Merge branch 'master' into save-images

comfy/ldm/modules/diffusionmodules/openaimodel.py

@@ -477,9 +477,9 @@ class UNetModel(nn.Module):
 
         if context_dim is not None:
             assert use_spatial_transformer, 'Fool!! You forgot to use the spatial transformer for your cross-attention conditioning...'
-            from omegaconf.listconfig import ListConfig
-            if type(context_dim) == ListConfig:
-                context_dim = list(context_dim)
+            # from omegaconf.listconfig import ListConfig
+            # if type(context_dim) == ListConfig:
+            #     context_dim = list(context_dim)
 
         if num_heads_upsample == -1:
             num_heads_upsample = num_heads

comfy/model_management.py

@@ -31,8 +31,25 @@ try:
 except:
     pass
 
-if "--cpu" in sys.argv:
-    vram_state = CPU
+if "--disable-xformers" in sys.argv:
+    XFORMERS_IS_AVAILBLE = False
+else:
+    try:
+        import xformers
+        import xformers.ops
+        XFORMERS_IS_AVAILBLE = True
+    except:
+        XFORMERS_IS_AVAILBLE = False
+
+ENABLE_PYTORCH_ATTENTION = False
+if "--use-pytorch-cross-attention" in sys.argv:
+    torch.backends.cuda.enable_math_sdp(True)
+    torch.backends.cuda.enable_flash_sdp(True)
+    torch.backends.cuda.enable_mem_efficient_sdp(True)
+    ENABLE_PYTORCH_ATTENTION = True
+    XFORMERS_IS_AVAILBLE = False
+
+
 if "--lowvram" in sys.argv:
     set_vram_to = LOW_VRAM
 if "--novram" in sys.argv:
@@ -54,6 +71,8 @@ if set_vram_to == LOW_VRAM or set_vram_to == NO_VRAM:
     total_vram_available_mb = (total_vram - 1024) // 2
     total_vram_available_mb = int(max(256, total_vram_available_mb))
 
+if "--cpu" in sys.argv:
+    vram_state = CPU
 
 print("Set vram state to:", ["CPU", "NO VRAM", "LOW VRAM", "NORMAL VRAM", "HIGH VRAM"][vram_state])
 
@@ -159,6 +178,14 @@ def get_autocast_device(dev):
         return dev.type
     return "cuda"
 
+def xformers_enabled():
+    if vram_state == CPU:
+        return False
+    return XFORMERS_IS_AVAILBLE
+
+def pytorch_attention_enabled():
+    return ENABLE_PYTORCH_ATTENTION
+
 def get_free_memory(dev=None, torch_free_too=False):
     if dev is None:
         dev = get_torch_device()

comfy/sd.py

@@ -6,7 +6,7 @@ import sd2_clip
 import model_management
 from .ldm.util import instantiate_from_config
 from .ldm.models.autoencoder import AutoencoderKL
-from omegaconf import OmegaConf
+import yaml
 from .cldm import cldm
 from .t2i_adapter import adapter
 
@@ -726,12 +726,19 @@ def load_clip(ckpt_path, embedding_directory=None):
     return clip
 
 def load_checkpoint(config_path, ckpt_path, output_vae=True, output_clip=True, embedding_directory=None):
-    config = OmegaConf.load(config_path)
+    with open(config_path, 'r') as stream:
+        config = yaml.safe_load(stream)
     model_config_params = config['model']['params']
     clip_config = model_config_params['cond_stage_config']
     scale_factor = model_config_params['scale_factor']
     vae_config = model_config_params['first_stage_config']
 
+    fp16 = False
+    if "unet_config" in model_config_params:
+        if "params" in model_config_params["unet_config"]:
+            if "use_fp16" in model_config_params["unet_config"]["params"]:
+                fp16 = model_config_params["unet_config"]["params"]["use_fp16"]
+
     clip = None
     vae = None
 
@@ -750,9 +757,13 @@ def load_checkpoint(config_path, ckpt_path, output_vae=True, output_clip=True, e
         w.cond_stage_model = clip.cond_stage_model
         load_state_dict_to = [w]
 
-    model = instantiate_from_config(config.model)
+    model = instantiate_from_config(config["model"])
     sd = load_torch_file(ckpt_path)
     model = load_model_weights(model, sd, verbose=False, load_state_dict_to=load_state_dict_to)
+
+    if fp16:
+        model = model.half()
+
     return (ModelPatcher(model), clip, vae)
 
