",

localizations/pt_BR.json

@@ -17,6 +17,7 @@
     "Checkpoint Merger": "Fusão de Checkpoint",
     "Train": "Treinar",
     "Settings": "Configurações",
+    "Extensions": "Extensions",
     "Prompt": "Prompt",
     "Negative prompt": "Prompt negativo",
     "Run": "Executar",
@@ -93,13 +94,13 @@
     "Eta": "Tempo estimado",
     "Clip skip": "Pular Clip",
     "Denoising": "Denoising",
+    "Cond. Image Mask Weight": "Peso da Máscara Condicional de Imagem",
     "X values": "Valores de X",
     "Y type": "Tipo de Y",
     "Y values": "Valores de Y",
     "Draw legend": "Desenhar a legenda",
     "Include Separate Images": "Incluir Imagens Separadas",
     "Keep -1 for seeds": "Manter em -1 para seeds",
-    "Drop Image Here": "Solte a imagem aqui",
     "Save": "Salvar",
     "Send to img2img": "Mandar para img2img",
     "Send to inpaint": "Mandar para inpaint",
@@ -110,6 +111,7 @@
     "Inpaint": "Inpaint",
     "Batch img2img": "Lote img2img",
     "Image for img2img": "Imagem para img2img",
+    "Drop Image Here": "Solte a imagem aqui",
     "Image for inpainting with mask": "Imagem para inpainting com máscara",
     "Mask": "Máscara",
     "Mask blur": "Desfoque da máscara",
@@ -166,16 +168,10 @@
     "Upscaler": "Ampliador",
     "Lanczos": "Lanczos",
     "LDSR": "LDSR",
-    "4x_foolhardy_Remacri": "4x_foolhardy_Remacri",
-    "Put ESRGAN models here": "Coloque modelos ESRGAN aqui",
-    "R-ESRGAN General 4xV3": "R-ESRGAN General 4xV3",
-    "R-ESRGAN AnimeVideo": "R-ESRGAN AnimeVideo",
-    "R-ESRGAN 4x+": "R-ESRGAN 4x+",
-    "R-ESRGAN 4x+ Anime6B": "R-ESRGAN 4x+ Anime6B",
-    "R-ESRGAN 2x+": "R-ESRGAN 2x+",
-    "ScuNET": "ScuNET",
+    "ESRGAN_4x": "ESRGAN_4x",
+    "ScuNET GAN": "ScuNET GAN",
     "ScuNET PSNR": "ScuNET PSNR",
-    "put_swinir_models_here": "put_swinir_models_here",
+    "SwinIR 4x": "SwinIR 4x",
     "Single Image": "Uma imagem",
     "Batch Process": "Processo em lote",
     "Batch from Directory": "Lote apartir de diretório",
@@ -189,7 +185,7 @@
     "GFPGAN visibility": "Visibilidade GFPGAN",
     "CodeFormer visibility": "Visibilidade CodeFormer",
     "CodeFormer weight (0 = maximum effect, 1 = minimum effect)": "Peso do CodeFormer (0 = efeito máximo, 1 = efeito mínimo)",
-    "Open output directory": "Abrir diretório de saída",
+    "Upscale Before Restoring Faces": "Ampliar Antes de Refinar Rostos",
     "Send to txt2img": "Mandar para txt2img",
     "A merger of the two checkpoints will be generated in your": "Uma fusão dos dois checkpoints será gerada em seu",
     "checkpoint": "checkpoint",
@@ -216,6 +212,7 @@
     "Modules": "Módulos",
     "Enter hypernetwork layer structure": "Entrar na estrutura de camadas da hypernetwork",
     "Select activation function of hypernetwork": "Selecionar a função de ativação de hypernetwork",
+    "linear": "linear",
     "relu": "relu",
     "leakyrelu": "leakyrelu",
     "elu": "elu",
@@ -227,12 +224,10 @@
     "glu": "glu",
     "hardshrink": "hardshrink",
     "hardsigmoid": "hardsigmoid",
-    "hardswish": "hardswish",
     "hardtanh": "hardtanh",
     "logsigmoid": "logsigmoid",
     "logsoftmax": "logsoftmax",
     "mish": "mish",
-    "multiheadattention": "multiheadattention",
     "prelu": "prelu",
     "rrelu": "rrelu",
     "relu6": "relu6",
@@ -274,9 +269,9 @@
     "Focal point edges weight": "Peso de ponto focal para bordas",
     "Create debug image": "Criar imagem de depuração",
     "Preprocess": "Pré-processar",
-    "Train an embedding; must specify a directory with a set of 1:1 ratio images": "Treinar um embedding; precisa especificar um diretório com imagens de proporção 1:1",
+    "Train an embedding; must specify a directory with a set of 1:1 ratio images": "Treinar uma incorporação; precisa especificar um diretório com imagens de proporção 1:1",
     "[wiki]": "[wiki]",
-    "Embedding": "Embedding",
+    "Embedding": "Incorporação",
     "Embedding Learning rate": "Taxa de aprendizagem da incorporação",
     "Hypernetwork Learning rate": "Taxa de aprendizagem de Hypernetwork",
     "Dataset directory": "Diretório de Dataset",
@@ -345,9 +340,11 @@
     "Filename join string": "Nome de arquivo join string",
     "Number of repeats for a single input image per epoch; used only for displaying epoch number": "Número de repetições para entrada única de imagens por época; serve apenas para mostrar o número de época",
     "Save an csv containing the loss to log directory every N steps, 0 to disable": "Salvar um csv com as perdas para o diretório de log a cada N steps, 0 para desativar",
+    "Use cross attention optimizations while training": "Usar otimizações de atenção cruzada enquanto treinando",
     "Stable Diffusion": "Stable Diffusion",
     "Checkpoints to cache in RAM": "Checkpoints para manter no cache da RAM",
     "Hypernetwork strength": "Força da Hypernetwork",
+     "Inpainting conditioning mask strength": "Força do inpaint para máscaras condicioniais",
     "Apply color correction to img2img results to match original colors.": "Aplicar correção de cor nas imagens geradas em img2img, usando a imagem original como base.",
     "Save a copy of image before applying color correction to img2img results": "Salvar uma cópia das imagens geradas em img2img antes de aplicar a correção de cor",
     "With img2img, do exactly the amount of steps the slider specifies (normally you'd do less with less denoising).": "Durante gerações img2img, fazer examente o número de steps definidos na barra (normalmente você faz menos steps com denoising menor).",
@@ -379,6 +376,7 @@
     "Add model hash to generation information": "Adicionar hash do modelo para informação de geração",
     "Add model name to generation information": "Adicionar nome do modelo para informação de geração",
     "When reading generation parameters from text into UI (from PNG info or pasted text), do not change the selected model/checkpoint.": "Quando ler parâmetros de texto para a interface (de informações de PNG ou texto copiado), não alterar o modelo/intervalo selecionado.",
+    "Send seed when sending prompt or image to other interface": "Enviar seed quando enviar prompt ou imagem para outra interface",
     "Font for image grids that have text": "Fonte para grade de imagens que têm texto",
     "Enable full page image viewer": "Ativar visualizador de página inteira",
     "Show images zoomed in by default in full page image viewer": "Mostrar imagens com zoom por definição no visualizador de página inteira",
@@ -386,13 +384,17 @@
     "Quicksettings list": "Lista de configurações rapidas",
     "Localization (requires restart)": "Localização (precisa reiniciar)",
     "ar_AR": "ar_AR",
+    "de_DE": "de_DE",
     "es_ES": "es_ES",
-    "fr-FR": "fr-FR",
+    "fr_FR": "fr_FR",
+    "it_IT": "it_IT",
     "ja_JP": "ja_JP",
     "ko_KR": "ko_KR",
+    "pt_BR": "pt_BR",
     "ru_RU": "ru_RU",
     "tr_TR": "tr_TR",
     "zh_CN": "zh_CN",
+    "zh_TW": "zh_TW",
     "Sampler parameters": "Parâmetros de Amostragem",
     "Hide samplers in user interface (requires restart)": "Esconder amostragens na interface de usuário (precisa reiniciar)",
     "eta (noise multiplier) for DDIM": "tempo estimado (multiplicador de ruído) para DDIM",
@@ -408,6 +410,19 @@
     "Download localization template": "Baixar arquivo modelo de localização",
     "Reload custom script bodies (No ui updates, No restart)": "Recarregar scripts personalizados (Sem atualizar a interface, Sem reiniciar)",
     "Restart Gradio and Refresh components (Custom Scripts, ui.py, js and css only)": "Reiniciar Gradio e atualizar componentes (Scripts personalizados, ui.py, js e css)",
+    "Installed": "Instalado",
+    "Available": "Disponível",
+    "Install from URL": "Instalado de URL",
+    "Apply and restart UI": "Apicar e reiniciar a interface",
+    "Check for updates": "Procurar por atualizações",
+    "Extension": "Extensão",
+    "URL": "URL",
+    "Update": "Atualização",
+    "Load from:": "Carregar de:",
+    "Extension index URL": "Índice de extensão URL",
+    "URL for extension's git repository": "URL para repositório git da extensão",
+    "Local directory name": "Nome do diretório local",
+    "Install": "Instalar",
     "Prompt (press Ctrl+Enter or Alt+Enter to generate)": "Prompt (apertar Ctrl+Enter ou Alt+Enter para gerar)",
     "Negative prompt (press Ctrl+Enter or Alt+Enter to generate)": "Prompt Negativo (apertar Ctrl+Enter ou Alt+Enter para gerar)",
     "Add a random artist to the prompt.": "Adicionar um artista aleatório para o prompt.",
@@ -420,7 +435,7 @@
     "Do not do anything special": "Não faça nada de especial",
     "Which algorithm to use to produce the image": "O tipo de algoritmo para gerar imagens.",
     "Euler Ancestral - very creative, each can get a completely different picture depending on step count, setting steps to higher than 30-40 does not help": "Euler Ancestral - cria mais variações para as imagens em diferentes passos. Mais que 40 passos cancela o efeito.",
-    "Denoising Diffusion Implicit Models - Funciona melhor para inpainting.": "Denoising Diffusion Implicit Models - Funciona melhor para inpainting.",
+    "Denoising Diffusion Implicit Models - best at inpainting": "Denoising Diffusion Implicit Models - Funciona melhor para inpainting.",
     "Produce an image that can be tiled.": "Produz uma imagem que pode ser ladrilhada.",
     "Use a two step process to partially create an image at smaller resolution, upscale, and then improve details in it without changing composition": "Cria um processo em duas etapas, com uma imagem em baixa qualidade primeiro, aumenta a imagem e refina os detalhes sem alterar a composição da imagem",
     "Determines how little respect the algorithm should have for image's content. At 0, nothing will change, and at 1 you'll get an unrelated image. With values below 1.0, processing will take less steps than the Sampling Steps slider specifies.": "Quanto o algoritmo deve manter da imagem original. Em 0, nada muda. Em 1 o algoritmo ignora a imagem original. Valores menores que 1.0 demoram mais.",
@@ -438,7 +453,7 @@
     "Write image to a directory (default - log/images) and generation parameters into csv file.": "Salva a imagem no diretório padrão ou escolhido e cria um arquivo csv com os parâmetros da geração.",
     "Open images output directory": "Abre o diretório de saída de imagens.",
     "How much to blur the mask before processing, in pixels.": "Transição do contorno da máscara, em pixels.",
-    "What to put inside the masked area before processing it with Stable Diffusion.":  "O que vai dentro da máscara antes de processar.",
+    "What to put inside the masked area before processing it with Stable Diffusion.":  "O que vai dentro da máscara antes de processá-la com Stable Diffusion.",
 	"fill it with colors of the image": "Preenche usando as cores da imagem.",
     "keep whatever was there originally": "manter usando o que estava lá originalmente",
     "fill it with latent space noise": "Preenche com ruídos do espaço latente.",
@@ -463,6 +478,8 @@
     "Restore low quality faces using GFPGAN neural network": "Restaurar rostos de baixa qualidade usando a rede neural GFPGAN",
     "This regular expression will be used extract words from filename, and they will be joined using the option below into label text used for training. Leave empty to keep filename text as it is.": "Esta expressão regular vai retirar palavras do nome do arquivo e serão juntadas via regex usando a opção abaixo em etiquetas usadas em treinamento. Não mexer para manter os nomes como estão.",
     "This string will be used to join split words into a single line if the option above is enabled.": "Esta string será usada para unir palavras divididas em uma única linha se a opção acima estiver habilitada.",
+    "Only applies to inpainting models. Determines how strongly to mask off the original image for inpainting and img2img. 1.0 means fully masked, which is the default behaviour. 0.0 means a fully unmasked conditioning. Lower values will help preserve the overall composition of the image, but will struggle with large changes.": "Aplicável somente para modelos de inpaint. Determina quanto deve mascarar da imagem original para inpaint e img2img. 1.0 significa totalmente mascarado, que é o comportamento padrão. 0.0 significa uma condição totalmente não mascarada. Valores baixos ajudam a preservar a composição geral da imagem, mas vai encontrar dificuldades com grandes mudanças.",
     "List of setting names, separated by commas, for settings that should go to the quick access bar at the top, rather than the usual setting tab. See modules/shared.py for setting names. Requires restarting to apply.": "Lista de nomes de configurações, separados por vírgulas, para configurações que devem ir para a barra de acesso rápido na parte superior, em vez da guia de configuração usual. Veja modules/shared.py para nomes de configuração. Necessita reinicialização para aplicar.",
     "If this values is non-zero, it will be added to seed and used to initialize RNG for noises when using samplers with Eta. You can use this to produce even more variation of images, or you can use this to match images of other software if you know what you are doing.": "Se este valor for diferente de zero, ele será adicionado à seed e usado para inicializar o RNG para ruídos ao usar amostragens com Tempo Estimado. Você pode usar isso para produzir ainda mais variações de imagens ou pode usar isso para combinar imagens de outro software se souber o que está fazendo."
+    "Leave empty for auto": "Deixar desmarcado para automático"
 }

