Merge branch 'master' of github.com:comfyanonymous/ComfyUI into feat/linux-amd-setup

README.md

@@ -65,12 +65,13 @@ See what ComfyUI can do with the [example workflows](https://comfyanonymous.gith
    - [Flux](https://comfyanonymous.github.io/ComfyUI_examples/flux/)
    - [Lumina Image 2.0](https://comfyanonymous.github.io/ComfyUI_examples/lumina2/)
    - [HiDream](https://comfyanonymous.github.io/ComfyUI_examples/hidream/)
+   - [Cosmos Predict2](https://comfyanonymous.github.io/ComfyUI_examples/cosmos_predict2/)
 - Video Models
    - [Stable Video Diffusion](https://comfyanonymous.github.io/ComfyUI_examples/video/)
    - [Mochi](https://comfyanonymous.github.io/ComfyUI_examples/mochi/)
    - [LTX-Video](https://comfyanonymous.github.io/ComfyUI_examples/ltxv/)
    - [Hunyuan Video](https://comfyanonymous.github.io/ComfyUI_examples/hunyuan_video/)
-   - [Nvidia Cosmos](https://comfyanonymous.github.io/ComfyUI_examples/cosmos/)
+   - [Nvidia Cosmos](https://comfyanonymous.github.io/ComfyUI_examples/cosmos/) and [Cosmos Predict2](https://comfyanonymous.github.io/ComfyUI_examples/cosmos_predict2/)
    - [Wan 2.1](https://comfyanonymous.github.io/ComfyUI_examples/wan/)
 - Audio Models
    - [Stable Audio](https://comfyanonymous.github.io/ComfyUI_examples/audio/)
@@ -272,6 +273,8 @@ You can install ComfyUI in Apple Mac silicon (M1 or M2) with any recent macOS ve
 
 #### DirectML (AMD Cards on Windows)
 
+This is very badly supported and is not recommended. There are some unofficial builds of pytorch ROCm on windows that exist that will give you a much better experience than this. This readme will be updated once official pytorch ROCm builds for windows come out.
+
 ```pip install torch-directml``` Then you can launch ComfyUI with: ```python main.py --directml```
 
 #### Ascend NPUs

comfy/k_diffusion/sampling.py

@@ -781,6 +781,7 @@ def sample_dpmpp_2m(model, x, sigmas, extra_args=None, callback=None, disable=No
         old_denoised = denoised
     return x
 
+
 @torch.no_grad()
 def sample_dpmpp_2m_sde(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None, solver_type='midpoint'):
     """DPM-Solver++(2M) SDE."""
@@ -796,9 +797,12 @@ def sample_dpmpp_2m_sde(model, x, sigmas, extra_args=None, callback=None, disabl
     noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=seed, cpu=True) if noise_sampler is None else noise_sampler
     s_in = x.new_ones([x.shape[0]])
 
+    model_sampling = model.inner_model.model_patcher.get_model_object('model_sampling')
+    lambda_fn = partial(sigma_to_half_log_snr, model_sampling=model_sampling)
+    sigmas = offset_first_sigma_for_snr(sigmas, model_sampling)
+
     old_denoised = None
-    h_last = None
-    h = None
+    h, h_last = None, None
 
     for i in trange(len(sigmas) - 1, disable=disable):
         denoised = model(x, sigmas[i] * s_in, **extra_args)
@@ -809,26 +813,29 @@ def sample_dpmpp_2m_sde(model, x, sigmas, extra_args=None, callback=None, disabl
             x = denoised
         else:
             # DPM-Solver++(2M) SDE
-            t, s = -sigmas[i].log(), -sigmas[i + 1].log()
-            h = s - t
-            eta_h = eta * h
+            lambda_s, lambda_t = lambda_fn(sigmas[i]), lambda_fn(sigmas[i + 1])
+            h = lambda_t - lambda_s
+            h_eta = h * (eta + 1)
 
-            x = sigmas[i + 1] / sigmas[i] * (-eta_h).exp() * x + (-h - eta_h).expm1().neg() * denoised
+            alpha_t = sigmas[i + 1] * lambda_t.exp()
+
+            x = sigmas[i + 1] / sigmas[i] * (-h * eta).exp() * x + alpha_t * (-h_eta).expm1().neg() * denoised
 
