ComfyUI v0.19.5

Signed-off-by: bigcat88 <bigcat88@icloud.com>

comfy/model_base.py

@@ -578,8 +578,8 @@ class Stable_Zero123(BaseModel):
     def __init__(self, model_config, model_type=ModelType.EPS, device=None, cc_projection_weight=None, cc_projection_bias=None):
         super().__init__(model_config, model_type, device=device)
         self.cc_projection = comfy.ops.manual_cast.Linear(cc_projection_weight.shape[1], cc_projection_weight.shape[0], dtype=self.get_dtype(), device=device)
-        self.cc_projection.weight.copy_(cc_projection_weight)
-        self.cc_projection.bias.copy_(cc_projection_bias)
+        self.cc_projection.weight = torch.nn.Parameter(cc_projection_weight.clone())
+        self.cc_projection.bias = torch.nn.Parameter(cc_projection_bias.clone())
 
     def extra_conds(self, **kwargs):
         out = {}

comfy_api_nodes/apis/bytedance.py

@@ -122,6 +122,41 @@ class TaskStatusResponse(BaseModel):
     usage: TaskStatusUsage | None = Field(None)
 
 
+class GetAssetResponse(BaseModel):
+    id: str = Field(...)
+    name: str | None = Field(None)
+    url: str | None = Field(None)
+    asset_type: str = Field(...)
+    group_id: str = Field(...)
+    status: str = Field(...)
+    error: TaskStatusError | None = Field(None)
+
+
+class SeedanceCreateVisualValidateSessionResponse(BaseModel):
+    session_id: str = Field(...)
+    h5_link: str = Field(...)
+
+
+class SeedanceGetVisualValidateSessionResponse(BaseModel):
+    session_id: str = Field(...)
+    status: str = Field(...)
+    group_id: str | None = Field(None)
+    error_code: str | None = Field(None)
+    error_message: str | None = Field(None)
+
+
+class SeedanceCreateAssetRequest(BaseModel):
+    group_id: str = Field(...)
+    url: str = Field(...)
+    asset_type: str = Field(...)
+    name: str | None = Field(None, max_length=64)
+    project_name: str | None = Field(None)
+
+
+class SeedanceCreateAssetResponse(BaseModel):
+    asset_id: str = Field(...)
+
+
 # Dollars per 1K tokens, keyed by (model_id, has_video_input).
 SEEDANCE2_PRICE_PER_1K_TOKENS = {
     ("dreamina-seedance-2-0-260128", False): 0.007,
@@ -158,10 +193,17 @@ RECOMMENDED_PRESETS_SEEDREAM_4 = [
     ("Custom", None, None),
 ]
 
-# Seedance 2.0 reference video pixel count limits per model.
+# Seedance 2.0 reference video pixel count limits per model and output resolution.
 SEEDANCE2_REF_VIDEO_PIXEL_LIMITS = {
-    "dreamina-seedance-2-0-260128": {"min": 409_600, "max": 927_408},
-    "dreamina-seedance-2-0-fast-260128": {"min": 409_600, "max": 927_408},
+    "dreamina-seedance-2-0-260128": {
+        "480p": {"min": 409_600, "max": 927_408},
+        "720p": {"min": 409_600, "max": 927_408},
+        "1080p": {"min": 409_600, "max": 2_073_600},
+    },
+    "dreamina-seedance-2-0-fast-260128": {
+        "480p": {"min": 409_600, "max": 927_408},
+        "720p": {"min": 409_600, "max": 927_408},
+    },
 }
 
 # The time in this dictionary are given for 10 seconds duration.

comfy_api_nodes/nodes_bytedance.py

@@ -1,5 +1,6 @@
 import logging
 import math
+import re
 
 import torch
 from typing_extensions import override
@@ -11,9 +12,14 @@ from comfy_api_nodes.apis.bytedance import (
     SEEDANCE2_PRICE_PER_1K_TOKENS,
     SEEDANCE2_REF_VIDEO_PIXEL_LIMITS,
     VIDEO_TASKS_EXECUTION_TIME,
+    GetAssetResponse,
     Image2VideoTaskCreationRequest,
     ImageTaskCreationResponse,
     Seedance2TaskCreationRequest,
+    SeedanceCreateAssetRequest,
+    SeedanceCreateAssetResponse,
+    SeedanceCreateVisualValidateSessionResponse,
+    SeedanceGetVisualValidateSessionResponse,
     Seedream4Options,
     Seedream4TaskCreationRequest,
     TaskAudioContent,
@@ -35,6 +41,7 @@ from comfy_api_nodes.util import (
     get_number_of_images,
     image_tensor_pair_to_batch,
     poll_op,
+    resize_video_to_pixel_budget,
     sync_op,
     upload_audio_to_comfyapi,
     upload_image_to_comfyapi,
@@ -43,10 +50,16 @@ from comfy_api_nodes.util import (
     validate_image_aspect_ratio,
     validate_image_dimensions,
     validate_string,
+    validate_video_dimensions,
+    validate_video_duration,
 )
+from server import PromptServer
 
 BYTEPLUS_IMAGE_ENDPOINT = "/proxy/byteplus/api/v3/images/generations"
 
+_VERIFICATION_POLL_TIMEOUT_SEC = 120
+_VERIFICATION_POLL_INTERVAL_SEC = 3
+
 SEEDREAM_MODELS = {
     "seedream 5.0 lite": "seedream-5-0-260128",
     "seedream-4-5-251128": "seedream-4-5-251128",
@@ -69,9 +82,12 @@ DEPRECATED_MODELS = {"seedance-1-0-lite-t2v-250428", "seedance-1-0-lite-i2v-2504
 logger = logging.getLogger(__name__)
 
 
-def _validate_ref_video_pixels(video: Input.Video, model_id: str, index: int) -> None:
-    """Validate reference video pixel count against Seedance 2.0 model limits."""
-    limits = SEEDANCE2_REF_VIDEO_PIXEL_LIMITS.get(model_id)
+def _validate_ref_video_pixels(video: Input.Video, model_id: str, resolution: str, index: int) -> None:
+    """Validate reference video pixel count against Seedance 2.0 model limits for the selected resolution."""
+    model_limits = SEEDANCE2_REF_VIDEO_PIXEL_LIMITS.get(model_id)
+    if not model_limits:
+        return
+    limits = model_limits.get(resolution)
     if not limits:
         return
     try:
@@ -92,6 +108,169 @@ def _validate_ref_video_pixels(video: Input.Video, model_id: str, index: int) ->
         )
 