modules/api/api.py

@@ -1,12 +1,13 @@
+import base64
+import io
 import time
 import uvicorn
-from gradio.processing_utils import encode_pil_to_base64, decode_base64_to_file, decode_base64_to_image
+from gradio.processing_utils import decode_base64_to_file, decode_base64_to_image
 from fastapi import APIRouter, Depends, HTTPException
 import modules.shared as shared
-from modules import devices
 from modules.api.models import *
 from modules.processing import StableDiffusionProcessingTxt2Img, StableDiffusionProcessingImg2Img, process_images
-from modules.sd_samplers import all_samplers
+from modules.sd_samplers import all_samplers, sample_to_image, samples_to_image_grid
 from modules.extras import run_extras, run_pnginfo
 
 
@@ -29,6 +30,12 @@ def setUpscalers(req: dict):
     return reqDict
 
 
+def encode_pil_to_base64(image):
+    buffer = io.BytesIO()
+    image.save(buffer, format="png")
+    return base64.b64encode(buffer.getvalue())
+
+
 class Api:
     def __init__(self, app, queue_lock):
         self.router = APIRouter()
@@ -40,6 +47,7 @@ class Api:
         self.app.add_api_route("/sdapi/v1/extra-batch-images", self.extras_batch_images_api, methods=["POST"], response_model=ExtrasBatchImagesResponse)
         self.app.add_api_route("/sdapi/v1/png-info", self.pnginfoapi, methods=["POST"], response_model=PNGInfoResponse)
         self.app.add_api_route("/sdapi/v1/progress", self.progressapi, methods=["GET"], response_model=ProgressResponse)
+        self.app.add_api_route("/sdapi/v1/interrupt", self.interruptapi, methods=["POST"])
 
     def text2imgapi(self, txt2imgreq: StableDiffusionTxt2ImgProcessingAPI):
         sampler_index = sampler_to_index(txt2imgreq.sampler_index)
@@ -170,12 +178,19 @@ class Api:
 
         progress = min(progress, 1)
 
+        shared.state.set_current_image()
+
         current_image = None
         if shared.state.current_image and not req.skip_current_image:
             current_image = encode_pil_to_base64(shared.state.current_image)
 
         return ProgressResponse(progress=progress, eta_relative=eta_relative, state=shared.state.dict(), current_image=current_image)
 
+    def interruptapi(self):
+        shared.state.interrupt()
+
+        return {}
+
     def launch(self, server_name, port):
         self.app.include_router(self.router)
         uvicorn.run(self.app, host=server_name, port=port)

modules/esrgan_model.py

@@ -50,6 +50,7 @@ def mod2normal(state_dict):
 def resrgan2normal(state_dict, nb=23):
     # this code is copied from https://github.com/victorca25/iNNfer
     if "conv_first.weight" in state_dict and "body.0.rdb1.conv1.weight" in state_dict:
+        re8x = 0
         crt_net = {}
         items = []
         for k, v in state_dict.items():
@@ -75,10 +76,18 @@ def resrgan2normal(state_dict, nb=23):
         crt_net['model.3.bias'] = state_dict['conv_up1.bias']
         crt_net['model.6.weight'] = state_dict['conv_up2.weight']
         crt_net['model.6.bias'] = state_dict['conv_up2.bias']
-        crt_net['model.8.weight'] = state_dict['conv_hr.weight']
-        crt_net['model.8.bias'] = state_dict['conv_hr.bias']
-        crt_net['model.10.weight'] = state_dict['conv_last.weight']
-        crt_net['model.10.bias'] = state_dict['conv_last.bias']
+
+        if 'conv_up3.weight' in state_dict:
+            # modification supporting: https://github.com/ai-forever/Real-ESRGAN/blob/main/RealESRGAN/rrdbnet_arch.py
+            re8x = 3
+            crt_net['model.9.weight'] = state_dict['conv_up3.weight']
+            crt_net['model.9.bias'] = state_dict['conv_up3.bias']
+
+        crt_net[f'model.{8+re8x}.weight'] = state_dict['conv_hr.weight']
+        crt_net[f'model.{8+re8x}.bias'] = state_dict['conv_hr.bias']
+        crt_net[f'model.{10+re8x}.weight'] = state_dict['conv_last.weight']
+        crt_net[f'model.{10+re8x}.bias'] = state_dict['conv_last.bias']
+
         state_dict = crt_net
     return state_dict
 

modules/extensions.py

@@ -0,0 +1,83 @@
+import os
+import sys
+import traceback
+
+import git
+
+from modules import paths, shared
+
+
+extensions = []
+extensions_dir = os.path.join(paths.script_path, "extensions")
+
+
+def active():
+    return [x for x in extensions if x.enabled]
+
+
+class Extension:
+    def __init__(self, name, path, enabled=True):
+        self.name = name
+        self.path = path
+        self.enabled = enabled
+        self.status = ''
+        self.can_update = False
+
+        repo = None
+        try:
+            if os.path.exists(os.path.join(path, ".git")):
+                repo = git.Repo(path)
+        except Exception:
+            print(f"Error reading github repository info from {path}:", file=sys.stderr)
+            print(traceback.format_exc(), file=sys.stderr)
+
+        if repo is None or repo.bare:
+            self.remote = None
+        else:
+            self.remote = next(repo.remote().urls, None)
+            self.status = 'unknown'
+
+    def list_files(self, subdir, extension):
+        from modules import scripts
+
+        dirpath = os.path.join(self.path, subdir)
+        if not os.path.isdir(dirpath):
+            return []
+
+        res = []
+        for filename in sorted(os.listdir(dirpath)):
+            res.append(scripts.ScriptFile(self.path, filename, os.path.join(dirpath, filename)))
+
+        res = [x for x in res if os.path.splitext(x.path)[1].lower() == extension and os.path.isfile(x.path)]
+
+        return res
+
+    def check_updates(self):
+        repo = git.Repo(self.path)
+        for fetch in repo.remote().fetch("--dry-run"):
+            if fetch.flags != fetch.HEAD_UPTODATE:
+                self.can_update = True
+                self.status = "behind"
+                return
+
+        self.can_update = False
+        self.status = "latest"
+
+    def pull(self):
+        repo = git.Repo(self.path)
+        repo.remotes.origin.pull()
+
+
+def list_extensions():
+    extensions.clear()
+
+    if not os.path.isdir(extensions_dir):
+        return
+
+    for dirname in sorted(os.listdir(extensions_dir)):
+        path = os.path.join(extensions_dir, dirname)
+        if not os.path.isdir(path):
+            continue
+
+        extension = Extension(name=dirname, path=path, enabled=dirname not in shared.opts.disabled_extensions)
+        extensions.append(extension)

modules/extras.py

@@ -141,7 +141,7 @@ def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_
                             upscaling_resize_w, upscaling_resize_h, upscaling_crop)
             cache_key = LruCache.Key(image_hash=hash(np.array(image.getdata()).tobytes()),
                                      info_hash=hash(info),
-                                     args_hash=hash(upscale_args))
+                                     args_hash=hash((upscale_args, upscale_first)))
             cached_entry = cached_images.get(cache_key)
             if cached_entry is None:
                 res = upscale(image, *upscale_args)

modules/generation_parameters_copypaste.py

@@ -17,6 +17,11 @@ paste_fields = {}
 bind_list = []
 
 
+def reset():
+    paste_fields.clear()
+    bind_list.clear()
+
+
 def quote(text):
     if ',' not in str(text):
         return text

modules/images.py

@@ -510,8 +510,9 @@ def save_image(image, path, basename, seed=None, prompt=None, extension='png', i
 
     if extension.lower() == '.png':
         pnginfo_data = PngImagePlugin.PngInfo()
-        for k, v in params.pnginfo.items():
-            pnginfo_data.add_text(k, str(v))
+        if opts.enable_pnginfo:
+            for k, v in params.pnginfo.items():
+                pnginfo_data.add_text(k, str(v))
 
         image.save(fullfn, quality=opts.jpeg_quality, pnginfo=pnginfo_data)
 

modules/img2img.py

@@ -55,6 +55,7 @@ def process_batch(p, input_dir, output_dir, args):
                 filename = f"{left}-{n}{right}"
 
             if not save_normally:
+                os.makedirs(output_dir, exist_ok=True)
                 processed_image.save(os.path.join(output_dir, filename))
 
 
@@ -80,7 +81,8 @@ def img2img(mode: int, prompt: str, negative_prompt: str, prompt_style: str, pro
         mask = None
 
     # Use the EXIF orientation of photos taken by smartphones.
-    image = ImageOps.exif_transpose(image) 
+    if image is not None:
+        image = ImageOps.exif_transpose(image) 
 
     assert 0. <= denoising_strength <= 1., 'can only work with strength in [0.0, 1.0]'
 