             if old_denoised is not None:
                 r = h_last / h
                 if solver_type == 'heun':
-                    x = x + ((-h - eta_h).expm1().neg() / (-h - eta_h) + 1) * (1 / r) * (denoised - old_denoised)
+                    x = x + alpha_t * ((-h_eta).expm1().neg() / (-h_eta) + 1) * (1 / r) * (denoised - old_denoised)
                 elif solver_type == 'midpoint':
-                    x = x + 0.5 * (-h - eta_h).expm1().neg() * (1 / r) * (denoised - old_denoised)
+                    x = x + 0.5 * alpha_t * (-h_eta).expm1().neg() * (1 / r) * (denoised - old_denoised)
 
-            if eta:
-                x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * sigmas[i + 1] * (-2 * eta_h).expm1().neg().sqrt() * s_noise
+            if eta > 0 and s_noise > 0:
+                x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * sigmas[i + 1] * (-2 * h * eta).expm1().neg().sqrt() * s_noise
 
         old_denoised = denoised
         h_last = h
     return x
 
+
 @torch.no_grad()
 def sample_dpmpp_3m_sde(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None):
     """DPM-Solver++(3M) SDE."""
@@ -842,6 +849,10 @@ def sample_dpmpp_3m_sde(model, x, sigmas, extra_args=None, callback=None, disabl
     noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=seed, cpu=True) if noise_sampler is None else noise_sampler
     s_in = x.new_ones([x.shape[0]])
 
+    model_sampling = model.inner_model.model_patcher.get_model_object('model_sampling')
+    lambda_fn = partial(sigma_to_half_log_snr, model_sampling=model_sampling)
+    sigmas = offset_first_sigma_for_snr(sigmas, model_sampling)
+
     denoised_1, denoised_2 = None, None
     h, h_1, h_2 = None, None, None
 
@@ -853,13 +864,16 @@ def sample_dpmpp_3m_sde(model, x, sigmas, extra_args=None, callback=None, disabl
             # Denoising step
             x = denoised
         else:
-            t, s = -sigmas[i].log(), -sigmas[i + 1].log()
-            h = s - t
+            lambda_s, lambda_t = lambda_fn(sigmas[i]), lambda_fn(sigmas[i + 1])
+            h = lambda_t - lambda_s
             h_eta = h * (eta + 1)
 
-            x = torch.exp(-h_eta) * x + (-h_eta).expm1().neg() * denoised
+            alpha_t = sigmas[i + 1] * lambda_t.exp()
+
+            x = sigmas[i + 1] / sigmas[i] * (-h * eta).exp() * x + alpha_t * (-h_eta).expm1().neg() * denoised
 
             if h_2 is not None:
+                # DPM-Solver++(3M) SDE
                 r0 = h_1 / h
                 r1 = h_2 / h
                 d1_0 = (denoised - denoised_1) / r0
@@ -868,20 +882,22 @@ def sample_dpmpp_3m_sde(model, x, sigmas, extra_args=None, callback=None, disabl
                 d2 = (d1_0 - d1_1) / (r0 + r1)
                 phi_2 = h_eta.neg().expm1() / h_eta + 1
                 phi_3 = phi_2 / h_eta - 0.5
-                x = x + phi_2 * d1 - phi_3 * d2
+                x = x + (alpha_t * phi_2) * d1 - (alpha_t * phi_3) * d2
             elif h_1 is not None:
+                # DPM-Solver++(2M) SDE
                 r = h_1 / h
                 d = (denoised - denoised_1) / r
                 phi_2 = h_eta.neg().expm1() / h_eta + 1
-                x = x + phi_2 * d
+                x = x + (alpha_t * phi_2) * d
 
-            if eta:
+            if eta > 0 and s_noise > 0:
                 x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * sigmas[i + 1] * (-2 * h * eta).expm1().neg().sqrt() * s_noise
 
         denoised_1, denoised_2 = denoised, denoised_1