 
+async def _resolve_reference_assets(
+    cls: type[IO.ComfyNode],
+    asset_ids: list[str],
+) -> tuple[dict[str, str], dict[str, str], dict[str, str]]:
+    """Look up each asset, validate Active status, group by asset_type.
+
+    Returns (image_assets, video_assets, audio_assets), each mapping asset_id -> "asset://<asset_id>".
+    """
+    image_assets: dict[str, str] = {}
+    video_assets: dict[str, str] = {}
+    audio_assets: dict[str, str] = {}
+    for i, raw_id in enumerate(asset_ids, 1):
+        asset_id = (raw_id or "").strip()
+        if not asset_id:
+            continue
+        result = await sync_op(
+            cls,
+            ApiEndpoint(path=f"/proxy/seedance/assets/{asset_id}"),
+            response_model=GetAssetResponse,
+        )
+        if result.status != "Active":
+            extra = f" {result.error.code}: {result.error.message}" if result.error else ""
+            raise ValueError(f"Reference asset {i} (Id={asset_id}) is not Active (Status={result.status}).{extra}")
+        asset_uri = f"asset://{asset_id}"
+        if result.asset_type == "Image":
+            image_assets[asset_id] = asset_uri
+        elif result.asset_type == "Video":
+            video_assets[asset_id] = asset_uri
+        elif result.asset_type == "Audio":
+            audio_assets[asset_id] = asset_uri
+    return image_assets, video_assets, audio_assets
+
+
+_ASSET_REF_RE = re.compile(r"\basset ?(\d{1,2})\b", re.IGNORECASE)
+
+
+def _build_asset_labels(
+    reference_assets: dict[str, str],
+    image_asset_uris: dict[str, str],
+    video_asset_uris: dict[str, str],
+    audio_asset_uris: dict[str, str],
+    n_reference_images: int,
+    n_reference_videos: int,
+    n_reference_audios: int,
+) -> dict[int, str]:
+    """Map asset slot number (from 'asset_N' keys) to its positional label.
+
+    Asset entries are appended to `content` after the reference_images/videos/audios,
+    so their 1-indexed labels continue from the count of existing same-type refs:
+    one reference_images entry + one Image-type asset -> asset labelled "Image 2".
+    """
+    image_n = n_reference_images
+    video_n = n_reference_videos
+    audio_n = n_reference_audios
+    labels: dict[int, str] = {}
+    for slot_key, raw_id in reference_assets.items():
+        asset_id = (raw_id or "").strip()
+        if not asset_id:
+            continue
+        try:
+            slot_num = int(slot_key.rsplit("_", 1)[-1])
+        except ValueError:
+            continue
+        if asset_id in image_asset_uris:
+            image_n += 1
+            labels[slot_num] = f"Image {image_n}"
+        elif asset_id in video_asset_uris:
+            video_n += 1
+            labels[slot_num] = f"Video {video_n}"
+        elif asset_id in audio_asset_uris:
+            audio_n += 1
+            labels[slot_num] = f"Audio {audio_n}"
+    return labels
+
+
+def _rewrite_asset_refs(prompt: str, labels: dict[int, str]) -> str:
+    """Case-insensitively replace 'assetNN' (1-2 digit) tokens with their labels."""
+    if not labels:
+        return prompt
+
+    def _sub(m: "re.Match[str]") -> str:
+        return labels.get(int(m.group(1)), m.group(0))
+
+    return _ASSET_REF_RE.sub(_sub, prompt)
+
+
+async def _obtain_group_id_via_h5_auth(cls: type[IO.ComfyNode]) -> str:
+    session = await sync_op(
+        cls,
+        ApiEndpoint(path="/proxy/seedance/visual-validate/sessions", method="POST"),
+        response_model=SeedanceCreateVisualValidateSessionResponse,
+    )
+    logger.warning("Seedance authentication required. Open link: %s", session.h5_link)
+
+    h5_text = f"Open this link in your browser and complete face verification:\n\n{session.h5_link}"
+
+    result = await poll_op(
+        cls,
+        ApiEndpoint(path=f"/proxy/seedance/visual-validate/sessions/{session.session_id}"),
+        response_model=SeedanceGetVisualValidateSessionResponse,
+        status_extractor=lambda r: r.status,
+        completed_statuses=["completed"],
+        failed_statuses=["failed"],
+        poll_interval=_VERIFICATION_POLL_INTERVAL_SEC,
+        max_poll_attempts=(_VERIFICATION_POLL_TIMEOUT_SEC // _VERIFICATION_POLL_INTERVAL_SEC) - 1,
+        estimated_duration=_VERIFICATION_POLL_TIMEOUT_SEC - 1,
+        extra_text=h5_text,
+    )
+
+    if not result.group_id:
+        raise RuntimeError(f"Seedance session {session.session_id} completed without a group_id")
+
+    logger.warning("Seedance authentication complete. New GroupId: %s", result.group_id)
+    PromptServer.instance.send_progress_text(
+        f"Authentication complete. New GroupId: {result.group_id}", cls.hidden.unique_id
+    )
+    return result.group_id
+
+
+async def _resolve_group_id(cls: type[IO.ComfyNode], group_id: str) -> str:
+    if group_id and group_id.strip():
+        return group_id.strip()
+    return await _obtain_group_id_via_h5_auth(cls)
+
+
+async def _create_seedance_asset(
+    cls: type[IO.ComfyNode],
+    *,
+    group_id: str,
+    url: str,
+    name: str,
+    asset_type: str,
+) -> str:
+    req = SeedanceCreateAssetRequest(
+        group_id=group_id,
+        url=url,
+        asset_type=asset_type,
+        name=name or None,
+    )
+    result = await sync_op(
+        cls,
+        ApiEndpoint(path="/proxy/seedance/assets", method="POST"),
+        response_model=SeedanceCreateAssetResponse,
+        data=req,
+    )
+    return result.asset_id
+
+
+async def _wait_for_asset_active(cls: type[IO.ComfyNode], asset_id: str, group_id: str) -> GetAssetResponse:
+    """Poll the newly created asset until its status becomes Active."""
+    return await poll_op(
+        cls,
+        ApiEndpoint(path=f"/proxy/seedance/assets/{asset_id}"),
+        response_model=GetAssetResponse,
+        status_extractor=lambda r: r.status,
+        completed_statuses=["Active"],
+        failed_statuses=["Failed"],
+        poll_interval=5,
+        max_poll_attempts=1200,
+        extra_text=f"Waiting for asset pre-processing...\n\nasset_id: {asset_id}\n\ngroup_id: {group_id}",
+    )
+
+
 def _seedance2_price_extractor(model_id: str, has_video_input: bool):
     """Returns a price_extractor closure for Seedance 2.0 poll_op."""
     rate = SEEDANCE2_PRICE_PER_1K_TOKENS.get((model_id, has_video_input))
@@ -1224,12 +1403,27 @@ class ByteDance2FirstLastFrameNode(IO.ComfyNode):
                 IO.Image.Input(
                     "first_frame",
                     tooltip="First frame image for the video.",
+                    optional=True,
                 ),
                 IO.Image.Input(
                     "last_frame",
                     tooltip="Last frame image for the video.",
                     optional=True,
                 ),
+                IO.String.Input(
+                    "first_frame_asset_id",
+                    default="",
+                    tooltip="Seedance asset_id to use as the first frame. "
+                            "Mutually exclusive with the first_frame image input.",
+                    optional=True,
+                ),
+                IO.String.Input(
+                    "last_frame_asset_id",
+                    default="",
+                    tooltip="Seedance asset_id to use as the last frame. "
+                            "Mutually exclusive with the last_frame image input.",
+                    optional=True,
+                ),
                 IO.Int.Input(
                     "seed",
                     default=0,
@@ -1282,24 +1476,54 @@ class ByteDance2FirstLastFrameNode(IO.ComfyNode):
     async def execute(
         cls,
         model: dict,
-        first_frame: Input.Image,
         seed: int,
         watermark: bool,
+        first_frame: Input.Image | None = None,
         last_frame: Input.Image | None = None,
+        first_frame_asset_id: str = "",
+        last_frame_asset_id: str = "",
     ) -> IO.NodeOutput:
         validate_string(model["prompt"], strip_whitespace=True, min_length=1)
         model_id = SEEDANCE_MODELS[model["model"]]
 
+        first_frame_asset_id = first_frame_asset_id.strip()
+        last_frame_asset_id = last_frame_asset_id.strip()
+
+        if first_frame is not None and first_frame_asset_id:
+            raise ValueError("Provide only one of first_frame or first_frame_asset_id, not both.")
+        if first_frame is None and not first_frame_asset_id:
+            raise ValueError("Either first_frame or first_frame_asset_id is required.")
+        if last_frame is not None and last_frame_asset_id:
+            raise ValueError("Provide only one of last_frame or last_frame_asset_id, not both.")
+
+        asset_ids_to_resolve = [a for a in (first_frame_asset_id, last_frame_asset_id) if a]
+        image_assets: dict[str, str] = {}
+        if asset_ids_to_resolve:
+            image_assets, _, _ = await _resolve_reference_assets(cls, asset_ids_to_resolve)
+            for aid in asset_ids_to_resolve:
+                if aid not in image_assets:
+                    raise ValueError(f"Asset {aid} is not an Image asset.")
+
+        if first_frame_asset_id:
+            first_frame_url = image_assets[first_frame_asset_id]
+        else:
+            first_frame_url = await upload_image_to_comfyapi(cls, first_frame, wait_label="Uploading first frame.")
+
         content: list[TaskTextContent | TaskImageContent] = [
             TaskTextContent(text=model["prompt"]),
             TaskImageContent(
-                image_url=TaskImageContentUrl(
-                    url=await upload_image_to_comfyapi(cls, first_frame, wait_label="Uploading first frame.")
-                ),
+                image_url=TaskImageContentUrl(url=first_frame_url),
                 role="first_frame",
             ),
         ]
-        if last_frame is not None:
+        if last_frame_asset_id:
+            content.append(
+                TaskImageContent(
+                    image_url=TaskImageContentUrl(url=image_assets[last_frame_asset_id]),
+                    role="last_frame",
+                ),
+            )
+        elif last_frame is not None:
             content.append(
                 TaskImageContent(
                     image_url=TaskImageContentUrl(
@@ -1373,6 +1597,32 @@ def _seedance2_reference_inputs(resolutions: list[str]):
                 min=0,
             ),
         ),
+        IO.Boolean.Input(
+            "auto_downscale",
+            default=False,
+            advanced=True,
+            optional=True,
+            tooltip="Automatically downscale reference videos that exceed the model's pixel budget "
+            "for the selected resolution. Aspect ratio is preserved; videos already within limits are untouched.",
+        ),
+        IO.Autogrow.Input(
+            "reference_assets",
+            template=IO.Autogrow.TemplateNames(
+                IO.String.Input("reference_asset"),
+                names=[
+                    "asset_1",
+                    "asset_2",
+                    "asset_3",
+                    "asset_4",
+                    "asset_5",
+                    "asset_6",
+                    "asset_7",
+                    "asset_8",
+                    "asset_9",
+                ],
+                min=0,
+            ),
+        ),
     ]
 