@@ -136,6 +138,8 @@ def img2img(mode: int, prompt: str, negative_prompt: str, prompt_style: str, pro
         if processed is None:
             processed = process_images(p)
 
+    p.close()
+
     shared.total_tqdm.clear()
 
     generation_info_js = processed.js()

modules/interrogate.py

@@ -56,9 +56,9 @@ class InterrogateModels:
         import clip
 
         if self.running_on_cpu:
-            model, preprocess = clip.load(clip_model_name, device="cpu")
+            model, preprocess = clip.load(clip_model_name, device="cpu", download_root=shared.cmd_opts.clip_models_path)
         else:
-            model, preprocess = clip.load(clip_model_name)
+            model, preprocess = clip.load(clip_model_name, download_root=shared.cmd_opts.clip_models_path)
 
         model.eval()
         model = model.to(devices.device_interrogate)

modules/lowvram.py

@@ -38,13 +38,18 @@ def setup_for_low_vram(sd_model, use_medvram):
     # see below for register_forward_pre_hook;
     # first_stage_model does not use forward(), it uses encode/decode, so register_forward_pre_hook is
     # useless here, and we just replace those methods
-    def first_stage_model_encode_wrap(self, encoder, x):
-        send_me_to_gpu(self, None)
-        return encoder(x)
 
-    def first_stage_model_decode_wrap(self, decoder, z):
-        send_me_to_gpu(self, None)
-        return decoder(z)
+    first_stage_model = sd_model.first_stage_model
+    first_stage_model_encode = sd_model.first_stage_model.encode
+    first_stage_model_decode = sd_model.first_stage_model.decode
+
+    def first_stage_model_encode_wrap(x):
+        send_me_to_gpu(first_stage_model, None)
+        return first_stage_model_encode(x)
+
+    def first_stage_model_decode_wrap(z):
+        send_me_to_gpu(first_stage_model, None)
+        return first_stage_model_decode(z)
 
     # remove three big modules, cond, first_stage, and unet from the model and then
     # send the model to GPU. Then put modules back. the modules will be in CPU.
@@ -56,8 +61,8 @@ def setup_for_low_vram(sd_model, use_medvram):
     # register hooks for those the first two models
     sd_model.cond_stage_model.transformer.register_forward_pre_hook(send_me_to_gpu)
     sd_model.first_stage_model.register_forward_pre_hook(send_me_to_gpu)
-    sd_model.first_stage_model.encode = lambda x, en=sd_model.first_stage_model.encode: first_stage_model_encode_wrap(sd_model.first_stage_model, en, x)
-    sd_model.first_stage_model.decode = lambda z, de=sd_model.first_stage_model.decode: first_stage_model_decode_wrap(sd_model.first_stage_model, de, z)
+    sd_model.first_stage_model.encode = first_stage_model_encode_wrap
+    sd_model.first_stage_model.decode = first_stage_model_decode_wrap
     parents[sd_model.cond_stage_model.transformer] = sd_model.cond_stage_model
 
     if use_medvram:

modules/modelloader.py

@@ -85,6 +85,9 @@ def cleanup_models():
     src_path = os.path.join(root_path, "ESRGAN")
     dest_path = os.path.join(models_path, "ESRGAN")
     move_files(src_path, dest_path)
+    src_path = os.path.join(models_path, "BSRGAN")
+    dest_path = os.path.join(models_path, "ESRGAN")
+    move_files(src_path, dest_path, ".pth")
     src_path = os.path.join(root_path, "gfpgan")
     dest_path = os.path.join(models_path, "GFPGAN")
     move_files(src_path, dest_path)

modules/processing.py

@@ -199,9 +199,13 @@ class StableDiffusionProcessing():
     def init(self, all_prompts, all_seeds, all_subseeds):
         pass
 
-    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength):
+    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts):
         raise NotImplementedError()
 
+    def close(self):
+        self.sd_model = None
+        self.sampler = None
+
 
 class Processed:
     def __init__(self, p: StableDiffusionProcessing, images_list, seed=-1, info="", subseed=None, all_prompts=None, all_seeds=None, all_subseeds=None, index_of_first_image=0, infotexts=None):
@@ -517,7 +521,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed:
                 shared.state.job = f"Batch {n+1} out of {p.n_iter}"
 
             with devices.autocast():
-                samples_ddim = p.sample(conditioning=c, unconditional_conditioning=uc, seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength)
+                samples_ddim = p.sample(conditioning=c, unconditional_conditioning=uc, seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength, prompts=prompts)
 
             samples_ddim = samples_ddim.to(devices.dtype_vae)
             x_samples_ddim = decode_first_stage(p.sd_model, samples_ddim)
@@ -645,7 +649,7 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
             self.truncate_x = int(self.firstphase_width - firstphase_width_truncated) // opt_f
             self.truncate_y = int(self.firstphase_height - firstphase_height_truncated) // opt_f
 
-    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength):
+    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts):
         self.sampler = sd_samplers.create_sampler_with_index(sd_samplers.samplers, self.sampler_index, self.sd_model)
 
         if not self.enable_hr:
@@ -658,9 +662,21 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
 
         samples = samples[:, :, self.truncate_y//2:samples.shape[2]-self.truncate_y//2, self.truncate_x//2:samples.shape[3]-self.truncate_x//2]
 
+        """saves image before applying hires fix, if enabled in options; takes as an arguyment either an image or batch with latent space images"""
+        def save_intermediate(image, index):
+            if not opts.save or self.do_not_save_samples or not opts.save_images_before_highres_fix:
+                return
+
+            if not isinstance(image, Image.Image):
+                image = sd_samplers.sample_to_image(image, index)
+
+            images.save_image(image, self.outpath_samples, "", seeds[index], prompts[index], opts.samples_format, suffix="-before-highres-fix")
+
         if opts.use_scale_latent_for_hires_fix:
             samples = torch.nn.functional.interpolate(samples, size=(self.height // opt_f, self.width // opt_f), mode="bilinear")
 
+            for i in range(samples.shape[0]):
+                save_intermediate(samples, i)
         else:
             decoded_samples = decode_first_stage(self.sd_model, samples)
             lowres_samples = torch.clamp((decoded_samples + 1.0) / 2.0, min=0.0, max=1.0)
@@ -670,6 +686,9 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
                 x_sample = 255. * np.moveaxis(x_sample.cpu().numpy(), 0, 2)
                 x_sample = x_sample.astype(np.uint8)
                 image = Image.fromarray(x_sample)
+
+                save_intermediate(image, i)
+
                 image = images.resize_image(0, image, self.width, self.height)
                 image = np.array(image).astype(np.float32) / 255.0
                 image = np.moveaxis(image, 2, 0)
@@ -827,8 +846,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing):
 
         self.image_conditioning = self.img2img_image_conditioning(image, self.init_latent, self.image_mask)
 
-
-    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength):
+    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts):
         x = create_random_tensors([opt_C, self.height // opt_f, self.width // opt_f], seeds=seeds, subseeds=subseeds, subseed_strength=self.subseed_strength, seed_resize_from_h=self.seed_resize_from_h, seed_resize_from_w=self.seed_resize_from_w, p=self)
 
         samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning)
@@ -839,4 +857,4 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing):
         del x
         devices.torch_gc()
 
-        return samples
+        return samples

modules/safe.py

@@ -32,7 +32,7 @@ class RestrictedUnpickler(pickle.Unpickler):
             return getattr(collections, name)
         if module == 'torch._utils' and name in ['_rebuild_tensor_v2', '_rebuild_parameter']:
             return getattr(torch._utils, name)
-        if module == 'torch' and name in ['FloatStorage', 'HalfStorage', 'IntStorage', 'LongStorage', 'DoubleStorage']:
+        if module == 'torch' and name in ['FloatStorage', 'HalfStorage', 'IntStorage', 'LongStorage', 'DoubleStorage', 'ByteStorage']:
             return getattr(torch, name)
         if module == 'torch.nn.modules.container' and name in ['ParameterDict']:
             return getattr(torch.nn.modules.container, name)

modules/script_callbacks.py

@@ -2,7 +2,10 @@ import sys
 import traceback
 from collections import namedtuple
 import inspect
+from typing import Optional
 
+from fastapi import FastAPI
+from gradio import Blocks
 
 def report_exception(c, job):
     print(f"Error executing callback {job} for {c.script}", file=sys.stderr)
@@ -24,12 +27,32 @@ class ImageSaveParams:
         """dictionary with parameters for image's PNG info data; infotext will have the key 'parameters'"""
 
 
+class CFGDenoiserParams:
+    def __init__(self, x, image_cond, sigma, sampling_step, total_sampling_steps):
+        self.x = x
+        """Latent image representation in the process of being denoised"""
+        
+        self.image_cond = image_cond
+        """Conditioning image"""
+        
+        self.sigma = sigma
+        """Current sigma noise step value"""
+        
+        self.sampling_step = sampling_step
+        """Current Sampling step number"""
+        
+        self.total_sampling_steps = total_sampling_steps
+        """Total number of sampling steps planned"""
+
+
 ScriptCallback = namedtuple("ScriptCallback", ["script", "callback"])
+callbacks_app_started = []
 callbacks_model_loaded = []
 callbacks_ui_tabs = []
 callbacks_ui_settings = []
 callbacks_before_image_saved = []
 callbacks_image_saved = []
+callbacks_cfg_denoiser = []
 
 
 def clear_callbacks():
@@ -38,6 +61,14 @@ def clear_callbacks():
     callbacks_ui_settings.clear()
     callbacks_before_image_saved.clear()
     callbacks_image_saved.clear()
+    callbacks_cfg_denoiser.clear()
+
+def app_started_callback(demo: Optional[Blocks], app: FastAPI):
+    for c in callbacks_app_started:
+        try:
+            c.callback(demo, app)
+        except Exception:
+            report_exception(c, 'app_started_callback')
 
 
 def model_loaded_callback(sd_model):
@@ -69,7 +100,7 @@ def ui_settings_callback():
 
 
 def before_image_saved_callback(params: ImageSaveParams):
-    for c in callbacks_image_saved:
+    for c in callbacks_before_image_saved:
         try:
             c.callback(params)
         except Exception:
@@ -84,6 +115,14 @@ def image_saved_callback(params: ImageSaveParams):
             report_exception(c, 'image_saved_callback')
 
 
+def cfg_denoiser_callback(params: CFGDenoiserParams):
+    for c in callbacks_cfg_denoiser:
+        try:
+            c.callback(params)
+        except Exception:
+            report_exception(c, 'cfg_denoiser_callback')
+
+
 def add_callback(callbacks, fun):
     stack = [x for x in inspect.stack() if x.filename != __file__]
     filename = stack[0].filename if len(stack) > 0 else 'unknown file'
@@ -91,6 +130,12 @@ def add_callback(callbacks, fun):
     callbacks.append(ScriptCallback(filename, fun))
 