         h_1, h_2 = h, h_1
     return x
 
+
 @torch.no_grad()
 def sample_dpmpp_3m_sde_gpu(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None):
     if len(sigmas) <= 1:
@@ -891,6 +907,7 @@ def sample_dpmpp_3m_sde_gpu(model, x, sigmas, extra_args=None, callback=None, di
     noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=extra_args.get("seed", None), cpu=False) if noise_sampler is None else noise_sampler
     return sample_dpmpp_3m_sde(model, x, sigmas, extra_args=extra_args, callback=callback, disable=disable, eta=eta, s_noise=s_noise, noise_sampler=noise_sampler)
 
+
 @torch.no_grad()
 def sample_dpmpp_2m_sde_gpu(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None, solver_type='midpoint'):
     if len(sigmas) <= 1:
@@ -900,6 +917,7 @@ def sample_dpmpp_2m_sde_gpu(model, x, sigmas, extra_args=None, callback=None, di
     noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max, seed=extra_args.get("seed", None), cpu=False) if noise_sampler is None else noise_sampler
     return sample_dpmpp_2m_sde(model, x, sigmas, extra_args=extra_args, callback=callback, disable=disable, eta=eta, s_noise=s_noise, noise_sampler=noise_sampler, solver_type=solver_type)
 
+
 @torch.no_grad()
 def sample_dpmpp_sde_gpu(model, x, sigmas, extra_args=None, callback=None, disable=None, eta=1., s_noise=1., noise_sampler=None, r=1 / 2):
     if len(sigmas) <= 1:

comfy/ldm/flux/controlnet.py

@@ -123,6 +123,8 @@ class ControlNetFlux(Flux):
 
         if y is None:
             y = torch.zeros((img.shape[0], self.params.vec_in_dim), device=img.device, dtype=img.dtype)
+        else:
+            y = y[:, :self.params.vec_in_dim]
 
         # running on sequences img
         img = self.img_in(img)

comfy/ldm/flux/layers.py

@@ -118,7 +118,7 @@ class Modulation(nn.Module):
 def apply_mod(tensor, m_mult, m_add=None, modulation_dims=None):
     if modulation_dims is None:
         if m_add is not None:
-            return tensor * m_mult + m_add
+            return torch.addcmul(m_add, tensor, m_mult)
         else:
             return tensor * m_mult
     else:

comfy/ldm/modules/sub_quadratic_attention.py

@@ -31,7 +31,7 @@ def dynamic_slice(
     starts: List[int],
     sizes: List[int],
 ) -> Tensor:
-    slicing = [slice(start, start + size) for start, size in zip(starts, sizes)]
+    slicing = tuple(slice(start, start + size) for start, size in zip(starts, sizes))
     return x[slicing]
 
 class AttnChunk(NamedTuple):

comfy/model_base.py

@@ -1024,6 +1024,8 @@ class CosmosPredict2(BaseModel):
     def process_timestep(self, timestep, x, denoise_mask=None, **kwargs):
         if denoise_mask is None:
             return timestep
+        if denoise_mask.ndim <= 4:
+            return timestep
         condition_video_mask_B_1_T_1_1 = denoise_mask.mean(dim=[1, 3, 4], keepdim=True)
         c_noise_B_1_T_1_1 = 0.0 * (1.0 - condition_video_mask_B_1_T_1_1) + timestep.reshape(timestep.shape[0], 1, 1, 1, 1) * condition_video_mask_B_1_T_1_1
         out = c_noise_B_1_T_1_1.squeeze(dim=[1, 3, 4])

comfy/sd1_clip.py

@@ -462,7 +462,7 @@ class SDTokenizer:
             tokenizer_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "sd1_tokenizer")
         self.tokenizer = tokenizer_class.from_pretrained(tokenizer_path, **tokenizer_args)
         self.max_length = tokenizer_data.get("{}_max_length".format(embedding_key), max_length)
-        self.min_length = min_length
+        self.min_length = tokenizer_data.get("{}_min_length".format(embedding_key), min_length)
         self.end_token = None