 
@@ -1474,16 +1724,47 @@ class ByteDance2ReferenceNode(IO.ComfyNode):
         reference_images = model.get("reference_images", {})
         reference_videos = model.get("reference_videos", {})
         reference_audios = model.get("reference_audios", {})
+        reference_assets = model.get("reference_assets", {})
 
-        if not reference_images and not reference_videos:
-            raise ValueError("At least one reference image or video is required.")
+        reference_image_assets, reference_video_assets, reference_audio_assets = await _resolve_reference_assets(
+            cls, list(reference_assets.values())
+        )
+
+        if not reference_images and not reference_videos and not reference_image_assets and not reference_video_assets:
+            raise ValueError("At least one reference image or video or asset is required.")
+
+        total_images = len(reference_images) + len(reference_image_assets)
+        if total_images > 9:
+            raise ValueError(
+                f"Too many reference images: {total_images} "
+                f"(images={len(reference_images)}, image assets={len(reference_image_assets)}). Maximum is 9."
+            )
+        total_videos = len(reference_videos) + len(reference_video_assets)
+        if total_videos > 3:
+            raise ValueError(
+                f"Too many reference videos: {total_videos} "
+                f"(videos={len(reference_videos)}, video assets={len(reference_video_assets)}). Maximum is 3."
+            )
+        total_audios = len(reference_audios) + len(reference_audio_assets)
+        if total_audios > 3:
+            raise ValueError(
+                f"Too many reference audios: {total_audios} "
+                f"(audios={len(reference_audios)}, audio assets={len(reference_audio_assets)}). Maximum is 3."
+            )
 
         model_id = SEEDANCE_MODELS[model["model"]]
-        has_video_input = len(reference_videos) > 0
+        has_video_input = total_videos > 0
+
+        if model.get("auto_downscale") and reference_videos:
+            max_px = SEEDANCE2_REF_VIDEO_PIXEL_LIMITS.get(model_id, {}).get(model["resolution"], {}).get("max")
+            if max_px:
+                for key in reference_videos:
+                    reference_videos[key] = resize_video_to_pixel_budget(reference_videos[key], max_px)
+
         total_video_duration = 0.0
         for i, key in enumerate(reference_videos, 1):
             video = reference_videos[key]
-            _validate_ref_video_pixels(video, model_id, i)
+            _validate_ref_video_pixels(video, model_id, model["resolution"], i)
             try:
                 dur = video.get_duration()
                 if dur < 1.8:
@@ -1506,8 +1787,19 @@ class ByteDance2ReferenceNode(IO.ComfyNode):
         if total_audio_duration > 15.1:
             raise ValueError(f"Total reference audio duration is {total_audio_duration:.1f}s. Maximum is 15.1 seconds.")
 
+        asset_labels = _build_asset_labels(
+            reference_assets,
+            reference_image_assets,
+            reference_video_assets,
+            reference_audio_assets,
+            len(reference_images),
+            len(reference_videos),
+            len(reference_audios),
+        )
+        prompt_text = _rewrite_asset_refs(model["prompt"], asset_labels)
+
         content: list[TaskTextContent | TaskImageContent | TaskVideoContent | TaskAudioContent] = [
-            TaskTextContent(text=model["prompt"]),
+            TaskTextContent(text=prompt_text),
         ]
         for i, key in enumerate(reference_images, 1):
             content.append(
@@ -1548,6 +1840,21 @@ class ByteDance2ReferenceNode(IO.ComfyNode):
                     ),
                 ),
             )
+        for url in reference_image_assets.values():
+            content.append(
+                TaskImageContent(
+                    image_url=TaskImageContentUrl(url=url),
+                    role="reference_image",
+                ),
+            )
+        for url in reference_video_assets.values():
+            content.append(
+                TaskVideoContent(video_url=TaskVideoContentUrl(url=url)),
+            )
+        for url in reference_audio_assets.values():
+            content.append(
+                TaskAudioContent(audio_url=TaskAudioContentUrl(url=url)),
+            )
         initial_response = await sync_op(
             cls,
             ApiEndpoint(path=BYTEPLUS_TASK_ENDPOINT, method="POST"),
@@ -1602,6 +1909,156 @@ async def process_video_task(
     return IO.NodeOutput(await download_url_to_video_output(response.content.video_url))
 
 
+class ByteDanceCreateImageAsset(IO.ComfyNode):
+
+    @classmethod
+    def define_schema(cls) -> IO.Schema:
+        return IO.Schema(
+            node_id="ByteDanceCreateImageAsset",
+            display_name="ByteDance Create Image Asset",
+            category="api node/image/ByteDance",
+            description=(
+                "Create a Seedance 2.0 personal image asset. Uploads the input image and "
+                "registers it in the given asset group. If group_id is empty, runs a real-person "
+                "H5 authentication flow to create a new group before adding the asset."
+            ),
+            inputs=[
+                IO.Image.Input("image", tooltip="Image to register as a personal asset."),
+                IO.String.Input(
+                    "group_id",
+                    default="",
+                    tooltip="Reuse an existing Seedance asset group ID to skip repeated human verification for the "
+                    "same person. Leave empty to run real-person authentication in the browser and create a new group.",
+                ),
+                # IO.String.Input(
+                #     "name",
+                #     default="",
+                #     tooltip="Asset name (up to 64 characters).",
+                # ),
+            ],
+            outputs=[
+                IO.String.Output(display_name="asset_id"),
+                IO.String.Output(display_name="group_id"),
+            ],
+            hidden=[
+                IO.Hidden.auth_token_comfy_org,
+                IO.Hidden.api_key_comfy_org,
+                IO.Hidden.unique_id,
+            ],
+            # is_api_node=True,
+        )
+
+    @classmethod
+    async def execute(
+        cls,
+        image: Input.Image,
+        group_id: str = "",
+        # name: str = "",
+    ) -> IO.NodeOutput:
+        # if len(name) > 64:
+        #     raise ValueError("Name of asset can not be greater then 64 symbols")
+        validate_image_dimensions(image, min_width=300, max_width=6000, min_height=300, max_height=6000)
+        validate_image_aspect_ratio(image, min_ratio=(0.4, 1), max_ratio=(2.5, 1))
+        resolved_group = await _resolve_group_id(cls, group_id)
+        asset_id = await _create_seedance_asset(
+            cls,
+            group_id=resolved_group,
+            url=await upload_image_to_comfyapi(cls, image),
+            name="",
+            asset_type="Image",
+        )
+        await _wait_for_asset_active(cls, asset_id, resolved_group)
+        PromptServer.instance.send_progress_text(
+            f"Please save the asset_id and group_id for reuse.\n\nasset_id: {asset_id}\n\n"
+            f"group_id: {resolved_group}",
+            cls.hidden.unique_id,
+        )
+        return IO.NodeOutput(asset_id, resolved_group)
+
+
+class ByteDanceCreateVideoAsset(IO.ComfyNode):
+
+    @classmethod
+    def define_schema(cls) -> IO.Schema:
+        return IO.Schema(
+            node_id="ByteDanceCreateVideoAsset",
+            display_name="ByteDance Create Video Asset",
+            category="api node/video/ByteDance",
+            description=(
+                "Create a Seedance 2.0 personal video asset. Uploads the input video and "
+                "registers it in the given asset group. If group_id is empty, runs a real-person "
+                "H5 authentication flow to create a new group before adding the asset."
+            ),
+            inputs=[
+                IO.Video.Input("video", tooltip="Video to register as a personal asset."),
+                IO.String.Input(
+                    "group_id",
+                    default="",
+                    tooltip="Reuse an existing Seedance asset group ID to skip repeated human verification for the "
+                    "same person. Leave empty to run real-person authentication in the browser and create a new group.",
+                ),
+                # IO.String.Input(
+                #     "name",
+                #     default="",
+                #     tooltip="Asset name (up to 64 characters).",
+                # ),
+            ],
+            outputs=[
+                IO.String.Output(display_name="asset_id"),
+                IO.String.Output(display_name="group_id"),
+            ],
+            hidden=[
+                IO.Hidden.auth_token_comfy_org,
+                IO.Hidden.api_key_comfy_org,
+                IO.Hidden.unique_id,
+            ],
+            # is_api_node=True,
+        )
+
+    @classmethod
+    async def execute(
+        cls,
+        video: Input.Video,
+        group_id: str = "",
+        # name: str = "",
+    ) -> IO.NodeOutput:
+        # if len(name) > 64:
+        #     raise ValueError("Name of asset can not be greater then 64 symbols")
+        validate_video_duration(video, min_duration=2, max_duration=15)
+        validate_video_dimensions(video, min_width=300, max_width=6000, min_height=300, max_height=6000)
+
+        w, h = video.get_dimensions()
+        if h > 0:
+            ratio = w / h
+            if not (0.4 <= ratio <= 2.5):
+                raise ValueError(f"Asset video aspect ratio (W/H) must be in [0.4, 2.5], got {ratio:.3f} ({w}x{h}).")
+        pixels = w * h
+        if not (409_600 <= pixels <= 927_408):
+            raise ValueError(
+                f"Asset video total pixels (W×H) must be in [409600, 927408], " f"got {pixels:,} ({w}x{h})."
+            )
+
+        fps = float(video.get_frame_rate())
+        if not (24 <= fps <= 60):
+            raise ValueError(f"Asset video FPS must be in [24, 60], got {fps:.2f}.")
+
+        resolved_group = await _resolve_group_id(cls, group_id)
+        asset_id = await _create_seedance_asset(
+            cls,
+            group_id=resolved_group,
+            url=await upload_video_to_comfyapi(cls, video),
+            name="",
+            asset_type="Video",
+        )
+        await _wait_for_asset_active(cls, asset_id, resolved_group)
+        PromptServer.instance.send_progress_text(
+            f"Please save the asset_id and group_id for reuse.\n\nasset_id: {asset_id}\n\n"
+            f"group_id: {resolved_group}",
+            cls.hidden.unique_id,
+        )
+        return IO.NodeOutput(asset_id, resolved_group)
+
+
 class ByteDanceExtension(ComfyExtension):
     @override
     async def get_node_list(self) -> list[type[IO.ComfyNode]]:
@@ -1615,6 +2072,8 @@ class ByteDanceExtension(ComfyExtension):
             ByteDance2TextToVideoNode,
             ByteDance2FirstLastFrameNode,
             ByteDance2ReferenceNode,
+            ByteDanceCreateImageAsset,
+            ByteDanceCreateVideoAsset,
         ]
 