 
+def on_app_started(callback):
+    """register a function to be called when the webui started, the gradio `Block` component and
+    fastapi `FastAPI` object are passed as the arguments"""
+    add_callback(callbacks_app_started, callback)
+
+
 def on_model_loaded(callback):
     """register a function to be called when the stable diffusion model is created; the model is
     passed as an argument"""
@@ -130,3 +175,12 @@ def on_image_saved(callback):
         - params: ImageSaveParams - parameters the image was saved with. Changing fields in this object does nothing.
     """
     add_callback(callbacks_image_saved, callback)
+
+
+def on_cfg_denoiser(callback):
+    """register a function to be called in the kdiffussion cfg_denoiser method after building the inner model inputs.
+    The callback is called with one argument:
+        - params: CFGDenoiserParams - parameters to be passed to the inner model and sampling state details.
+    """
+    add_callback(callbacks_cfg_denoiser, callback)
+

modules/scripts.py

@@ -7,7 +7,7 @@ import modules.ui as ui
 import gradio as gr
 
 from modules.processing import StableDiffusionProcessing
-from modules import shared, paths, script_callbacks
+from modules import shared, paths, script_callbacks, extensions
 
 AlwaysVisible = object()
 
@@ -107,17 +107,8 @@ def list_scripts(scriptdirname, extension):
         for filename in sorted(os.listdir(basedir)):
             scripts_list.append(ScriptFile(paths.script_path, filename, os.path.join(basedir, filename)))
 
-    extdir = os.path.join(paths.script_path, "extensions")
-    if os.path.exists(extdir):
-        for dirname in sorted(os.listdir(extdir)):
-            dirpath = os.path.join(extdir, dirname)
-            scriptdirpath = os.path.join(dirpath, scriptdirname)
-
-            if not os.path.isdir(scriptdirpath):
-                continue
-
-            for filename in sorted(os.listdir(scriptdirpath)):
-                scripts_list.append(ScriptFile(dirpath, filename, os.path.join(scriptdirpath, filename)))
+    for ext in extensions.active():
+        scripts_list += ext.list_files(scriptdirname, extension)
 
     scripts_list = [x for x in scripts_list if os.path.splitext(x.path)[1].lower() == extension and os.path.isfile(x.path)]
 
@@ -127,11 +118,7 @@ def list_scripts(scriptdirname, extension):
 def list_files_with_name(filename):
     res = []
 
-    dirs = [paths.script_path]
-
-    extdir = os.path.join(paths.script_path, "extensions")
-    if os.path.exists(extdir):
-        dirs += [os.path.join(extdir, d) for d in sorted(os.listdir(extdir))]
+    dirs = [paths.script_path] + [ext.path for ext in extensions.active()]
 
     for dirpath in dirs:
         if not os.path.isdir(dirpath):

modules/sd_hijack.py

@@ -94,6 +94,10 @@ class StableDiffusionModelHijack:
         if type(model_embeddings.token_embedding) == EmbeddingsWithFixes:
             model_embeddings.token_embedding = model_embeddings.token_embedding.wrapped
 
+        self.layers = None
+        self.circular_enabled = False
+        self.clip = None
+
     def apply_circular(self, enable):
         if self.circular_enabled == enable:
             return

modules/sd_models.py

@@ -1,6 +1,7 @@
 import collections
 import os.path
 import sys
+import gc
 from collections import namedtuple
 import torch
 import re
@@ -8,7 +9,7 @@ from omegaconf import OmegaConf
 
 from ldm.util import instantiate_from_config
 
-from modules import shared, modelloader, devices, script_callbacks
+from modules import shared, modelloader, devices, script_callbacks, sd_vae
 from modules.paths import models_path
 from modules.sd_hijack_inpainting import do_inpainting_hijack, should_hijack_inpainting
 
@@ -158,14 +159,15 @@ def get_state_dict_from_checkpoint(pl_sd):
     return pl_sd
 
 
-vae_ignore_keys = {"model_ema.decay", "model_ema.num_updates"}
-
-
-def load_model_weights(model, checkpoint_info):
+def load_model_weights(model, checkpoint_info, vae_file="auto"):
     checkpoint_file = checkpoint_info.filename
     sd_model_hash = checkpoint_info.hash
 
-    if checkpoint_info not in checkpoints_loaded:
+    vae_file = sd_vae.resolve_vae(checkpoint_file, vae_file=vae_file)
+
+    checkpoint_key = checkpoint_info
+
+    if checkpoint_key not in checkpoints_loaded:
         print(f"Loading weights [{sd_model_hash}] from {checkpoint_file}")
 
         pl_sd = torch.load(checkpoint_file, map_location=shared.weight_load_location)
@@ -181,37 +183,38 @@ def load_model_weights(model, checkpoint_info):
             model.to(memory_format=torch.channels_last)
 
         if not shared.cmd_opts.no_half:
+            vae = model.first_stage_model
+
+            # with --no-half-vae, remove VAE from model when doing half() to prevent its weights from being converted to float16
+            if shared.cmd_opts.no_half_vae:
+                model.first_stage_model = None
+
             model.half()
+            model.first_stage_model = vae
 
         devices.dtype = torch.float32 if shared.cmd_opts.no_half else torch.float16
         devices.dtype_vae = torch.float32 if shared.cmd_opts.no_half or shared.cmd_opts.no_half_vae else torch.float16
 
-        vae_file = os.path.splitext(checkpoint_file)[0] + ".vae.pt"
-
-        if not os.path.exists(vae_file) and shared.cmd_opts.vae_path is not None:
-            vae_file = shared.cmd_opts.vae_path
-
-        if os.path.exists(vae_file):
-            print(f"Loading VAE weights from: {vae_file}")
-            vae_ckpt = torch.load(vae_file, map_location=shared.weight_load_location)
-            vae_dict = {k: v for k, v in vae_ckpt["state_dict"].items() if k[0:4] != "loss" and k not in vae_ignore_keys}
-            model.first_stage_model.load_state_dict(vae_dict)
-
         model.first_stage_model.to(devices.dtype_vae)
 
         if shared.opts.sd_checkpoint_cache > 0:
-            checkpoints_loaded[checkpoint_info] = model.state_dict().copy()
+            # if PR #4035 were to get merged, restore base VAE first before caching
+            checkpoints_loaded[checkpoint_key] = model.state_dict().copy()
             while len(checkpoints_loaded) > shared.opts.sd_checkpoint_cache:
                 checkpoints_loaded.popitem(last=False)  # LRU
+
     else:
-        print(f"Loading weights [{sd_model_hash}] from cache")
-        checkpoints_loaded.move_to_end(checkpoint_info)
-        model.load_state_dict(checkpoints_loaded[checkpoint_info])
+        vae_name = sd_vae.get_filename(vae_file)
+        print(f"Loading weights [{sd_model_hash}] with {vae_name} VAE from cache")
+        checkpoints_loaded.move_to_end(checkpoint_key)
+        model.load_state_dict(checkpoints_loaded[checkpoint_key])
 
     model.sd_model_hash = sd_model_hash
     model.sd_model_checkpoint = checkpoint_file
     model.sd_checkpoint_info = checkpoint_info
 
+    sd_vae.load_vae(model, vae_file)
+
 
 def load_model(checkpoint_info=None):
     from modules import lowvram, sd_hijack
@@ -220,6 +223,12 @@ def load_model(checkpoint_info=None):
     if checkpoint_info.config != shared.cmd_opts.config:
         print(f"Loading config from: {checkpoint_info.config}")
 
+    if shared.sd_model:
+        sd_hijack.model_hijack.undo_hijack(shared.sd_model)
+        shared.sd_model = None
+        gc.collect()
+        devices.torch_gc()
+
     sd_config = OmegaConf.load(checkpoint_info.config)
     
     if should_hijack_inpainting(checkpoint_info):
@@ -233,6 +242,7 @@ def load_model(checkpoint_info=None):
         checkpoint_info = checkpoint_info._replace(config=checkpoint_info.config.replace(".yaml", "-inpainting.yaml"))
 
     do_inpainting_hijack()
+
     sd_model = instantiate_from_config(sd_config.model)
     load_model_weights(sd_model, checkpoint_info)
 
@@ -252,14 +262,18 @@ def load_model(checkpoint_info=None):
     return sd_model
 
 
-def reload_model_weights(sd_model, info=None):
+def reload_model_weights(sd_model=None, info=None):
     from modules import lowvram, devices, sd_hijack
     checkpoint_info = info or select_checkpoint()
+ 
+    if not sd_model:
+        sd_model = shared.sd_model
 
     if sd_model.sd_model_checkpoint == checkpoint_info.filename:
         return
 
     if sd_model.sd_checkpoint_info.config != checkpoint_info.config or should_hijack_inpainting(checkpoint_info) != should_hijack_inpainting(sd_model.sd_checkpoint_info):
+        del sd_model
         checkpoints_loaded.clear()
         load_model(checkpoint_info)
         return shared.sd_model

modules/sd_samplers.py

@@ -1,5 +1,6 @@
 from collections import namedtuple
 import numpy as np
+from math import floor
 import torch
 import tqdm
 from PIL import Image
@@ -11,6 +12,7 @@ from modules import prompt_parser, devices, processing, images
 
 from modules.shared import opts, cmd_opts, state
 import modules.shared as shared
+from modules.script_callbacks import CFGDenoiserParams, cfg_denoiser_callback
 
 
 SamplerData = namedtuple('SamplerData', ['name', 'constructor', 'aliases', 'options'])
@@ -91,8 +93,8 @@ def single_sample_to_image(sample):
     return Image.fromarray(x_sample)
 
 
-def sample_to_image(samples):
-    return single_sample_to_image(samples[0])
+def sample_to_image(samples, index=0):
+    return single_sample_to_image(samples[index])
 
 
 def samples_to_image_grid(samples):
@@ -205,17 +207,22 @@ class VanillaStableDiffusionSampler:
         self.mask = p.mask if hasattr(p, 'mask') else None
         self.nmask = p.nmask if hasattr(p, 'nmask') else None
 
+
+    def adjust_steps_if_invalid(self, p, num_steps):
+        if  (self.config.name == 'DDIM' and p.ddim_discretize == 'uniform') or (self.config.name == 'PLMS'):
+            valid_step = 999 / (1000 // num_steps)
+            if valid_step == floor(valid_step):
+                return int(valid_step) + 1
+        
+        return num_steps
+
+
     def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning, steps=None, image_conditioning=None):
         steps, t_enc = setup_img2img_steps(p, steps)
-
+        steps = self.adjust_steps_if_invalid(p, steps)
         self.initialize(p)
 
-        # existing code fails with certain step counts, like 9
-        try:
-            self.sampler.make_schedule(ddim_num_steps=steps,  ddim_eta=self.eta, ddim_discretize=p.ddim_discretize, verbose=False)
-        except Exception:
-            self.sampler.make_schedule(ddim_num_steps=steps+1, ddim_eta=self.eta, ddim_discretize=p.ddim_discretize, verbose=False)
-
+        self.sampler.make_schedule(ddim_num_steps=steps, ddim_eta=self.eta, ddim_discretize=p.ddim_discretize, verbose=False)
         x1 = self.sampler.stochastic_encode(x, torch.tensor([t_enc] * int(x.shape[0])).to(shared.device), noise=noise)
 
         self.init_latent = x
@@ -239,18 +246,14 @@ class VanillaStableDiffusionSampler:
         self.last_latent = x
         self.step = 0
 
-        steps = steps or p.steps
+        steps = self.adjust_steps_if_invalid(p, steps or p.steps)
 
         # Wrap the conditioning models with additional image conditioning for inpainting model
         if image_conditioning is not None:
             conditioning = {"c_concat": [image_conditioning], "c_crossattn": [conditioning]}
             unconditional_conditioning = {"c_concat": [image_conditioning], "c_crossattn": [unconditional_conditioning]}
 