 
@@ -853,4 +864,7 @@ def load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, e
     model = instantiate_from_config(model_config)
     model = load_model_weights(model, sd, verbose=False, load_state_dict_to=load_state_dict_to)
 
+    if fp16:
+        model = model.half()
+
     return (ModelPatcher(model), clip, vae)

main.py

 import os
 import sys
+import shutil
 
 import threading
 import asyncio
@@ -8,9 +9,6 @@ if os.name == "nt":
     import logging
     logging.getLogger("xformers").addFilter(lambda record: 'A matching Triton is not available' not in record.getMessage())
 
-import execution
-import server
-
 if __name__ == "__main__":
     if '--help' in sys.argv:
         print("Valid Command line Arguments:")
@@ -18,6 +16,8 @@ if __name__ == "__main__":
         print("\t--port 8188\t\t\tSet the listen port.")
         print("\t--dont-upcast-attention\t\tDisable upcasting of attention \n\t\t\t\t\tcan boost speed but increase the chances of black images.\n")
         print("\t--use-split-cross-attention\tUse the split cross attention optimization instead of the sub-quadratic one.\n\t\t\t\t\tIgnored when xformers is used.")
+        print("\t--use-pytorch-cross-attention\tUse the new pytorch 2.0 cross attention function.")
+        print("\t--disable-xformers\t\tdisables xformers")
         print()
         print("\t--highvram\t\t\tBy default models will be unloaded to CPU memory after being used.\n\t\t\t\t\tThis option keeps them in GPU memory.\n")
         print("\t--normalvram\t\t\tUsed to force normal vram use if lowvram gets automatically enabled.")
@@ -31,6 +31,9 @@ if __name__ == "__main__":
         print("disabling upcasting of attention")
         os.environ['ATTN_PRECISION'] = "fp16"
 
+import execution
+import server
+
 def prompt_worker(q, server):
     e = execution.PromptExecutor(server)
     while True:
@@ -38,8 +41,8 @@ def prompt_worker(q, server):
         e.execute(item[-2], item[-1])
         q.task_done(item_id, e.outputs)
 
-async def run(server, address='', port=8188, verbose=True):
-    await asyncio.gather(server.start(address, port, verbose), server.publish_loop())
+async def run(server, address='', port=8188, verbose=True, call_on_start=None):
+    await asyncio.gather(server.start(address, port, verbose, call_on_start), server.publish_loop())
 
 def hijack_progress(server):
     from tqdm.auto import tqdm
@@ -51,7 +54,14 @@ def hijack_progress(server):
         return v
     setattr(tqdm, "update", wrapped_func)
 
+def cleanup_temp():
+    temp_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "temp")
+    if os.path.exists(temp_dir):
+        shutil.rmtree(temp_dir, ignore_errors=True)
+
 if __name__ == "__main__":
+    cleanup_temp()
+
     loop = asyncio.new_event_loop()
     asyncio.set_event_loop(loop)
     server = server.PromptServer(loop)
@@ -76,11 +86,22 @@ if __name__ == "__main__":
     except:
         pass
 
+    if '--quick-test-for-ci' in sys.argv:
+        exit(0)
+
+    call_on_start = None
+    if "--windows-standalone-build" in sys.argv:
+        def startup_server(address, port):
+            import webbrowser
+            webbrowser.open("http://{}:{}".format(address, port))
+        call_on_start = startup_server
+
     if os.name == "nt":
         try:
-            loop.run_until_complete(run(server, address=address, port=port, verbose=not dont_print))
+            loop.run_until_complete(run(server, address=address, port=port, verbose=not dont_print, call_on_start=call_on_start))
         except KeyboardInterrupt:
             pass
     else:
-        loop.run_until_complete(run(server, address=address, port=port, verbose=not dont_print))
+        loop.run_until_complete(run(server, address=address, port=port, verbose=not dont_print, call_on_start=call_on_start))
 
+    cleanup_temp()

nodes.py

@@ -775,6 +775,7 @@ class KSamplerAdvanced:
 class SaveImage:
     def __init__(self):
         self.output_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "output")
+        self.url_suffix = ""
 
     @classmethod
     def INPUT_TYPES(s):
@@ -818,6 +819,9 @@ class SaveImage:
             os.makedirs(full_output_folder, exist_ok=True)
             counter = 1
 