 

comfy_api_nodes/nodes_kling.py

@@ -276,6 +276,7 @@ async def finish_omni_video_task(cls: type[IO.ComfyNode], response: TaskStatusRe
         cls,
         ApiEndpoint(path=f"/proxy/kling/v1/videos/omni-video/{response.data.task_id}"),
         response_model=TaskStatusResponse,
+        max_poll_attempts=280,
         status_extractor=lambda r: (r.data.task_status if r.data else None),
     )
     return IO.NodeOutput(await download_url_to_video_output(final_response.data.task_result.videos[0].url))
@@ -862,7 +863,7 @@ class OmniProTextToVideoNode(IO.ComfyNode):
                 ),
                 IO.Combo.Input("aspect_ratio", options=["16:9", "9:16", "1:1"]),
                 IO.Int.Input("duration", default=5, min=3, max=15, display_mode=IO.NumberDisplay.slider),
-                IO.Combo.Input("resolution", options=["1080p", "720p"], optional=True),
+                IO.Combo.Input("resolution", options=["4k", "1080p", "720p"], default="1080p", optional=True),
                 IO.DynamicCombo.Input(
                     "storyboards",
                     options=[
@@ -904,12 +905,13 @@ class OmniProTextToVideoNode(IO.ComfyNode):
                 depends_on=IO.PriceBadgeDepends(widgets=["duration", "resolution", "model_name", "generate_audio"]),
                 expr="""
                 (
-                  $mode := (widgets.resolution = "720p") ? "std" : "pro";
+                  $res := widgets.resolution;
+                  $mode := $res = "4k" ? "4k" : ($res = "720p" ? "std" : "pro");
                   $isV3 := $contains(widgets.model_name, "v3");
                   $audio := $isV3 and widgets.generate_audio;
                   $rates := $audio
-                    ? {"std": 0.112, "pro": 0.14}
-                    : {"std": 0.084, "pro": 0.112};
+                    ? {"std": 0.112, "pro": 0.14, "4k": 0.42}
+                    : {"std": 0.084, "pro": 0.112, "4k": 0.42};
                   {"type":"usd","usd": $lookup($rates, $mode) * widgets.duration}
                 )
                 """,
@@ -934,6 +936,8 @@ class OmniProTextToVideoNode(IO.ComfyNode):
                 raise ValueError("kling-video-o1 only supports durations of 5 or 10 seconds.")
             if generate_audio:
                 raise ValueError("kling-video-o1 does not support audio generation.")
+            if resolution == "4k":
+                raise ValueError("kling-video-o1 does not support 4k resolution.")
         stories_enabled = storyboards is not None and storyboards["storyboards"] != "disabled"
         if stories_enabled and model_name == "kling-video-o1":
             raise ValueError("kling-video-o1 does not support storyboards.")
@@ -963,6 +967,12 @@ class OmniProTextToVideoNode(IO.ComfyNode):
                     f"must equal the global duration ({duration}s)."
                 )
 