-        # existing code fails with certain step counts, like 9
-        try:
-            samples_ddim = self.launch_sampling(steps, lambda: self.sampler.sample(S=steps, conditioning=conditioning, batch_size=int(x.shape[0]), shape=x[0].shape, verbose=False, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning, x_T=x, eta=self.eta)[0])
-        except Exception:
-            samples_ddim = self.launch_sampling(steps, lambda: self.sampler.sample(S=steps+1, conditioning=conditioning, batch_size=int(x.shape[0]), shape=x[0].shape, verbose=False, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning, x_T=x, eta=self.eta)[0])
+        samples_ddim = self.launch_sampling(steps, lambda: self.sampler.sample(S=steps, conditioning=conditioning, batch_size=int(x.shape[0]), shape=x[0].shape, verbose=False, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning, x_T=x, eta=self.eta)[0])
 
         return samples_ddim
 
@@ -278,6 +281,12 @@ class CFGDenoiser(torch.nn.Module):
         image_cond_in = torch.cat([torch.stack([image_cond[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [image_cond])
         sigma_in = torch.cat([torch.stack([sigma[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [sigma])
 
+        denoiser_params = CFGDenoiserParams(x_in, image_cond_in, sigma_in, state.sampling_step, state.sampling_steps)
+        cfg_denoiser_callback(denoiser_params)
+        x_in = denoiser_params.x
+        image_cond_in = denoiser_params.image_cond
+        sigma_in = denoiser_params.sigma
+
         if tensor.shape[1] == uncond.shape[1]:
             cond_in = torch.cat([tensor, uncond])
 

modules/sd_vae.py

@@ -0,0 +1,207 @@
+import torch
+import os
+from collections import namedtuple
+from modules import shared, devices, script_callbacks
+from modules.paths import models_path
+import glob
+
+
+model_dir = "Stable-diffusion"
+model_path = os.path.abspath(os.path.join(models_path, model_dir))
+vae_dir = "VAE"
+vae_path = os.path.abspath(os.path.join(models_path, vae_dir))
+
+
+vae_ignore_keys = {"model_ema.decay", "model_ema.num_updates"}
+
+
+default_vae_dict = {"auto": "auto", "None": "None"}
+default_vae_list = ["auto", "None"]
+
+
+default_vae_values = [default_vae_dict[x] for x in default_vae_list]
+vae_dict = dict(default_vae_dict)
+vae_list = list(default_vae_list)
+first_load = True
+
+
+base_vae = None
+loaded_vae_file = None
+checkpoint_info = None
+
+
+def get_base_vae(model):
+    if base_vae is not None and checkpoint_info == model.sd_checkpoint_info and model:
+        return base_vae
+    return None
+
+
+def store_base_vae(model):
+    global base_vae, checkpoint_info
+    if checkpoint_info != model.sd_checkpoint_info:
+        base_vae = model.first_stage_model.state_dict().copy()
+        checkpoint_info = model.sd_checkpoint_info
+
+
+def delete_base_vae():
+    global base_vae, checkpoint_info
+    base_vae = None
+    checkpoint_info = None
+
+
+def restore_base_vae(model):
+    global base_vae, checkpoint_info
+    if base_vae is not None and checkpoint_info == model.sd_checkpoint_info:
+        load_vae_dict(model, base_vae)
+    delete_base_vae()
+
+
+def get_filename(filepath):
+    return os.path.splitext(os.path.basename(filepath))[0]
+
+
+def refresh_vae_list(vae_path=vae_path, model_path=model_path):
+    global vae_dict, vae_list
+    res = {}
+    candidates = [
+        *glob.iglob(os.path.join(model_path, '**/*.vae.ckpt'), recursive=True),
+        *glob.iglob(os.path.join(model_path, '**/*.vae.pt'), recursive=True),
+        *glob.iglob(os.path.join(vae_path, '**/*.ckpt'), recursive=True),
+        *glob.iglob(os.path.join(vae_path, '**/*.pt'), recursive=True)
+    ]
+    if shared.cmd_opts.vae_path is not None and os.path.isfile(shared.cmd_opts.vae_path):
+        candidates.append(shared.cmd_opts.vae_path)
+    for filepath in candidates:
+        name = get_filename(filepath)
+        res[name] = filepath
+    vae_list.clear()
+    vae_list.extend(default_vae_list)
+    vae_list.extend(list(res.keys()))
+    vae_dict.clear()
+    vae_dict.update(res)
+    vae_dict.update(default_vae_dict)
+    return vae_list
+
+
+def resolve_vae(checkpoint_file, vae_file="auto"):
+    global first_load, vae_dict, vae_list
+
+    # if vae_file argument is provided, it takes priority, but not saved
+    if vae_file and vae_file not in default_vae_list:
+        if not os.path.isfile(vae_file):
+            vae_file = "auto"
+            print("VAE provided as function argument doesn't exist")
+    # for the first load, if vae-path is provided, it takes priority, saved, and failure is reported
+    if first_load and shared.cmd_opts.vae_path is not None:
+        if os.path.isfile(shared.cmd_opts.vae_path):
+            vae_file = shared.cmd_opts.vae_path
+            shared.opts.data['sd_vae'] = get_filename(vae_file)
+        else:
+            print("VAE provided as command line argument doesn't exist")
+    # else, we load from settings
+    if vae_file == "auto" and shared.opts.sd_vae is not None:
+        # if saved VAE settings isn't recognized, fallback to auto
+        vae_file = vae_dict.get(shared.opts.sd_vae, "auto")
+        # if VAE selected but not found, fallback to auto
+        if vae_file not in default_vae_values and not os.path.isfile(vae_file):
+            vae_file = "auto"
+            print("Selected VAE doesn't exist")
+    # vae-path cmd arg takes priority for auto
+    if vae_file == "auto" and shared.cmd_opts.vae_path is not None:
+        if os.path.isfile(shared.cmd_opts.vae_path):
+            vae_file = shared.cmd_opts.vae_path
+            print("Using VAE provided as command line argument")
+    # if still not found, try look for ".vae.pt" beside model
+    model_path = os.path.splitext(checkpoint_file)[0]
+    if vae_file == "auto":
+        vae_file_try = model_path + ".vae.pt"
+        if os.path.isfile(vae_file_try):
+            vae_file = vae_file_try
+            print("Using VAE found beside selected model")
+    # if still not found, try look for ".vae.ckpt" beside model
+    if vae_file == "auto":
+        vae_file_try = model_path + ".vae.ckpt"
+        if os.path.isfile(vae_file_try):
+            vae_file = vae_file_try
+            print("Using VAE found beside selected model")
+    # No more fallbacks for auto
+    if vae_file == "auto":
+        vae_file = None
+    # Last check, just because
+    if vae_file and not os.path.exists(vae_file):
+        vae_file = None
+
+    return vae_file
+
+
+def load_vae(model, vae_file=None):
+    global first_load, vae_dict, vae_list, loaded_vae_file
+    # save_settings = False
+
+    if vae_file:
+        print(f"Loading VAE weights from: {vae_file}")
+        vae_ckpt = torch.load(vae_file, map_location=shared.weight_load_location)
+        vae_dict_1 = {k: v for k, v in vae_ckpt["state_dict"].items() if k[0:4] != "loss" and k not in vae_ignore_keys}
+        load_vae_dict(model, vae_dict_1)
+
+        # If vae used is not in dict, update it
+        # It will be removed on refresh though
+        vae_opt = get_filename(vae_file)
+        if vae_opt not in vae_dict:
+            vae_dict[vae_opt] = vae_file
+            vae_list.append(vae_opt)
+
+    loaded_vae_file = vae_file
+
+    """
+    # Save current VAE to VAE settings, maybe? will it work?
+    if save_settings:
+        if vae_file is None:
+            vae_opt = "None"
+
+        # shared.opts.sd_vae = vae_opt
+    """
+
+    first_load = False
+
+
+# don't call this from outside
+def load_vae_dict(model, vae_dict_1=None):
+    if vae_dict_1:
+        store_base_vae(model)
+        model.first_stage_model.load_state_dict(vae_dict_1)
+    else:
+        restore_base_vae()
+    model.first_stage_model.to(devices.dtype_vae)
+
+
+def reload_vae_weights(sd_model=None, vae_file="auto"):
+    from modules import lowvram, devices, sd_hijack
+
+    if not sd_model:
+        sd_model = shared.sd_model
+
+    checkpoint_info = sd_model.sd_checkpoint_info
+    checkpoint_file = checkpoint_info.filename
+    vae_file = resolve_vae(checkpoint_file, vae_file=vae_file)
+
+    if loaded_vae_file == vae_file:
+        return
+
+    if shared.cmd_opts.lowvram or shared.cmd_opts.medvram:
+        lowvram.send_everything_to_cpu()
+    else:
+        sd_model.to(devices.cpu)
+
+    sd_hijack.model_hijack.undo_hijack(sd_model)
+
+    load_vae(sd_model, vae_file)
+
+    sd_hijack.model_hijack.hijack(sd_model)
+    script_callbacks.model_loaded_callback(sd_model)
+
+    if not shared.cmd_opts.lowvram and not shared.cmd_opts.medvram:
+        sd_model.to(devices.device)
+
+    print(f"VAE Weights loaded.")
+    return sd_model

modules/shared.py

@@ -4,6 +4,7 @@ import json
 import os
 import sys
 from collections import OrderedDict
+import time
 
 import gradio as gr
 import tqdm
@@ -14,7 +15,7 @@ import modules.memmon
 import modules.sd_models
 import modules.styles
 import modules.devices as devices
-from modules import sd_samplers, sd_models, localization
+from modules import sd_samplers, sd_models, localization, sd_vae
 from modules.hypernetworks import hypernetwork
 from modules.paths import models_path, script_path, sd_path
 
@@ -40,7 +41,7 @@ parser.add_argument("--lowram", action='store_true', help="load stable diffusion
 parser.add_argument("--always-batch-cond-uncond", action='store_true', help="disables cond/uncond batching that is enabled to save memory with --medvram or --lowvram")
 parser.add_argument("--unload-gfpgan", action='store_true', help="does not do anything.")
 parser.add_argument("--precision", type=str, help="evaluate at this precision", choices=["full", "autocast"], default="autocast")
-parser.add_argument("--share", action='store_true', help="use share=True for gradio and make the UI accessible through their site (doesn't work for me but you might have better luck)")
+parser.add_argument("--share", action='store_true', help="use share=True for gradio and make the UI accessible through their site")
 parser.add_argument("--ngrok", type=str, help="ngrok authtoken, alternative to gradio --share", default=None)
 parser.add_argument("--ngrok-region", type=str, help="The region in which ngrok should start.", default="us")
 parser.add_argument("--codeformer-models-path", type=str, help="Path to directory with codeformer model file(s).", default=os.path.join(models_path, 'Codeformer'))
@@ -51,6 +52,7 @@ parser.add_argument("--realesrgan-models-path", type=str, help="Path to director
 parser.add_argument("--scunet-models-path", type=str, help="Path to directory with ScuNET model file(s).", default=os.path.join(models_path, 'ScuNET'))
 parser.add_argument("--swinir-models-path", type=str, help="Path to directory with SwinIR model file(s).", default=os.path.join(models_path, 'SwinIR'))
 parser.add_argument("--ldsr-models-path", type=str, help="Path to directory with LDSR model file(s).", default=os.path.join(models_path, 'LDSR'))
+parser.add_argument("--clip-models-path", type=str, help="Path to directory with CLIP model file(s).", default=None)
 parser.add_argument("--xformers", action='store_true', help="enable xformers for cross attention layers")
 parser.add_argument("--force-enable-xformers", action='store_true', help="enable xformers for cross attention layers regardless of whether the checking code thinks you can run it; do not make bug reports if this fails to work")
 parser.add_argument("--deepdanbooru", action='store_true', help="enable deepdanbooru interrogator")
@@ -97,6 +99,8 @@ restricted_opts = {
     "outdir_save",
 }
 