         self.min_padding = min_padding
 

comfy_config/config_parser.py

@@ -11,6 +11,43 @@ from comfy_config.types import (
     PyProjectSettings
 )
 
+def validate_and_extract_os_classifiers(classifiers: list) -> list:
+    os_classifiers = [c for c in classifiers if c.startswith("Operating System :: ")]
+    if not os_classifiers:
+        return []
+
+    os_values = [c[len("Operating System :: ") :] for c in os_classifiers]
+    valid_os_prefixes = {"Microsoft", "POSIX", "MacOS", "OS Independent"}
+
+    for os_value in os_values:
+        if not any(os_value.startswith(prefix) for prefix in valid_os_prefixes):
+            return []
+
+    return os_values
+
+
+def validate_and_extract_accelerator_classifiers(classifiers: list) -> list:
+    accelerator_classifiers = [c for c in classifiers if c.startswith("Environment ::")]
+    if not accelerator_classifiers:
+        return []
+
+    accelerator_values = [c[len("Environment :: ") :] for c in accelerator_classifiers]
+
+    valid_accelerators = {
+        "GPU :: NVIDIA CUDA",
+        "GPU :: AMD ROCm",
+        "GPU :: Intel Arc",
+        "NPU :: Huawei Ascend",
+        "GPU :: Apple Metal",
+    }
+
+    for accelerator_value in accelerator_values:
+        if accelerator_value not in valid_accelerators:
+            return []
+
+    return accelerator_values
+
+
 """
 Extract configuration from a custom node directory's pyproject.toml file or a Python file.
 
@@ -78,6 +115,24 @@ def extract_node_configuration(path) -> Optional[PyProjectConfig]:
     tool_data = raw_settings.tool
     comfy_data = tool_data.get("comfy", {}) if tool_data else {}
 
+    dependencies = project_data.get("dependencies", [])
+    supported_comfyui_frontend_version = ""
+    for dep in dependencies:
+        if isinstance(dep, str) and dep.startswith("comfyui-frontend-package"):
+            supported_comfyui_frontend_version = dep.removeprefix("comfyui-frontend-package")
+            break
+
+    supported_comfyui_version = comfy_data.get("requires-comfyui", "")
+
+    classifiers = project_data.get('classifiers', [])
+    supported_os = validate_and_extract_os_classifiers(classifiers)
+    supported_accelerators = validate_and_extract_accelerator_classifiers(classifiers)
+
+    project_data['supported_os'] = supported_os
+    project_data['supported_accelerators'] = supported_accelerators
+    project_data['supported_comfyui_frontend_version'] = supported_comfyui_frontend_version
+    project_data['supported_comfyui_version'] = supported_comfyui_version
+
     return PyProjectConfig(project=project_data, tool_comfy=comfy_data)
 
 

comfy_config/types.py

@@ -51,7 +51,7 @@ class ComfyConfig(BaseModel):
     models: List[Model] = Field(default_factory=list, alias="Models")
     includes: List[str] = Field(default_factory=list)
     web: Optional[str] = None
-
+    banner_url: str = ""
 
 class License(BaseModel):
     file: str = ""
@@ -66,6 +66,10 @@ class ProjectConfig(BaseModel):
     dependencies: List[str] = Field(default_factory=list)
     license: License = Field(default_factory=License)
     urls: URLs = Field(default_factory=URLs)
+    supported_os: List[str] = Field(default_factory=list)
+    supported_accelerators: List[str] = Field(default_factory=list)
+    supported_comfyui_version: str = ""
+    supported_comfyui_frontend_version: str = ""
 
     @field_validator('license', mode='before')
     @classmethod

comfy_extras/nodes_images.py

@@ -304,10 +304,23 @@ Optional spacing can be added between images.