+        if resolution == "4k":
+            mode = "4k"
+        elif resolution == "1080p":
+            mode = "pro"
+        else:
+            mode = "std"
         response = await sync_op(
             cls,
             ApiEndpoint(path="/proxy/kling/v1/videos/omni-video", method="POST"),
@@ -972,7 +982,7 @@ class OmniProTextToVideoNode(IO.ComfyNode):
                 prompt=prompt,
                 aspect_ratio=aspect_ratio,
                 duration=str(duration),
-                mode="pro" if resolution == "1080p" else "std",
+                mode=mode,
                 multi_shot=multi_shot,
                 multi_prompt=multi_prompt_list,
                 shot_type="customize" if multi_shot else None,
@@ -1014,7 +1024,7 @@ class OmniProFirstLastFrameNode(IO.ComfyNode):
                     optional=True,
                     tooltip="Up to 6 additional reference images.",
                 ),
-                IO.Combo.Input("resolution", options=["1080p", "720p"], optional=True),
+                IO.Combo.Input("resolution", options=["4k", "1080p", "720p"], default="1080p", optional=True),
                 IO.DynamicCombo.Input(
                     "storyboards",
                     options=[
@@ -1061,12 +1071,13 @@ class OmniProFirstLastFrameNode(IO.ComfyNode):
                 depends_on=IO.PriceBadgeDepends(widgets=["duration", "resolution", "model_name", "generate_audio"]),
                 expr="""
                 (
-                  $mode := (widgets.resolution = "720p") ? "std" : "pro";
+                  $res := widgets.resolution;
+                  $mode := $res = "4k" ? "4k" : ($res = "720p" ? "std" : "pro");
                   $isV3 := $contains(widgets.model_name, "v3");
                   $audio := $isV3 and widgets.generate_audio;
                   $rates := $audio
-                    ? {"std": 0.112, "pro": 0.14}
-                    : {"std": 0.084, "pro": 0.112};
+                    ? {"std": 0.112, "pro": 0.14, "4k": 0.42}
+                    : {"std": 0.084, "pro": 0.112, "4k": 0.42};
                   {"type":"usd","usd": $lookup($rates, $mode) * widgets.duration}
                 )
                 """,
@@ -1093,6 +1104,8 @@ class OmniProFirstLastFrameNode(IO.ComfyNode):
                 raise ValueError("kling-video-o1 does not support durations greater than 10 seconds.")
             if generate_audio:
                 raise ValueError("kling-video-o1 does not support audio generation.")
+            if resolution == "4k":
+                raise ValueError("kling-video-o1 does not support 4k resolution.")
         stories_enabled = storyboards is not None and storyboards["storyboards"] != "disabled"
         if stories_enabled and model_name == "kling-video-o1":
             raise ValueError("kling-video-o1 does not support storyboards.")
@@ -1161,6 +1174,12 @@ class OmniProFirstLastFrameNode(IO.ComfyNode):
                 validate_image_aspect_ratio(i, (1, 2.5), (2.5, 1))
             for i in await upload_images_to_comfyapi(cls, reference_images, wait_label="Uploading reference frame(s)"):
                 image_list.append(OmniParamImage(image_url=i))
+        if resolution == "4k":
+            mode = "4k"
+        elif resolution == "1080p":
+            mode = "pro"
+        else:
+            mode = "std"
         response = await sync_op(
             cls,
             ApiEndpoint(path="/proxy/kling/v1/videos/omni-video", method="POST"),
@@ -1170,7 +1189,7 @@ class OmniProFirstLastFrameNode(IO.ComfyNode):
                 prompt=prompt,
                 duration=str(duration),
                 image_list=image_list,
-                mode="pro" if resolution == "1080p" else "std",
+                mode=mode,
                 sound="on" if generate_audio else "off",
                 multi_shot=multi_shot,
                 multi_prompt=multi_prompt_list,
@@ -1204,7 +1223,7 @@ class OmniProImageToVideoNode(IO.ComfyNode):
                     "reference_images",
                     tooltip="Up to 7 reference images.",
                 ),
-                IO.Combo.Input("resolution", options=["1080p", "720p"], optional=True),
+                IO.Combo.Input("resolution", options=["4k", "1080p", "720p"], default="1080p", optional=True),
                 IO.DynamicCombo.Input(
                     "storyboards",
                     options=[
@@ -1251,12 +1270,13 @@ class OmniProImageToVideoNode(IO.ComfyNode):
                 depends_on=IO.PriceBadgeDepends(widgets=["duration", "resolution", "model_name", "generate_audio"]),
                 expr="""
                 (
-                  $mode := (widgets.resolution = "720p") ? "std" : "pro";
+                  $res := widgets.resolution;
+                  $mode := $res = "4k" ? "4k" : ($res = "720p" ? "std" : "pro");
                   $isV3 := $contains(widgets.model_name, "v3");
                   $audio := $isV3 and widgets.generate_audio;
                   $rates := $audio
-                    ? {"std": 0.112, "pro": 0.14}
-                    : {"std": 0.084, "pro": 0.112};
+                    ? {"std": 0.112, "pro": 0.14, "4k": 0.42}
+                    : {"std": 0.084, "pro": 0.112, "4k": 0.42};
                   {"type":"usd","usd": $lookup($rates, $mode) * widgets.duration}
                 )
                 """,
@@ -1282,6 +1302,8 @@ class OmniProImageToVideoNode(IO.ComfyNode):
                 raise ValueError("kling-video-o1 does not support durations greater than 10 seconds.")
             if generate_audio:
                 raise ValueError("kling-video-o1 does not support audio generation.")
+            if resolution == "4k":
+                raise ValueError("kling-video-o1 does not support 4k resolution.")
         stories_enabled = storyboards is not None and storyboards["storyboards"] != "disabled"
         if stories_enabled and model_name == "kling-video-o1":
             raise ValueError("kling-video-o1 does not support storyboards.")
@@ -1320,6 +1342,12 @@ class OmniProImageToVideoNode(IO.ComfyNode):
         image_list: list[OmniParamImage] = []
         for i in await upload_images_to_comfyapi(cls, reference_images, wait_label="Uploading reference image"):
             image_list.append(OmniParamImage(image_url=i))
+        if resolution == "4k":
+            mode = "4k"
+        elif resolution == "1080p":
+            mode = "pro"
+        else:
+            mode = "std"
         response = await sync_op(
             cls,
             ApiEndpoint(path="/proxy/kling/v1/videos/omni-video", method="POST"),
@@ -1330,7 +1358,7 @@ class OmniProImageToVideoNode(IO.ComfyNode):
                 aspect_ratio=aspect_ratio,
                 duration=str(duration),
                 image_list=image_list,
-                mode="pro" if resolution == "1080p" else "std",
+                mode=mode,
                 sound="on" if generate_audio else "off",
                 multi_shot=multi_shot,
                 multi_prompt=multi_prompt_list,
@@ -2860,7 +2888,7 @@ class KlingVideoNode(IO.ComfyNode):
                         IO.DynamicCombo.Option(
                             "kling-v3",
                             [
-                                IO.Combo.Input("resolution", options=["1080p", "720p"]),
+                                IO.Combo.Input("resolution", options=["4k", "1080p", "720p"], default="1080p"),
                                 IO.Combo.Input(
                                     "aspect_ratio",
                                     options=["16:9", "9:16", "1:1"],
@@ -2913,7 +2941,11 @@ class KlingVideoNode(IO.ComfyNode):
                 ),
                 expr="""
                 (
-                  $rates := {"1080p": {"off": 0.112, "on": 0.168}, "720p": {"off": 0.084, "on": 0.126}};
+                  $rates := {
+                    "4k": {"off": 0.42, "on": 0.42},
+                    "1080p": {"off": 0.112, "on": 0.168},
+                    "720p": {"off": 0.084, "on": 0.126}
+                  };
                   $res := $lookup(widgets, "model.resolution");
                   $audio := widgets.generate_audio ? "on" : "off";
                   $rate := $lookup($lookup($rates, $res), $audio);
@@ -2943,7 +2975,12 @@ class KlingVideoNode(IO.ComfyNode):
         start_frame: Input.Image | None = None,
     ) -> IO.NodeOutput:
         _ = seed
-        mode = "pro" if model["resolution"] == "1080p" else "std"
+        if model["resolution"] == "4k":
+            mode = "4k"
+        elif model["resolution"] == "1080p":
+            mode = "pro"
+        else:
+            mode = "std"
         custom_multi_shot = False
         if multi_shot["multi_shot"] == "disabled":
             shot_type = None
@@ -3025,6 +3062,7 @@ class KlingVideoNode(IO.ComfyNode):
             cls,
             ApiEndpoint(path=poll_path),
             response_model=TaskStatusResponse,
+            max_poll_attempts=280,
             status_extractor=lambda r: (r.data.task_status if r.data else None),
         )
         return IO.NodeOutput(await download_url_to_video_output(final_response.data.task_result.videos[0].url))
@@ -3057,7 +3095,7 @@ class KlingFirstLastFrameNode(IO.ComfyNode):
                         IO.DynamicCombo.Option(
                             "kling-v3",
                             [
-                                IO.Combo.Input("resolution", options=["1080p", "720p"]),
+                                IO.Combo.Input("resolution", options=["4k", "1080p", "720p"], default="1080p"),
                             ],
                         ),
                     ],
@@ -3089,7 +3127,11 @@ class KlingFirstLastFrameNode(IO.ComfyNode):
                 ),
                 expr="""
                 (
-                  $rates := {"1080p": {"off": 0.112, "on": 0.168}, "720p": {"off": 0.084, "on": 0.126}};
+                  $rates := {
+                    "4k": {"off": 0.42, "on": 0.42},
+                    "1080p": {"off": 0.112, "on": 0.168},
+                    "720p": {"off": 0.084, "on": 0.126}
+                  };
                   $res := $lookup(widgets, "model.resolution");
                   $audio := widgets.generate_audio ? "on" : "off";
                   $rate := $lookup($lookup($rates, $res), $audio);
@@ -3118,6 +3160,12 @@ class KlingFirstLastFrameNode(IO.ComfyNode):
         validate_image_aspect_ratio(end_frame, (1, 2.5), (2.5, 1))
         image_url = await upload_image_to_comfyapi(cls, first_frame, wait_label="Uploading first frame")
         image_tail_url = await upload_image_to_comfyapi(cls, end_frame, wait_label="Uploading end frame")
+        if model["resolution"] == "4k":
+            mode = "4k"
+        elif model["resolution"] == "1080p":
+            mode = "pro"
+        else:
+            mode = "std"
         response = await sync_op(
             cls,
             ApiEndpoint(path="/proxy/kling/v1/videos/image2video", method="POST"),
@@ -3127,7 +3175,7 @@ class KlingFirstLastFrameNode(IO.ComfyNode):
                 image=image_url,
                 image_tail=image_tail_url,
                 prompt=prompt,
-                mode="pro" if model["resolution"] == "1080p" else "std",
+                mode=mode,
                 duration=str(duration),
                 sound="on" if generate_audio else "off",
             ),
@@ -3140,6 +3188,7 @@ class KlingFirstLastFrameNode(IO.ComfyNode):
             cls,
             ApiEndpoint(path=f"/proxy/kling/v1/videos/image2video/{response.data.task_id}"),
             response_model=TaskStatusResponse,
+            max_poll_attempts=280,
             status_extractor=lambda r: (r.data.task_status if r.data else None),
         )
         return IO.NodeOutput(await download_url_to_video_output(final_response.data.task_result.videos[0].url))

comfy_api_nodes/nodes_openai.py

@@ -357,13 +357,17 @@ def calculate_tokens_price_image_1_5(response: OpenAIImageGenerationResponse) ->
     return ((response.usage.input_tokens * 8.0) + (response.usage.output_tokens * 32.0)) / 1_000_000.0
 