+cmd_opts.disable_extension_access = cmd_opts.share or cmd_opts.listen
+
 devices.device, devices.device_interrogate, devices.device_gfpgan, devices.device_swinir, devices.device_esrgan, devices.device_scunet, devices.device_codeformer = \
 (devices.cpu if any(y in cmd_opts.use_cpu for y in [x, 'all']) else devices.get_optimal_device() for x in ['sd', 'interrogate', 'gfpgan', 'swinir', 'esrgan', 'scunet', 'codeformer'])
 
@@ -132,6 +136,8 @@ class State:
     current_image = None
     current_image_sampling_step = 0
     textinfo = None
+    time_start = None
+    need_restart = False
 
     def skip(self):
         self.skipped = True
@@ -168,6 +174,7 @@ class State:
         self.skipped = False
         self.interrupted = False
         self.textinfo = None
+        self.time_start = time.time()
 
         devices.torch_gc()
 
@@ -177,6 +184,20 @@ class State:
 
         devices.torch_gc()
 
+    """sets self.current_image from self.current_latent if enough sampling steps have been made after the last call to this"""
+    def set_current_image(self):
+        if not parallel_processing_allowed:
+            return
+
+        if self.sampling_step - self.current_image_sampling_step >= opts.show_progress_every_n_steps and self.current_latent is not None:
+            if opts.show_progress_grid:
+                self.current_image = sd_samplers.samples_to_image_grid(self.current_latent)
+            else:
+                self.current_image = sd_samplers.sample_to_image(self.current_latent)
+
+            self.current_image_sampling_step = self.sampling_step
+
+
 state = State()
 
 artist_db = modules.artists.ArtistsDatabase(os.path.join(script_path, 'artists.csv'))
@@ -234,6 +255,8 @@ options_templates.update(options_section(('saving-images', "Saving images/grids"
     "enable_pnginfo": OptionInfo(True, "Save text information about generation parameters as chunks to png files"),
     "save_txt": OptionInfo(False, "Create a text file next to every image with generation parameters."),
     "save_images_before_face_restoration": OptionInfo(False, "Save a copy of image before doing face restoration."),
+    "save_images_before_highres_fix": OptionInfo(False, "Save a copy of image before applying highres fix."),
+    "save_images_before_color_correction": OptionInfo(False, "Save a copy of image before applying color correction to img2img results"),
     "jpeg_quality": OptionInfo(80, "Quality for saved jpeg images", gr.Slider, {"minimum": 1, "maximum": 100, "step": 1}),
     "export_for_4chan": OptionInfo(True, "If PNG image is larger than 4MB or any dimension is larger than 4000, downscale and save copy as JPG"),
 
@@ -290,16 +313,17 @@ options_templates.update(options_section(('training', "Training"), {
     "dataset_filename_join_string": OptionInfo(" ", "Filename join string"),
     "training_image_repeats_per_epoch": OptionInfo(1, "Number of repeats for a single input image per epoch; used only for displaying epoch number", gr.Number, {"precision": 0}),
     "training_write_csv_every": OptionInfo(500, "Save an csv containing the loss to log directory every N steps, 0 to disable"),
+    "training_xattention_optimizations": OptionInfo(False, "Use cross attention optimizations while training"),
 }))
 
 options_templates.update(options_section(('sd', "Stable Diffusion"), {
     "sd_model_checkpoint": OptionInfo(None, "Stable Diffusion checkpoint", gr.Dropdown, lambda: {"choices": modules.sd_models.checkpoint_tiles()}, refresh=sd_models.list_models),
     "sd_checkpoint_cache": OptionInfo(0, "Checkpoints to cache in RAM", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}),
+    "sd_vae": OptionInfo("auto", "SD VAE", gr.Dropdown, lambda: {"choices": list(sd_vae.vae_list)}, refresh=sd_vae.refresh_vae_list),
     "sd_hypernetwork": OptionInfo("None", "Hypernetwork", gr.Dropdown, lambda: {"choices": ["None"] + [x for x in hypernetworks.keys()]}, refresh=reload_hypernetworks),
     "sd_hypernetwork_strength": OptionInfo(1.0, "Hypernetwork strength", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.001}),
     "inpainting_mask_weight": OptionInfo(1.0, "Inpainting conditioning mask strength", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}),
     "img2img_color_correction": OptionInfo(False, "Apply color correction to img2img results to match original colors."),
-    "save_images_before_color_correction": OptionInfo(False, "Save a copy of image before applying color correction to img2img results"),
     "img2img_fix_steps": OptionInfo(False, "With img2img, do exactly the amount of steps the slider specifies (normally you'd do less with less denoising)."),
     "enable_quantization": OptionInfo(False, "Enable quantization in K samplers for sharper and cleaner results. This may change existing seeds. Requires restart to apply."),
     "enable_emphasis": OptionInfo(True, "Emphasis: use (text) to make model pay more attention to text and [text] to make it pay less attention"),
@@ -354,6 +378,12 @@ options_templates.update(options_section(('sampler-params', "Sampler parameters"
     'eta_noise_seed_delta': OptionInfo(0, "Eta noise seed delta", gr.Number, {"precision": 0}),
 }))
 
+options_templates.update(options_section((None, "Hidden options"), {
+    "disabled_extensions": OptionInfo([], "Disable those extensions"),
+}))
+
+options_templates.update()
+
 
 class Options:
     data = None
@@ -365,8 +395,9 @@ class Options:
 
     def __setattr__(self, key, value):
         if self.data is not None:
-            if key in self.data:
+            if key in self.data or key in self.data_labels:
                 self.data[key] = value
+                return
 
         return super(Options, self).__setattr__(key, value)
 
@@ -407,11 +438,12 @@ class Options:
         if bad_settings > 0:
             print(f"The program is likely to not work with bad settings.\nSettings file: {filename}\nEither fix the file, or delete it and restart.", file=sys.stderr)
 
-    def onchange(self, key, func):
+    def onchange(self, key, func, call=True):
         item = self.data_labels.get(key)
         item.onchange = func
 
-        func()
+        if call:
+            func()
 
     def dumpjson(self):
         d = {k: self.data.get(k, self.data_labels.get(k).default) for k in self.data_labels.keys()}

modules/textual_inversion/textual_inversion.py

@@ -235,6 +235,7 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc
     filename = os.path.join(shared.cmd_opts.embeddings_dir, f'{embedding_name}.pt')
 
     log_directory = os.path.join(log_directory, datetime.datetime.now().strftime("%Y-%m-%d"), embedding_name)
+    unload = shared.opts.unload_models_when_training
 
     if save_embedding_every > 0:
         embedding_dir = os.path.join(log_directory, "embeddings")
@@ -272,6 +273,8 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc
     shared.state.textinfo = f"Preparing dataset from {html.escape(data_root)}..."
     with torch.autocast("cuda"):
         ds = modules.textual_inversion.dataset.PersonalizedBase(data_root=data_root, width=training_width, height=training_height, repeats=shared.opts.training_image_repeats_per_epoch, placeholder_token=embedding_name, model=shared.sd_model, device=devices.device, template_file=template_file, batch_size=batch_size)
+    if unload:
+        shared.sd_model.first_stage_model.to(devices.cpu)
 
     embedding.vec.requires_grad = True
     optimizer = torch.optim.AdamW([embedding.vec], lr=scheduler.learn_rate)
@@ -328,6 +331,9 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc
         if images_dir is not None and steps_done % create_image_every == 0:
             forced_filename = f'{embedding_name}-{steps_done}'
             last_saved_image = os.path.join(images_dir, forced_filename)
+
+            shared.sd_model.first_stage_model.to(devices.device)
+
             p = processing.StableDiffusionProcessingTxt2Img(
                 sd_model=shared.sd_model,
                 do_not_save_grid=True,
@@ -355,6 +361,9 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc
             processed = processing.process_images(p)
             image = processed.images[0]
 
+            if unload:
+                shared.sd_model.first_stage_model.to(devices.cpu)
+
             shared.state.current_image = image
 
             if save_image_with_stored_embedding and os.path.exists(last_saved_file) and embedding_yet_to_be_embedded:
@@ -400,6 +409,7 @@ Last saved image: {html.escape(last_saved_image)}<br/>
 
     filename = os.path.join(shared.cmd_opts.embeddings_dir, f'{embedding_name}.pt')
     save_embedding(embedding, checkpoint, embedding_name, filename, remove_cached_checksum=True)
+    shared.sd_model.first_stage_model.to(devices.device)
 
     return embedding, filename
 

modules/textual_inversion/ui.py

@@ -25,8 +25,10 @@ def train_embedding(*args):
 
     assert not shared.cmd_opts.lowvram, 'Training models with lowvram not possible'
 
+    apply_optimizations = shared.opts.training_xattention_optimizations
     try:
-        sd_hijack.undo_optimizations()
+        if not apply_optimizations:
+            sd_hijack.undo_optimizations()
 
         embedding, filename = modules.textual_inversion.textual_inversion.train_embedding(*args)
 
@@ -38,5 +40,6 @@ Embedding saved to {html.escape(filename)}
     except Exception:
         raise
     finally:
-        sd_hijack.apply_optimizations()
+        if not apply_optimizations:
+            sd_hijack.apply_optimizations()
 

modules/txt2img.py

@@ -47,6 +47,8 @@ def txt2img(prompt: str, negative_prompt: str, prompt_style: str, prompt_style2:
     if processed is None:
         processed = process_images(p)
 
+    p.close()
+
     shared.total_tqdm.clear()
 
     generation_info_js = processed.js()

modules/ui.py

@@ -19,7 +19,7 @@ import numpy as np
 from PIL import Image, PngImagePlugin
 
 
-from modules import sd_hijack, sd_models, localization, script_callbacks
+from modules import sd_hijack, sd_models, localization, script_callbacks, ui_extensions
 from modules.paths import script_path
 
 from modules.shared import opts, cmd_opts, restricted_opts
@@ -277,15 +277,7 @@ def check_progress_call(id_part):
     preview_visibility = gr_show(False)
 
     if opts.show_progress_every_n_steps > 0:
-        if shared.parallel_processing_allowed:
-
-            if shared.state.sampling_step - shared.state.current_image_sampling_step >= opts.show_progress_every_n_steps and shared.state.current_latent is not None:
-                if opts.show_progress_grid:
-                    shared.state.current_image = modules.sd_samplers.samples_to_image_grid(shared.state.current_latent)
-                else:
-                    shared.state.current_image = modules.sd_samplers.sample_to_image(shared.state.current_latent)
-                shared.state.current_image_sampling_step = shared.state.sampling_step
-
+        shared.state.set_current_image()
         image = shared.state.current_image
 
         if image is None:
@@ -671,6 +663,9 @@ def create_ui(wrap_gradio_gpu_call):
     import modules.img2img
     import modules.txt2img
 