                 image2.movedim(-1, 1), target_w, target_h, "lanczos", "disabled"
             ).movedim(1, -1)
 
+        color_map = {
+            "white": 1.0,
+            "black": 0.0,
+            "red": (1.0, 0.0, 0.0),
+            "green": (0.0, 1.0, 0.0),
+            "blue": (0.0, 0.0, 1.0),
+        }
+
+        color_val = color_map[spacing_color]
+
         # When not matching sizes, pad to align non-concat dimensions
         if not match_image_size:
             h1, w1 = image1.shape[1:3]
             h2, w2 = image2.shape[1:3]
+            pad_value = 0.0
+            if not isinstance(color_val, tuple):
+                pad_value = color_val
 
             if direction in ["left", "right"]:
                 # For horizontal concat, pad heights to match
@@ -316,11 +329,11 @@ Optional spacing can be added between images.
                     if h1 < target_h:
                         pad_h = target_h - h1
                         pad_top, pad_bottom = pad_h // 2, pad_h - pad_h // 2
-                        image1 = torch.nn.functional.pad(image1, (0, 0, 0, 0, pad_top, pad_bottom), mode='constant', value=0.0)
+                        image1 = torch.nn.functional.pad(image1, (0, 0, 0, 0, pad_top, pad_bottom), mode='constant', value=pad_value)
                     if h2 < target_h:
                         pad_h = target_h - h2
                         pad_top, pad_bottom = pad_h // 2, pad_h - pad_h // 2
-                        image2 = torch.nn.functional.pad(image2, (0, 0, 0, 0, pad_top, pad_bottom), mode='constant', value=0.0)
+                        image2 = torch.nn.functional.pad(image2, (0, 0, 0, 0, pad_top, pad_bottom), mode='constant', value=pad_value)
             else:  # up, down
                 # For vertical concat, pad widths to match
                 if w1 != w2:
@@ -328,11 +341,11 @@ Optional spacing can be added between images.
                     if w1 < target_w:
                         pad_w = target_w - w1
                         pad_left, pad_right = pad_w // 2, pad_w - pad_w // 2
-                        image1 = torch.nn.functional.pad(image1, (0, 0, pad_left, pad_right), mode='constant', value=0.0)
+                        image1 = torch.nn.functional.pad(image1, (0, 0, pad_left, pad_right), mode='constant', value=pad_value)
                     if w2 < target_w:
                         pad_w = target_w - w2
                         pad_left, pad_right = pad_w // 2, pad_w - pad_w // 2
-                        image2 = torch.nn.functional.pad(image2, (0, 0, pad_left, pad_right), mode='constant', value=0.0)
+                        image2 = torch.nn.functional.pad(image2, (0, 0, pad_left, pad_right), mode='constant', value=pad_value)
 
         # Ensure same number of channels
         if image1.shape[-1] != image2.shape[-1]:
@@ -366,15 +379,6 @@ Optional spacing can be added between images.
         if spacing_width > 0:
             spacing_width = spacing_width + (spacing_width % 2)  # Ensure even
 
-            color_map = {
-                "white": 1.0,
-                "black": 0.0,
-                "red": (1.0, 0.0, 0.0),
-                "green": (0.0, 1.0, 0.0),
-                "blue": (0.0, 0.0, 1.0),
-            }
-            color_val = color_map[spacing_color]
-
             if direction in ["left", "right"]:
                 spacing_shape = (
                     image1.shape[0],
@@ -410,6 +414,62 @@ Optional spacing can be added between images.