 
+def calculate_tokens_price_image_2_0(response: OpenAIImageGenerationResponse) -> float | None:
+    return ((response.usage.input_tokens * 8.0) + (response.usage.output_tokens * 30.0)) / 1_000_000.0
+
+
 class OpenAIGPTImage1(IO.ComfyNode):
 
     @classmethod
     def define_schema(cls):
         return IO.Schema(
             node_id="OpenAIGPTImage1",
-            display_name="OpenAI GPT Image 1.5",
+            display_name="OpenAI GPT Image 2",
             category="api node/image/OpenAI",
             description="Generates images synchronously via OpenAI's GPT Image endpoint.",
             inputs=[
@@ -401,7 +405,17 @@ class OpenAIGPTImage1(IO.ComfyNode):
                 IO.Combo.Input(
                     "size",
                     default="auto",
-                    options=["auto", "1024x1024", "1024x1536", "1536x1024"],
+                    options=[
+                        "auto",
+                        "1024x1024",
+                        "1024x1536",
+                        "1536x1024",
+                        "2048x2048",
+                        "2048x1152",
+                        "1152x2048",
+                        "3840x2160",
+                        "2160x3840",
+                    ],
                     tooltip="Image size",
                     optional=True,
                 ),
@@ -427,8 +441,8 @@ class OpenAIGPTImage1(IO.ComfyNode):
                 ),
                 IO.Combo.Input(
                     "model",
-                    options=["gpt-image-1", "gpt-image-1.5"],
-                    default="gpt-image-1.5",
+                    options=["gpt-image-1", "gpt-image-1.5", "gpt-image-2"],
+                    default="gpt-image-2",
                     optional=True,
                 ),
             ],
@@ -442,23 +456,36 @@ class OpenAIGPTImage1(IO.ComfyNode):
             ],
             is_api_node=True,
             price_badge=IO.PriceBadge(
-                depends_on=IO.PriceBadgeDepends(widgets=["quality", "n"]),
+                depends_on=IO.PriceBadgeDepends(widgets=["quality", "n", "model"]),
                 expr="""
                 (
                   $ranges := {
-                    "low":    [0.011, 0.02],
-                    "medium": [0.046, 0.07],
-                    "high":   [0.167, 0.3]
+                    "gpt-image-1": {
+                      "low":    [0.011, 0.02],
+                      "medium": [0.042, 0.07],
+                      "high":   [0.167, 0.25]
+                    },
+                    "gpt-image-1.5": {
+                      "low":    [0.009, 0.02],
+                      "medium": [0.034, 0.062],
+                      "high":   [0.133, 0.22]
+                    },
+                    "gpt-image-2": {
+                      "low":    [0.0048, 0.012],
+                      "medium": [0.041, 0.112],
+                      "high":   [0.165, 0.43]
+                    }
                   };
-                  $range := $lookup($ranges, widgets.quality);
-                  $n := widgets.n;
+                  $range := $lookup($lookup($ranges, widgets.model), widgets.quality);
+                  $nRaw := widgets.n;
+                  $n := ($nRaw != null and $nRaw != 0) ? $nRaw : 1;
                   ($n = 1)
-                    ? {"type":"range_usd","min_usd": $range[0], "max_usd": $range[1]}
+                    ? {"type":"range_usd","min_usd": $range[0], "max_usd": $range[1], "format": {"approximate": true}}
                     : {
                         "type":"range_usd",
-                        "min_usd": $range[0],
-                        "max_usd": $range[1],
-                        "format": { "suffix": " x " & $string($n) & "/Run" }
+                        "min_usd": $range[0] * $n,
+                        "max_usd": $range[1] * $n,
+                        "format": { "suffix": "/Run", "approximate": true }
                       }
                 )
                 """,
@@ -483,10 +510,18 @@ class OpenAIGPTImage1(IO.ComfyNode):
         if mask is not None and image is None:
             raise ValueError("Cannot use a mask without an input image")
 
+        if model in ("gpt-image-1", "gpt-image-1.5"):
+            if size not in ("auto", "1024x1024", "1024x1536", "1536x1024"):
+                raise ValueError(f"Resolution {size} is only supported by GPT Image 2 model")
+
         if model == "gpt-image-1":
             price_extractor = calculate_tokens_price_image_1
         elif model == "gpt-image-1.5":
             price_extractor = calculate_tokens_price_image_1_5
+        elif model == "gpt-image-2":
+            price_extractor = calculate_tokens_price_image_2_0
+            if background == "transparent":
+                raise ValueError("Transparent background is not supported for GPT Image 2 model")
         else:
             raise ValueError(f"Unknown model: {model}")
 

comfy_api_nodes/nodes_veo2.py

@@ -24,8 +24,9 @@ from comfy_api_nodes.util import (
 AVERAGE_DURATION_VIDEO_GEN = 32
 MODELS_MAP = {
     "veo-2.0-generate-001": "veo-2.0-generate-001",
-    "veo-3.1-generate": "veo-3.1-generate-preview",
-    "veo-3.1-fast-generate": "veo-3.1-fast-generate-preview",
+    "veo-3.1-generate": "veo-3.1-generate-001",
+    "veo-3.1-fast-generate": "veo-3.1-fast-generate-001",
+    "veo-3.1-lite": "veo-3.1-lite-generate-001",
     "veo-3.0-generate-001": "veo-3.0-generate-001",
     "veo-3.0-fast-generate-001": "veo-3.0-fast-generate-001",
 }
@@ -247,17 +248,8 @@ class VeoVideoGenerationNode(IO.ComfyNode):
         raise Exception("Video generation completed but no video was returned")
 
 
-class Veo3VideoGenerationNode(VeoVideoGenerationNode):
-    """
-    Generates videos from text prompts using Google's Veo 3 API.
-
-    Supported models:
-    - veo-3.0-generate-001
-    - veo-3.0-fast-generate-001
-
-    This node extends the base Veo node with Veo 3 specific features including
-    audio generation and fixed 8-second duration.
-    """
+class Veo3VideoGenerationNode(IO.ComfyNode):
+    """Generates videos from text prompts using Google's Veo 3 API."""
 
     @classmethod
     def define_schema(cls):
@@ -279,6 +271,13 @@ class Veo3VideoGenerationNode(VeoVideoGenerationNode):
                     default="16:9",
                     tooltip="Aspect ratio of the output video",
                 ),
+                IO.Combo.Input(
+                    "resolution",
+                    options=["720p", "1080p", "4k"],
+                    default="720p",
+                    tooltip="Output video resolution. 4K is not available for veo-3.1-lite and veo-3.0 models.",
+                    optional=True,
+                ),
                 IO.String.Input(
                     "negative_prompt",
                     multiline=True,
@@ -289,11 +288,11 @@ class Veo3VideoGenerationNode(VeoVideoGenerationNode):
                 IO.Int.Input(
                     "duration_seconds",
                     default=8,
-                    min=8,
+                    min=4,
                     max=8,
-                    step=1,
+                    step=2,
                     display_mode=IO.NumberDisplay.number,
-                    tooltip="Duration of the output video in seconds (Veo 3 only supports 8 seconds)",
+                    tooltip="Duration of the output video in seconds",
                     optional=True,
                 ),
                 IO.Boolean.Input(
@@ -332,10 +331,10 @@ class Veo3VideoGenerationNode(VeoVideoGenerationNode):
                     options=[
                         "veo-3.1-generate",
                         "veo-3.1-fast-generate",
+                        "veo-3.1-lite",
                         "veo-3.0-generate-001",
                         "veo-3.0-fast-generate-001",
                     ],
-                    default="veo-3.0-generate-001",
                     tooltip="Veo 3 model to use for video generation",
                     optional=True,
                 ),
@@ -356,21 +355,111 @@ class Veo3VideoGenerationNode(VeoVideoGenerationNode):
             ],
             is_api_node=True,
             price_badge=IO.PriceBadge(
-                depends_on=IO.PriceBadgeDepends(widgets=["model", "generate_audio"]),
+                depends_on=IO.PriceBadgeDepends(widgets=["model", "generate_audio", "resolution", "duration_seconds"]),
                 expr="""
                 (
                   $m := widgets.model;
+                  $r := widgets.resolution;
                   $a := widgets.generate_audio;
-                  ($contains($m,"veo-3.0-fast-generate-001") or $contains($m,"veo-3.1-fast-generate"))
-                    ? {"type":"usd","usd": ($a ? 1.2 : 0.8)}
-                    : ($contains($m,"veo-3.0-generate-001") or $contains($m,"veo-3.1-generate"))
-                      ? {"type":"usd","usd": ($a ? 3.2 : 1.6)}
-                      : {"type":"range_usd","min_usd":0.8,"max_usd":3.2}
+                  $seconds := widgets.duration_seconds;
+                  $pps :=
+                    $contains($m, "lite")
+                      ? ($r = "1080p" ? ($a ? 0.08 : 0.05) : ($a ? 0.05 : 0.03))
+                    : $contains($m, "3.1-fast")
+                      ? ($r = "4k" ? ($a ? 0.30 : 0.25) : $r = "1080p" ? ($a ? 0.12 : 0.10) : ($a ? 0.10 : 0.08))
+                    : $contains($m, "3.1-generate")
+                      ? ($r = "4k" ? ($a ? 0.60 : 0.40) : ($a ? 0.40 : 0.20))
+                    : $contains($m, "3.0-fast")
+                      ? ($a ? 0.15 : 0.10)
+                    : ($a ? 0.40 : 0.20);
+                  {"type":"usd","usd": $pps * $seconds}
                 )
                 """,
             ),
         )
 