+    reload_javascript()
+
+    parameters_copypaste.reset()
 
     with gr.Blocks(analytics_enabled=False) as txt2img_interface:
         txt2img_prompt, roll, txt2img_prompt_style, txt2img_negative_prompt, txt2img_prompt_style2, submit, _, _, txt2img_prompt_style_apply, txt2img_save_style, txt2img_paste, token_counter, token_button = create_toprow(is_img2img=False)
@@ -1059,7 +1054,7 @@ def create_ui(wrap_gradio_gpu_call):
 
                 with gr.Tabs(elem_id="extras_resize_mode"):
                     with gr.TabItem('Scale by'):
-                        upscaling_resize = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Resize", value=2)
+                        upscaling_resize = gr.Slider(minimum=1.0, maximum=8.0, step=0.05, label="Resize", value=4)
                     with gr.TabItem('Scale to'):
                         with gr.Group():
                             with gr.Row():
@@ -1511,8 +1506,9 @@ def create_ui(wrap_gradio_gpu_call):
         column = None
         with gr.Row(elem_id="settings").style(equal_height=False):
             for i, (k, item) in enumerate(opts.data_labels.items()):
+                section_must_be_skipped = item.section[0] is None
 
-                if previous_section != item.section:
+                if previous_section != item.section and not section_must_be_skipped:
                     if cols_displayed < settings_cols and (items_displayed >= items_per_col or previous_section is None):
                         if column is not None:
                             column.__exit__()
@@ -1531,6 +1527,8 @@ def create_ui(wrap_gradio_gpu_call):
                 if k in quicksettings_names and not shared.cmd_opts.freeze_settings:
                     quicksettings_list.append((i, k, item))
                     components.append(dummy_component)
+                elif section_must_be_skipped:
+                    components.append(dummy_component)
                 else:
                     component = create_setting_component(k)
                     component_dict[k] = component
@@ -1566,19 +1564,19 @@ def create_ui(wrap_gradio_gpu_call):
         reload_script_bodies.click(
             fn=reload_scripts,
             inputs=[],
-            outputs=[],
-            _js='function(){}'
+            outputs=[]
         )
 
         def request_restart():
             shared.state.interrupt()
-            settings_interface.gradio_ref.do_restart = True
+            shared.state.need_restart = True
 
         restart_gradio.click(
+
             fn=request_restart,
             inputs=[],
             outputs=[],
-            _js='function(){restart_reload()}'
+            _js='restart_reload'
         )
 
         if column is not None:
@@ -1612,14 +1610,15 @@ def create_ui(wrap_gradio_gpu_call):
     interfaces += script_callbacks.ui_tabs_callback()
     interfaces += [(settings_interface, "Settings", "settings")]
 
+    extensions_interface = ui_extensions.create_ui()
+    interfaces += [(extensions_interface, "Extensions", "extensions")]
+
     with gr.Blocks(css=css, analytics_enabled=False, title="Stable Diffusion") as demo:
         with gr.Row(elem_id="quicksettings"):
             for i, k, item in quicksettings_list:
                 component = create_setting_component(k, is_quicksettings=True)
                 component_dict[k] = component
 
-        settings_interface.gradio_ref = demo
-
         parameters_copypaste.integrate_settings_paste_fields(component_dict)
         parameters_copypaste.run_bind()
 
@@ -1776,4 +1775,3 @@ def load_javascript(raw_response):
 
 
 reload_javascript = partial(load_javascript, gradio.routes.templates.TemplateResponse)
-reload_javascript()

modules/ui_extensions.py

@@ -0,0 +1,268 @@
+import json
+import os.path
+import shutil
+import sys
+import time
+import traceback
+
+import git
+
+import gradio as gr
+import html
+
+from modules import extensions, shared, paths
+
+
+available_extensions = {"extensions": []}
+
+
+def check_access():
+    assert not shared.cmd_opts.disable_extension_access, "extension access disabed because of commandline flags"
+
+
+def apply_and_restart(disable_list, update_list):
+    check_access()
+
+    disabled = json.loads(disable_list)
+    assert type(disabled) == list, f"wrong disable_list data for apply_and_restart: {disable_list}"
+
+    update = json.loads(update_list)
+    assert type(update) == list, f"wrong update_list data for apply_and_restart: {update_list}"
+
+    update = set(update)
+
+    for ext in extensions.extensions:
+        if ext.name not in update:
+            continue
+
+        try:
+            ext.pull()
+        except Exception:
+            print(f"Error pulling updates for {ext.name}:", file=sys.stderr)
+            print(traceback.format_exc(), file=sys.stderr)
+
+    shared.opts.disabled_extensions = disabled
+    shared.opts.save(shared.config_filename)
+
+    shared.state.interrupt()
+    shared.state.need_restart = True
+
+
+def check_updates():
+    check_access()
+
+    for ext in extensions.extensions:
+        if ext.remote is None:
+            continue
+
+        try:
+            ext.check_updates()
+        except Exception:
+            print(f"Error checking updates for {ext.name}:", file=sys.stderr)
+            print(traceback.format_exc(), file=sys.stderr)
+
+    return extension_table()
+
+
+def extension_table():
+    code = f"""<!-- {time.time()} -->
+    <table id="extensions">
+        <thead>
+            <tr>
+                <th><abbr title="Use checkbox to enable the extension; it will be enabled or disabled when you click apply button">Extension</abbr></th>
+                <th>URL</th>
+                <th><abbr title="Use checkbox to mark the extension for update; it will be updated when you click apply button">Update</abbr></th>
+            </tr>
+        </thead>
+        <tbody>
+    """
+
+    for ext in extensions.extensions:
+        if ext.can_update:
+            ext_status = f"""<label><input class="gr-check-radio gr-checkbox" name="update_{html.escape(ext.name)}" checked="checked" type="checkbox">{html.escape(ext.status)}</label>"""
+        else:
+            ext_status = ext.status
+
+        code += f"""
+            <tr>
+                <td><label><input class="gr-check-radio gr-checkbox" name="enable_{html.escape(ext.name)}" type="checkbox" {'checked="checked"' if ext.enabled else ''}>{html.escape(ext.name)}</label></td>
+                <td><a href="{html.escape(ext.remote or '')}">{html.escape(ext.remote or '')}</a></td>
+                <td{' class="extension_status"' if ext.remote is not None else ''}>{ext_status}</td>
+            </tr>
+    """
+
+    code += """
+        </tbody>
+    </table>
+    """
+
+    return code
+
+
+def normalize_git_url(url):
+    if url is None:
+        return ""
+
+    url = url.replace(".git", "")
+    return url
+
+
+def install_extension_from_url(dirname, url):
+    check_access()
+
+    assert url, 'No URL specified'
+
+    if dirname is None or dirname == "":
+        *parts, last_part = url.split('/')
+        last_part = normalize_git_url(last_part)
+
+        dirname = last_part
+
+    target_dir = os.path.join(extensions.extensions_dir, dirname)
+    assert not os.path.exists(target_dir), f'Extension directory already exists: {target_dir}'
+
+    normalized_url = normalize_git_url(url)
+    assert len([x for x in extensions.extensions if normalize_git_url(x.remote) == normalized_url]) == 0, 'Extension with this URL is already installed'
+
+    tmpdir = os.path.join(paths.script_path, "tmp", dirname)
+
+    try:
+        shutil.rmtree(tmpdir, True)
+
+        repo = git.Repo.clone_from(url, tmpdir)
+        repo.remote().fetch()
+
+        os.rename(tmpdir, target_dir)
+
+        extensions.list_extensions()
+        return [extension_table(), html.escape(f"Installed into {target_dir}. Use Installed tab to restart.")]
+    finally:
+        shutil.rmtree(tmpdir, True)
+
+
+def install_extension_from_index(url):
+    ext_table, message = install_extension_from_url(None, url)
+
+    return refresh_available_extensions_from_data(), ext_table, message
+
+
+def refresh_available_extensions(url):
+    global available_extensions
+
+    import urllib.request
+    with urllib.request.urlopen(url) as response:
+        text = response.read()
+
+    available_extensions = json.loads(text)
+
+    return url, refresh_available_extensions_from_data(), ''
+
+
+def refresh_available_extensions_from_data():
+    extlist = available_extensions["extensions"]
+    installed_extension_urls = {normalize_git_url(extension.remote): extension.name for extension in extensions.extensions}
+
+    code = f"""<!-- {time.time()} -->
+    <table id="available_extensions">
+        <thead>
+            <tr>
+                <th>Extension</th>
+                <th>Description</th>
+                <th>Action</th>
+            </tr>
+        </thead>
+        <tbody>
+    """
+
+    for ext in extlist:
+        name = ext.get("name", "noname")
+        url = ext.get("url", None)
+        description = ext.get("description", "")
+
+        if url is None:
+            continue
+
+        existing = installed_extension_urls.get(normalize_git_url(url), None)
+
+        install_code = f"""<input onclick="install_extension_from_index(this, '{html.escape(url)}')" type="button" value="{"Install" if not existing else "Installed"}" {"disabled=disabled" if existing else ""} class="gr-button gr-button-lg gr-button-secondary">"""
+
+        code += f"""
+            <tr>
+                <td><a href="{html.escape(url)}">{html.escape(name)}</a></td>
+                <td>{html.escape(description)}</td>
+                <td>{install_code}</td>
+            </tr>
+    """
+
+    code += """
+        </tbody>
+    </table>
+    """
+
+    return code
+
+
+def create_ui():
+    import modules.ui
+
+    with gr.Blocks(analytics_enabled=False) as ui:
+        with gr.Tabs(elem_id="tabs_extensions") as tabs:
+            with gr.TabItem("Installed"):
+
+                with gr.Row():
+                    apply = gr.Button(value="Apply and restart UI", variant="primary")
+                    check = gr.Button(value="Check for updates")
+                    extensions_disabled_list = gr.Text(elem_id="extensions_disabled_list", visible=False).style(container=False)
+                    extensions_update_list = gr.Text(elem_id="extensions_update_list", visible=False).style(container=False)
+
+                extensions_table = gr.HTML(lambda: extension_table())
+
+                apply.click(
+                    fn=apply_and_restart,
+                    _js="extensions_apply",
+                    inputs=[extensions_disabled_list, extensions_update_list],
+                    outputs=[],
+                )
+
+                check.click(
+                    fn=check_updates,
+                    _js="extensions_check",
+                    inputs=[],
+                    outputs=[extensions_table],
+                )
+
+            with gr.TabItem("Available"):
+                with gr.Row():
+                    refresh_available_extensions_button = gr.Button(value="Load from:", variant="primary")
+                    available_extensions_index = gr.Text(value="https://raw.githubusercontent.com/wiki/AUTOMATIC1111/stable-diffusion-webui/Extensions-index.md", label="Extension index URL").style(container=False)
+                    extension_to_install = gr.Text(elem_id="extension_to_install", visible=False)
+                    install_extension_button = gr.Button(elem_id="install_extension_button", visible=False)
+
+                install_result = gr.HTML()
+                available_extensions_table = gr.HTML()
+
+                refresh_available_extensions_button.click(
+                    fn=modules.ui.wrap_gradio_call(refresh_available_extensions, extra_outputs=[gr.update(), gr.update()]),
+                    inputs=[available_extensions_index],
+                    outputs=[available_extensions_index, available_extensions_table, install_result],
+                )
+
+                install_extension_button.click(
+                    fn=modules.ui.wrap_gradio_call(install_extension_from_index, extra_outputs=[gr.update(), gr.update()]),
+                    inputs=[extension_to_install],
+                    outputs=[available_extensions_table, extensions_table, install_result],
+                )
+
+            with gr.TabItem("Install from URL"):
+                install_url = gr.Text(label="URL for extension's git repository")
+                install_dirname = gr.Text(label="Local directory name", placeholder="Leave empty for auto")
+                install_button = gr.Button(value="Install", variant="primary")
+                install_result = gr.HTML(elem_id="extension_install_result")
+
+                install_button.click(
+                    fn=modules.ui.wrap_gradio_call(install_extension_from_url, extra_outputs=[gr.update()]),
+                    inputs=[install_dirname, install_url],
+                    outputs=[extensions_table, install_result],
+                )
+
+    return ui

modules/upscaler.py

@@ -10,6 +10,7 @@ import modules.shared
 from modules import modelloader, shared
 
 LANCZOS = (Image.Resampling.LANCZOS if hasattr(Image, 'Resampling') else Image.LANCZOS)
+NEAREST = (Image.Resampling.NEAREST if hasattr(Image, 'Resampling') else Image.NEAREST)
 from modules.paths import models_path
 