+        if not os.path.exists(self.output_dir):
+            os.makedirs(self.output_dir)
+
         paths = list()
         for image in images:
             i = 255. * image.cpu().numpy()
@@ -828,12 +832,25 @@ class SaveImage:
             if extra_pnginfo is not None:
                 for x in extra_pnginfo:
                     metadata.add_text(x, json.dumps(extra_pnginfo[x]))
+
             file = f"{filename}_{counter:05}_.png"
             img.save(os.path.join(full_output_folder, file), pnginfo=metadata, optimize=True)
-            paths.append(os.path.join(subfolder, file))
+            paths.append(os.path.join(subfolder, file + self.url_suffix))
             counter += 1
         return { "ui": { "images": paths } }
 
+class PreviewImage(SaveImage):
+    def __init__(self):
+        self.output_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "temp")
+        self.url_suffix = "?type=temp"
+
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required":
+                    {"images": ("IMAGE", ), },
+                "hidden": {"prompt": "PROMPT", "extra_pnginfo": "EXTRA_PNGINFO"},
+                }
+
 class LoadImage:
     input_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "input")
     @classmethod
@@ -954,6 +971,7 @@ NODE_CLASS_MAPPINGS = {
     "EmptyLatentImage": EmptyLatentImage,
     "LatentUpscale": LatentUpscale,
     "SaveImage": SaveImage,
+    "PreviewImage": PreviewImage,
     "LoadImage": LoadImage,
     "LoadImageMask": LoadImageMask,
     "ImageScale": ImageScale,

requirements.txt

 torch
 torchdiffeq
 torchsde
-omegaconf
 einops
 open-clip-torch
 transformers
@@ -9,3 +8,4 @@ safetensors
 pytorch_lightning
 aiohttp
 accelerate
+pyyaml

server.py

@@ -113,7 +113,7 @@ class PromptServer():
         async def view_image(request):
             if "file" in request.rel_url.query:
                 type = request.rel_url.query.get("type", "output")
-                if type != "output" and type != "input":
+                if type not in ["output", "input", "temp"]:
                     return web.Response(status=400)
 
                 output_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), type)
@@ -267,7 +267,7 @@ class PromptServer():
             msg = await self.messages.get()
             await self.send(*msg)
 
-    async def start(self, address, port, verbose=True):
+    async def start(self, address, port, verbose=True, call_on_start=None):
         runner = web.AppRunner(self.app)
         await runner.setup()
         site = web.TCPSite(runner, address, port)
@@ -278,3 +278,6 @@ class PromptServer():
         if verbose:
             print("Starting server\n")
             print("To see the GUI go to: http://{}:{}".format(address, port))
+        if call_on_start is not None:
+            call_on_start(address, port)
+

web/scripts/app.js

@@ -144,7 +144,14 @@ class ComfyApp {
 					if (numImages === 1 && !imageIndex) {
 						this.imageIndex = imageIndex = 0;
 					}
-					let shiftY = this.type === "SaveImage" ? 55 : this.imageOffset || 0;
+
+					let shiftY;
+					if (this.imageOffset != null) {
+						shiftY = this.imageOffset;
+					} else {
+						shiftY = this.computeSize()[1];
+					}
+
 					let dw = this.size[0];
 					let dh = this.size[1];
 					dh -= shiftY;
@@ -400,6 +407,15 @@ class ComfyApp {
 		api.init();
 	}
 
+	#addKeyboardHandler() {
+		window.addEventListener("keydown", (e) => {
+			// Queue prompt using ctrl or command + enter
+			if ((e.ctrlKey || e.metaKey) && (e.key === "Enter" || e.keyCode === 13 || e.keyCode === 10)) {
+				this.queuePrompt(e.shiftKey ? -1 : 0);
+			}
+		});
+	}
+
 	/**
 	 * Loads all extensions from the API into the window
 	 */
@@ -466,6 +482,7 @@ class ComfyApp {
 		this.#addApiUpdateHandlers();
 		this.#addDropHandler();
 		this.#addPasteHandler();
+		this.#addKeyboardHandler();
 
 		await this.#invokeExtensionsAsync("setup");
 	}
@@ -499,7 +516,11 @@ class ComfyApp {
 
 						if (Array.isArray(type)) {
 							// Enums e.g. latent rotation
-							this.addWidget("combo", inputName, type[0], () => {}, { values: type });
+							let defaultValue = type[0];
+							if (inputData[1] && inputData[1].default) {
+								defaultValue = inputData[1].default;
+							}
+							this.addWidget("combo", inputName, defaultValue, () => {}, { values: type });
 						} else if (`${type}:${inputName}` in widgets) {
 							// Support custom widgets by Type:Name
 							Object.assign(config, widgets[`${type}:${inputName}`](this, inputName, inputData, app) || {});