+    @classmethod
+    async def execute(
+        cls,
+        prompt,
+        aspect_ratio="16:9",
+        resolution="720p",
+        negative_prompt="",
+        duration_seconds=8,
+        enhance_prompt=True,
+        person_generation="ALLOW",
+        seed=0,
+        image=None,
+        model="veo-3.0-generate-001",
+        generate_audio=False,
+    ):
+        if resolution == "4k" and ("lite" in model or "3.0" in model):
+            raise Exception("4K resolution is not supported by the veo-3.1-lite or veo-3.0 models.")
+
+        model = MODELS_MAP[model]
+
+        instances = [{"prompt": prompt}]
+        if image is not None:
+            image_base64 = tensor_to_base64_string(image)
+            if image_base64:
+                instances[0]["image"] = {"bytesBase64Encoded": image_base64, "mimeType": "image/png"}
+
+        parameters = {
+            "aspectRatio": aspect_ratio,
+            "personGeneration": person_generation,
+            "durationSeconds": duration_seconds,
+            "enhancePrompt": True,
+            "generateAudio": generate_audio,
+        }
+        if negative_prompt:
+            parameters["negativePrompt"] = negative_prompt
+        if seed > 0:
+            parameters["seed"] = seed
+        if "veo-3.1" in model:
+            parameters["resolution"] = resolution
+
+        initial_response = await sync_op(
+            cls,
+            ApiEndpoint(path=f"/proxy/veo/{model}/generate", method="POST"),
+            response_model=VeoGenVidResponse,
+            data=VeoGenVidRequest(
+                instances=instances,
+                parameters=parameters,
+            ),
+        )
+
+        poll_response = await poll_op(
+            cls,
+            ApiEndpoint(path=f"/proxy/veo/{model}/poll", method="POST"),
+            response_model=VeoGenVidPollResponse,
+            status_extractor=lambda r: "completed" if r.done else "pending",
+            data=VeoGenVidPollRequest(operationName=initial_response.name),
+            poll_interval=9.0,
+            estimated_duration=AVERAGE_DURATION_VIDEO_GEN,
+        )
+
+        if poll_response.error:
+            raise Exception(f"Veo API error: {poll_response.error.message} (code: {poll_response.error.code})")
+
+        response = poll_response.response
+        filtered_count = response.raiMediaFilteredCount
+        if filtered_count:
+            reasons = response.raiMediaFilteredReasons or []
+            reason_part = f": {reasons[0]}" if reasons else ""
+            raise Exception(
+                f"Content blocked by Google's Responsible AI filters{reason_part} "
+                f"({filtered_count} video{'s' if filtered_count != 1 else ''} filtered)."
+            )
+
+        if response.videos:
+            video = response.videos[0]
+            if video.bytesBase64Encoded:
+                return IO.NodeOutput(InputImpl.VideoFromFile(BytesIO(base64.b64decode(video.bytesBase64Encoded))))
+            if video.gcsUri:
+                return IO.NodeOutput(await download_url_to_video_output(video.gcsUri))
+            raise Exception("Video returned but no data or URL was provided")
+        raise Exception("Video generation completed but no video was returned")
+
 
 class Veo3FirstLastFrameNode(IO.ComfyNode):
 
@@ -394,7 +483,7 @@ class Veo3FirstLastFrameNode(IO.ComfyNode):
                     default="",
                     tooltip="Negative text prompt to guide what to avoid in the video",
                 ),
-                IO.Combo.Input("resolution", options=["720p", "1080p"]),
+                IO.Combo.Input("resolution", options=["720p", "1080p", "4k"]),
                 IO.Combo.Input(
                     "aspect_ratio",
                     options=["16:9", "9:16"],
@@ -424,8 +513,7 @@ class Veo3FirstLastFrameNode(IO.ComfyNode):
                 IO.Image.Input("last_frame", tooltip="End frame"),
                 IO.Combo.Input(
                     "model",
-                    options=["veo-3.1-generate", "veo-3.1-fast-generate"],
-                    default="veo-3.1-fast-generate",
+                    options=["veo-3.1-generate", "veo-3.1-fast-generate", "veo-3.1-lite"],
                 ),
                 IO.Boolean.Input(
                     "generate_audio",
@@ -443,26 +531,20 @@ class Veo3FirstLastFrameNode(IO.ComfyNode):
             ],
             is_api_node=True,
             price_badge=IO.PriceBadge(
-                depends_on=IO.PriceBadgeDepends(widgets=["model", "generate_audio", "duration"]),
+                depends_on=IO.PriceBadgeDepends(widgets=["model", "generate_audio", "duration", "resolution"]),
                 expr="""
                 (
-                  $prices := {
-                    "veo-3.1-fast-generate": { "audio": 0.15, "no_audio": 0.10 },
-                    "veo-3.1-generate":      { "audio": 0.40, "no_audio": 0.20 }
-                  };
                   $m := widgets.model;
-                  $ga := (widgets.generate_audio = "true");
+                  $r := widgets.resolution;
+                  $ga := widgets.generate_audio;
                   $seconds := widgets.duration;
-                  $modelKey :=
-                    $contains($m, "veo-3.1-fast-generate") ? "veo-3.1-fast-generate" :
-                    $contains($m, "veo-3.1-generate")      ? "veo-3.1-generate" :
-                    "";
-                  $audioKey := $ga ? "audio" : "no_audio";
-                  $modelPrices := $lookup($prices, $modelKey);
-                  $pps := $lookup($modelPrices, $audioKey);
-                  ($pps != null)
-                    ? {"type":"usd","usd": $pps * $seconds}
-                    : {"type":"range_usd","min_usd": 0.4, "max_usd": 3.2}
+                  $pps :=
+                    $contains($m, "lite")
+                      ? ($r = "1080p" ? ($ga ? 0.08 : 0.05) : ($ga ? 0.05 : 0.03))
+                    : $contains($m, "fast")
+                      ? ($r = "4k" ? ($ga ? 0.30 : 0.25) : $r = "1080p" ? ($ga ? 0.12 : 0.10) : ($ga ? 0.10 : 0.08))
+                    : ($r = "4k" ? ($ga ? 0.60 : 0.40) : ($ga ? 0.40 : 0.20));
+                  {"type":"usd","usd": $pps * $seconds}
                 )
                 """,
             ),
@@ -482,6 +564,9 @@ class Veo3FirstLastFrameNode(IO.ComfyNode):
         model: str,
         generate_audio: bool,
     ):
+        if "lite" in model and resolution == "4k":
+            raise Exception("4K resolution is not supported by the veo-3.1-lite model.")
+
         model = MODELS_MAP[model]
         initial_response = await sync_op(
             cls,
@@ -519,7 +604,7 @@ class Veo3FirstLastFrameNode(IO.ComfyNode):
             data=VeoGenVidPollRequest(
                 operationName=initial_response.name,
             ),
-            poll_interval=5.0,
+            poll_interval=9.0,
             estimated_duration=AVERAGE_DURATION_VIDEO_GEN,
         )
 

comfy_api_nodes/util/__init__.py

@@ -19,6 +19,7 @@ from .conversions import (
     image_tensor_pair_to_batch,
     pil_to_bytesio,
     resize_mask_to_image,
+    resize_video_to_pixel_budget,
     tensor_to_base64_string,
     tensor_to_bytesio,
     tensor_to_pil,
@@ -90,6 +91,7 @@ __all__ = [
     "image_tensor_pair_to_batch",
     "pil_to_bytesio",
     "resize_mask_to_image",
+    "resize_video_to_pixel_budget",
     "tensor_to_base64_string",
     "tensor_to_bytesio",
     "tensor_to_pil",