 
@@ -57,7 +58,7 @@ class Upscaler:
         dest_w = img.width * scale
         dest_h = img.height * scale
         for i in range(3):
-            if img.width >= dest_w and img.height >= dest_h:
+            if img.width > dest_w and img.height > dest_h:
                 break
             img = self.do_upscale(img, selected_model)
         if img.width != dest_w or img.height != dest_h:
@@ -120,3 +121,17 @@ class UpscalerLanczos(Upscaler):
         self.name = "Lanczos"
         self.scalers = [UpscalerData("Lanczos", None, self)]
 
+
+class UpscalerNearest(Upscaler):
+    scalers = []
+
+    def do_upscale(self, img, selected_model=None):
+        return img.resize((int(img.width * self.scale), int(img.height * self.scale)), resample=NEAREST)
+
+    def load_model(self, _):
+        pass
+
+    def __init__(self, dirname=None):
+        super().__init__(False)
+        self.name = "Nearest"
+        self.scalers = [UpscalerData("Nearest", None, self)]

requirements.txt

@@ -4,7 +4,7 @@ fairscale==0.4.4
 fonts
 font-roboto
 gfpgan
-gradio==3.5
+gradio==3.8
 invisible-watermark
 numpy
 omegaconf
@@ -12,7 +12,7 @@ opencv-python
 requests
 piexif
 Pillow
-pytorch_lightning
+pytorch_lightning==1.7.7
 realesrgan
 scikit-image>=0.19
 timm==0.4.12
@@ -26,3 +26,4 @@ torchdiffeq
 kornia
 lark
 inflection
+GitPython

requirements_versions.txt

@@ -2,7 +2,7 @@ transformers==4.19.2
 diffusers==0.3.0
 basicsr==1.4.2
 gfpgan==1.3.8
-gradio==3.5
+gradio==3.8
 numpy==1.23.3
 Pillow==9.2.0
 realesrgan==0.3.0
@@ -23,3 +23,4 @@ torchdiffeq==0.2.3
 kornia==0.6.7
 lark==1.1.2
 inflection==0.5.1
+GitPython==3.1.27

scripts/img2imgalt.py

@@ -166,8 +166,7 @@ class Script(scripts.Script):
         if override_strength:
             p.denoising_strength = 1.0
 
-
-        def sample_extra(conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength):
+        def sample_extra(conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts):
             lat = (p.init_latent.cpu().numpy() * 10).astype(int)
 
             same_params = self.cache is not None and self.cache.cfg_scale == cfg and self.cache.steps == st \

scripts/prompts_from_file.py

@@ -96,6 +96,7 @@ class Script(scripts.Script):
 
     def ui(self, is_img2img):
         checkbox_iterate = gr.Checkbox(label="Iterate seed every line", value=False)
+        checkbox_iterate_batch = gr.Checkbox(label="Use same random seed for all lines", value=False)
 
         prompt_txt = gr.Textbox(label="List of prompt inputs", lines=1)
         file = gr.File(label="Upload prompt inputs", type='bytes')
@@ -106,9 +107,9 @@ class Script(scripts.Script):
         # We don't shrink back to 1, because that causes the control to ignore [enter], and it may
         # be unclear to the user that shift-enter is needed.
         prompt_txt.change(lambda tb: gr.update(lines=7) if ("\n" in tb) else gr.update(lines=2), inputs=[prompt_txt], outputs=[prompt_txt])
-        return [checkbox_iterate, file, prompt_txt]
+        return [checkbox_iterate, checkbox_iterate_batch, file, prompt_txt]
 
-    def run(self, p, checkbox_iterate, file, prompt_txt: str):
+    def run(self, p, checkbox_iterate, checkbox_iterate_batch, file, prompt_txt: str):
         lines = [x.strip() for x in prompt_txt.splitlines()]
         lines = [x for x in lines if len(x) > 0]
 
@@ -137,7 +138,7 @@ class Script(scripts.Script):
             jobs.append(args)
 
         print(f"Will process {len(lines)} lines in {job_count} jobs.")
-        if (checkbox_iterate and p.seed == -1):
+        if (checkbox_iterate or checkbox_iterate_batch) and p.seed == -1:
             p.seed = int(random.randrange(4294967294))
 
         state.job_count = job_count
@@ -153,7 +154,7 @@ class Script(scripts.Script):
             proc = process_images(copy_p)
             images += proc.images
             
-            if (checkbox_iterate):
+            if checkbox_iterate:
                 p.seed = p.seed + (p.batch_size * p.n_iter)
 
 

style.css

@@ -260,6 +260,16 @@ input[type="range"]{
 #txt2img_negative_prompt, #img2img_negative_prompt{
 }
 
+/* gradio 3.8 adds opacity to progressbar which makes it blink; disable it here */
+.transition.opacity-20 {
+  opacity: 1 !important;
+}
+
+/* more gradio's garbage cleanup */
+.min-h-\[4rem\] {
+  min-height: unset !important;
+}
+
 #txt2img_progressbar, #img2img_progressbar, #ti_progressbar{
   position: absolute;
   z-index: 1000;
@@ -491,7 +501,7 @@ input[type="range"]{
     padding: 0;
 }
 
-#refresh_sd_model_checkpoint, #refresh_sd_hypernetwork, #refresh_train_hypernetwork_name, #refresh_train_embedding_name, #refresh_localization{
+#refresh_sd_model_checkpoint, #refresh_sd_vae, #refresh_sd_hypernetwork, #refresh_train_hypernetwork_name, #refresh_train_embedding_name, #refresh_localization{
     max-width: 2.5em;
     min-width: 2.5em;
     height: 2.4em;
@@ -530,6 +540,29 @@ img2maskimg, #img2maskimg > .h-60, #img2maskimg > .h-60 > div, #img2maskimg > .h
     min-height: 480px !important;
 }
 
+/* Extensions */
+
+#tab_extensions table{
+    border-collapse: collapse;
+}
+
+#tab_extensions table td, #tab_extensions table th{
+    border: 1px solid #ccc;
+    padding: 0.25em 0.5em;
+}
+
+#tab_extensions table input[type="checkbox"]{
+    margin-right: 0.5em;
+}
+
+#tab_extensions button{
+    max-width: 16em;
+}
+
+#tab_extensions input[disabled="disabled"]{
+    opacity: 0.5;
+}
+
 /* The following handles localization for right-to-left (RTL) languages like Arabic.
 The rtl media type will only be activated by the logic in javascript/localization.js.
 If you change anything above, you need to make sure it is RTL compliant by just running
@@ -607,4 +640,4 @@ Then, you will need to add the RTL counterpart only if needed in the rtl section
         right: unset;
         left: 0.5em;
     }
-}
+}

webui.py

@@ -9,7 +9,7 @@ from fastapi.middleware.gzip import GZipMiddleware
 
 from modules.paths import script_path
 
-from modules import devices, sd_samplers, upscaler
+from modules import devices, sd_samplers, upscaler, extensions
 import modules.codeformer_model as codeformer
 import modules.extras
 import modules.face_restoration
@@ -21,8 +21,10 @@ import modules.paths
 import modules.scripts
 import modules.sd_hijack
 import modules.sd_models
+import modules.sd_vae
 import modules.shared as shared
 import modules.txt2img
+import modules.script_callbacks
 
 import modules.ui
 from modules import devices
@@ -60,6 +62,8 @@ def wrap_gradio_gpu_call(func, extra_outputs=None):
 
 
 def initialize():
+    extensions.list_extensions()
+
     if cmd_opts.ui_debug_mode:
         shared.sd_upscalers = upscaler.UpscalerLanczos().scalers
         modules.scripts.load_scripts()
@@ -74,8 +78,10 @@ def initialize():
 
     modules.scripts.load_scripts()
 
+    modules.sd_vae.refresh_vae_list()
     modules.sd_models.load_model()
-    shared.opts.onchange("sd_model_checkpoint", wrap_queued_call(lambda: modules.sd_models.reload_model_weights(shared.sd_model)))
+    shared.opts.onchange("sd_model_checkpoint", wrap_queued_call(lambda: modules.sd_models.reload_model_weights()))
+    shared.opts.onchange("sd_vae", wrap_queued_call(lambda: modules.sd_vae.reload_vae_weights()), call=False)
     shared.opts.onchange("sd_hypernetwork", wrap_queued_call(lambda: modules.hypernetworks.hypernetwork.load_hypernetwork(shared.opts.sd_hypernetwork)))
     shared.opts.onchange("sd_hypernetwork_strength", modules.hypernetworks.hypernetwork.apply_strength)
 
@@ -92,15 +98,18 @@ def create_api(app):
     api = Api(app, queue_lock)
     return api
 
+
 def wait_on_server(demo=None):
     while 1:
         time.sleep(0.5)
-        if demo and getattr(demo, 'do_restart', False):
+        if shared.state.need_restart:
+            shared.state.need_restart = False
             time.sleep(0.5)
             demo.close()
             time.sleep(0.5)
             break
 
+
 def api_only():
     initialize()
 
@@ -108,6 +117,8 @@ def api_only():
     app.add_middleware(GZipMiddleware, minimum_size=1000)
     api = create_api(app)
 
+    modules.script_callbacks.app_started_callback(None, app)
+
     api.launch(server_name="0.0.0.0" if cmd_opts.listen else "127.0.0.1", port=cmd_opts.port if cmd_opts.port else 7861)
 
 
@@ -132,14 +143,18 @@ def webui():
 
         app.add_middleware(GZipMiddleware, minimum_size=1000)
 
-        if (launch_api):
+        if launch_api:
             create_api(app)
 
+        modules.script_callbacks.app_started_callback(demo, app)
+
         wait_on_server(demo)
 
         sd_samplers.set_samplers()
 
-        print('Reloading Custom Scripts')
+        print('Reloading extensions')
+        extensions.list_extensions()
+        print('Reloading custom scripts')
         modules.scripts.reload_scripts()
         print('Reloading modules: modules.ui')
         importlib.reload(modules.ui)
@@ -148,8 +163,6 @@ def webui():
         print('Restarting Gradio')
 
 
-
-task = []
 if __name__ == "__main__":
     if cmd_opts.nowebui:
         api_only()