         concat_dim = 2 if direction in ["left", "right"] else 1
         return (torch.cat(images, dim=concat_dim),)
 
+class ResizeAndPadImage:
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "image": ("IMAGE",),
+                "target_width": ("INT", {
+                    "default": 512,
+                    "min": 1,
+                    "max": MAX_RESOLUTION,
+                    "step": 1
+                }),
+                "target_height": ("INT", {
+                    "default": 512,
+                    "min": 1,
+                    "max": MAX_RESOLUTION,
+                    "step": 1
+                }),
+                "padding_color": (["white", "black"],),
+                "interpolation": (["area", "bicubic", "nearest-exact", "bilinear", "lanczos"],),
+            }
+        }
+
+    RETURN_TYPES = ("IMAGE",)
+    FUNCTION = "resize_and_pad"
+    CATEGORY = "image/transform"
+
+    def resize_and_pad(self, image, target_width, target_height, padding_color, interpolation):
+        batch_size, orig_height, orig_width, channels = image.shape
+
+        scale_w = target_width / orig_width
+        scale_h = target_height / orig_height
+        scale = min(scale_w, scale_h)
+
+        new_width = int(orig_width * scale)
+        new_height = int(orig_height * scale)
+
+        image_permuted = image.permute(0, 3, 1, 2)
+
+        resized = comfy.utils.common_upscale(image_permuted, new_width, new_height, interpolation, "disabled")
+
+        pad_value = 0.0 if padding_color == "black" else 1.0
+        padded = torch.full(
+            (batch_size, channels, target_height, target_width),
+            pad_value,
+            dtype=image.dtype,
+            device=image.device
+        )
+
+        y_offset = (target_height - new_height) // 2
+        x_offset = (target_width - new_width) // 2
+
+        padded[:, :, y_offset:y_offset + new_height, x_offset:x_offset + new_width] = resized
+
+        output = padded.permute(0, 2, 3, 1)
+        return (output,)
 
 class SaveSVGNode:
     """
@@ -532,5 +592,6 @@ NODE_CLASS_MAPPINGS = {
     "SaveAnimatedPNG": SaveAnimatedPNG,
     "SaveSVGNode": SaveSVGNode,
     "ImageStitch": ImageStitch,
+    "ResizeAndPadImage": ResizeAndPadImage,
     "GetImageSize": GetImageSize,
 }

comfyui_version.py

@@ -1,3 +1,3 @@
 # This file is automatically generated by the build process when version is
 # updated in pyproject.toml.
-__version__ = "0.3.40"
+__version__ = "0.3.41"

execution.py

@@ -429,17 +429,20 @@ def execute(server, dynprompt, caches, current_item, extra_data, executed, promp
 
         logging.error(f"!!! Exception during processing !!! {ex}")
         logging.error(traceback.format_exc())
+        tips = ""
+
+        if isinstance(ex, comfy.model_management.OOM_EXCEPTION):
+            tips = "This error means you ran out of memory on your GPU.\n\nTIPS: If the workflow worked before you might have accidentally set the batch_size to a large number."
+            logging.error("Got an OOM, unloading all loaded models.")
+            comfy.model_management.unload_all_models()
 
         error_details = {
             "node_id": real_node_id,
-            "exception_message": str(ex),
+            "exception_message": "{}\n{}".format(ex, tips),
             "exception_type": exception_type,
             "traceback": traceback.format_tb(tb),
             "current_inputs": input_data_formatted
         }
-        if isinstance(ex, comfy.model_management.OOM_EXCEPTION):
-            logging.error("Got an OOM, unloading all loaded models.")
-            comfy.model_management.unload_all_models()
 
         return (ExecutionResult.FAILURE, error_details, ex)
 

main.py

@@ -185,7 +185,13 @@ def prompt_worker(q, server_instance):
 
             current_time = time.perf_counter()
             execution_time = current_time - execution_start_time
-            logging.info("Prompt executed in {:.2f} seconds".format(execution_time))
+
+            # Log Time in a more readable way after 10 minutes
+            if execution_time > 600:
+                execution_time = time.strftime("%H:%M:%S", time.gmtime(execution_time))
+                logging.info(f"Prompt executed in {execution_time}")
+            else:
+                logging.info("Prompt executed in {:.2f} seconds".format(execution_time))
 
         flags = q.get_flags()
         free_memory = flags.get("free_memory", False)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ComfyUI"
-version = "0.3.40"
+version = "0.3.41"
 readme = "README.md"
 license = { file = "LICENSE" }
 requires-python = ">=3.9"

requirements.txt

@@ -1,5 +1,5 @@
-comfyui-frontend-package==1.21.7
-comfyui-workflow-templates==0.1.28
+comfyui-frontend-package==1.22.2
+comfyui-workflow-templates==0.1.29
 comfyui-embedded-docs==0.2.2
 torch
 torchsde