comfy_api_nodes/util/client.py

@@ -156,6 +156,7 @@ async def poll_op(
     estimated_duration: int | None = None,
     cancel_endpoint: ApiEndpoint | None = None,
     cancel_timeout: float = 10.0,
+    extra_text: str | None = None,
 ) -> M:
     raw = await poll_op_raw(
         cls,
@@ -176,6 +177,7 @@ async def poll_op(
         estimated_duration=estimated_duration,
         cancel_endpoint=cancel_endpoint,
         cancel_timeout=cancel_timeout,
+        extra_text=extra_text,
     )
     if not isinstance(raw, dict):
         raise Exception("Expected JSON response to validate into a Pydantic model, got non-JSON (binary or text).")
@@ -260,6 +262,7 @@ async def poll_op_raw(
     estimated_duration: int | None = None,
     cancel_endpoint: ApiEndpoint | None = None,
     cancel_timeout: float = 10.0,
+    extra_text: str | None = None,
 ) -> dict[str, Any]:
     """
     Polls an endpoint until the task reaches a terminal state. Displays time while queued/processing,
@@ -299,6 +302,7 @@ async def poll_op_raw(
                     price=state.price,
                     is_queued=state.is_queued,
                     processing_elapsed_seconds=int(proc_elapsed),
+                    extra_text=extra_text,
                 )
                 await asyncio.sleep(1.0)
         except Exception as exc:
@@ -389,6 +393,7 @@ async def poll_op_raw(
                     price=state.price,
                     is_queued=False,
                     processing_elapsed_seconds=int(state.base_processing_elapsed),
+                    extra_text=extra_text,
                 )
                 return resp_json
 
@@ -462,6 +467,7 @@ def _display_time_progress(
     price: float | None = None,
     is_queued: bool | None = None,
     processing_elapsed_seconds: int | None = None,
+    extra_text: str | None = None,
 ) -> None:
     if estimated_total is not None and estimated_total > 0 and is_queued is False:
         pe = processing_elapsed_seconds if processing_elapsed_seconds is not None else elapsed_seconds
@@ -469,7 +475,8 @@ def _display_time_progress(
         time_line = f"Time elapsed: {int(elapsed_seconds)}s (~{remaining}s remaining)"
     else:
         time_line = f"Time elapsed: {int(elapsed_seconds)}s"
-    _display_text(node_cls, time_line, status=status, price=price)
+    text = f"{time_line}\n\n{extra_text}" if extra_text else time_line
+    _display_text(node_cls, text, status=status, price=price)
 
 
 async def _diagnose_connectivity() -> dict[str, bool]:

comfy_api_nodes/util/conversions.py

@@ -129,22 +129,38 @@ def pil_to_bytesio(img: Image.Image, mime_type: str = "image/png") -> BytesIO:
     return img_byte_arr
 
 
+def _compute_downscale_dims(src_w: int, src_h: int, total_pixels: int) -> tuple[int, int] | None:
+    """Return downscaled (w, h) with even dims fitting ``total_pixels``, or None if already fits.
+
+    Source aspect ratio is preserved; output may drift by a fraction of a percent because both dimensions
+    are rounded down to even values (many  codecs require divisible-by-2).
+    """
+    pixels = src_w * src_h
+    if pixels <= total_pixels:
+        return None
+    scale = math.sqrt(total_pixels / pixels)
+    new_w = max(2, int(src_w * scale))
+    new_h = max(2, int(src_h * scale))
+    new_w -= new_w % 2
+    new_h -= new_h % 2
+    return new_w, new_h
+
+
 def downscale_image_tensor(image: torch.Tensor, total_pixels: int = 1536 * 1024) -> torch.Tensor:
-    """Downscale input image tensor to roughly the specified total pixels."""
+    """Downscale input image tensor to roughly the specified total pixels.
+
+    Output dimensions are rounded down to even values so that the result is guaranteed to fit within ``total_pixels``
+    and is compatible with codecs that require even dimensions (e.g. yuv420p).
+    """
     samples = image.movedim(-1, 1)
-    total = int(total_pixels)
-    scale_by = math.sqrt(total / (samples.shape[3] * samples.shape[2]))
-    if scale_by >= 1:
+    dims = _compute_downscale_dims(samples.shape[3], samples.shape[2], int(total_pixels))
+    if dims is None:
         return image
-    width = round(samples.shape[3] * scale_by)
-    height = round(samples.shape[2] * scale_by)
-
-    s = common_upscale(samples, width, height, "lanczos", "disabled")
-    s = s.movedim(1, -1)
-    return s
+    new_w, new_h = dims
+    return common_upscale(samples, new_w, new_h, "lanczos", "disabled").movedim(1, -1)
 
 
-def downscale_image_tensor_by_max_side(image: torch.Tensor, *,  max_side: int) -> torch.Tensor:
+def downscale_image_tensor_by_max_side(image: torch.Tensor, *, max_side: int) -> torch.Tensor:
     """Downscale input image tensor so the largest dimension is at most max_side pixels."""
     samples = image.movedim(-1, 1)
     height, width = samples.shape[2], samples.shape[3]
@@ -399,6 +415,72 @@ def trim_video(video: Input.Video, duration_sec: float) -> Input.Video:
         raise RuntimeError(f"Failed to trim video: {str(e)}") from e
 
 
+def resize_video_to_pixel_budget(video: Input.Video, total_pixels: int) -> Input.Video:
+    """Downscale a video to fit within ``total_pixels`` (w * h), preserving aspect ratio.
+
+    Returns the original video object untouched when it already fits. Preserves frame rate, duration, and audio.
+    Aspect ratio is preserved up to a fraction of a percent (even-dim rounding).
+    """
+    src_w, src_h = video.get_dimensions()
+    scale_dims = _compute_downscale_dims(src_w, src_h, total_pixels)
+    if scale_dims is None:
+        return video
+    return _apply_video_scale(video, scale_dims)
+
+
+def _apply_video_scale(video: Input.Video, scale_dims: tuple[int, int]) -> Input.Video:
+    """Re-encode ``video`` scaled to ``scale_dims`` with a single decode/encode pass."""
+    out_w, out_h = scale_dims
+    output_buffer = BytesIO()
+    input_container = None
+    output_container = None
+
+    try:
+        input_source = video.get_stream_source()
+        input_container = av.open(input_source, mode="r")
+        output_container = av.open(output_buffer, mode="w", format="mp4")
+
+        video_stream = output_container.add_stream("h264", rate=video.get_frame_rate())
+        video_stream.width = out_w
+        video_stream.height = out_h
+        video_stream.pix_fmt = "yuv420p"
+
+        audio_stream = None
+        for stream in input_container.streams:
+            if isinstance(stream, av.AudioStream):
+                audio_stream = output_container.add_stream("aac", rate=stream.sample_rate)
+                audio_stream.sample_rate = stream.sample_rate
+                audio_stream.layout = stream.layout
+                break
+
+        for frame in input_container.decode(video=0):
+            frame = frame.reformat(width=out_w, height=out_h, format="yuv420p")
+            for packet in video_stream.encode(frame):
+                output_container.mux(packet)
+        for packet in video_stream.encode():
+            output_container.mux(packet)
+
+        if audio_stream is not None:
+            input_container.seek(0)
+            for audio_frame in input_container.decode(audio=0):
+                for packet in audio_stream.encode(audio_frame):
+                    output_container.mux(packet)
+            for packet in audio_stream.encode():
+                output_container.mux(packet)
+
+        output_container.close()
+        input_container.close()
+        output_buffer.seek(0)
+        return InputImpl.VideoFromFile(output_buffer)
+
+    except Exception as e:
+        if input_container is not None:
+            input_container.close()
+        if output_container is not None:
+            output_container.close()
+        raise RuntimeError(f"Failed to resize video: {str(e)}") from e
+
+
 def _f32_pcm(wav: torch.Tensor) -> torch.Tensor:
     """Convert audio to float 32 bits PCM format. Copy-paste from nodes_audio.py file."""
     if wav.dtype.is_floating_point:

comfyui_version.py

@@ -1,3 +1,3 @@
 # This file is automatically generated by the build process when version is
 # updated in pyproject.toml.
-__version__ = "0.19.3"
+__version__ = "0.19.5"

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ComfyUI"
-version = "0.19.3"
+version = "0.19.5"
 readme = "README.md"
 license = { file = "LICENSE" }
 requires-python = ">=3.10"

requirements.txt

@@ -1,5 +1,5 @@
-comfyui-frontend-package==1.42.11
-comfyui-workflow-templates==0.9.57
+comfyui-frontend-package==1.42.14
+comfyui-workflow-templates==0.9.61
 comfyui-embedded-docs==0.4.3
 torch
 torchsde