ComfyUI backport release v0.33.4

comfy/model_prefetch.py

@@ -47,7 +47,7 @@ def cleanup_prefetch_queues():
     GRAPH_CAPTURE_STREAMS = {}
 
 def prefetch_queue_pop(queue, device, module, dtype=None, core=None, enable_graph=False, generator=None):
-    enable_graph = enable_graph and not args.disable_cuda_graphs and comfy.model_management.is_device_cuda(device)
+    enable_graph = enable_graph and not args.disable_cuda_graphs and comfy.model_management.is_device_cuda(device) and getattr(module, "_v_block", None) is not None
     if queue is None:
         if core is not None:
             core()

comfy_api_nodes/apis/bfl.py

@@ -166,3 +166,13 @@ class Flux3VideoContinuationRequest(Flux3VideoRequest):
     start_video: str = Field(
         ..., description="MP4 (URL or base64); the new clip carries on from its final frames."
     )
+
+
+class BFLFluxVideoUpscaleRequest(BaseModel):
+    model_config = ConfigDict(extra="forbid")
+
+    input_video: str = Field(..., description="MP4 (URL or base64), 1 to 20 seconds.")
+    upscale_factor: float = Field(2.0, ge=1.5, le=3.0)
+    creativity: int = Field(1, description="0 preserves the source precisely, 1 enhances detail.")
+    prompt: str | None = Field(None)
+    safety_tolerance: int = Field(2, ge=0, le=4)

comfy_api_nodes/apis/bytedance.py

@@ -18,7 +18,8 @@ class Seedream4Options(BaseModel):
 
 
 class Seedream5OptimizePromptOptions(BaseModel):
-    thinking: Literal["auto", "enabled", "disabled"] = Field(...)
+    thinking: Literal["auto", "enabled", "disabled"] | None = Field(None)
+    mode: Literal["standard", "fast"] | None = Field(None)
 
 
 class Seedream4TaskCreationRequest(BaseModel):
@@ -116,6 +117,7 @@ class Seedance2TaskCreationRequest(BaseModel):
     seed: int | None = Field(None, ge=0, le=2147483647)
     watermark: bool | None = Field(None)
     output_format: str | None = Field(None)
+    omni_reference_task_type: str | None = Field(None, description="One of: auto, reference, edit, extend.")
 
 
 class TaskCreationResponse(BaseModel):
@@ -186,35 +188,6 @@ class SeedanceVirtualLibraryCreateAssetRequest(BaseModel):
     asset_type: str | None = Field(None, description="BytePlus asset type. Defaults to Image server-side when omitted.")
 
 
-# Dollars per 1K tokens, keyed by (model_id, has_video_input, resolution).
-SEEDANCE2_PRICE_PER_1K_TOKENS = {
-    ("dreamina-seedance-2-0-260128", False, "480p"): 0.007,
-    ("dreamina-seedance-2-0-260128", True, "480p"): 0.0043,
-    ("dreamina-seedance-2-0-260128", False, "720p"): 0.007,
-    ("dreamina-seedance-2-0-260128", True, "720p"): 0.0043,
-    ("dreamina-seedance-2-0-260128", False, "1080p"): 0.0077,
-    ("dreamina-seedance-2-0-260128", True, "1080p"): 0.0047,
-    ("dreamina-seedance-2-0-260128", False, "4k"): 0.004,
-    ("dreamina-seedance-2-0-260128", True, "4k"): 0.0024,
-    ("dreamina-seedance-2-0-fast-260128", False, "480p"): 0.0056,
-    ("dreamina-seedance-2-0-fast-260128", True, "480p"): 0.0033,
-    ("dreamina-seedance-2-0-fast-260128", False, "720p"): 0.0056,
-    ("dreamina-seedance-2-0-fast-260128", True, "720p"): 0.0033,
-    ("dreamina-seedance-2-0-mini", False, "480p"): 0.0035,
-    ("dreamina-seedance-2-0-mini", True, "480p"): 0.0021,
-    ("dreamina-seedance-2-0-mini", False, "720p"): 0.0035,
-    ("dreamina-seedance-2-0-mini", True, "720p"): 0.0021,
-    ("dreamina-seedance-2-5-260628", False, "480p"): 0.0107,
-    ("dreamina-seedance-2-5-260628", True, "480p"): 0.0064,
-    ("dreamina-seedance-2-5-260628", False, "720p"): 0.0107,
-    ("dreamina-seedance-2-5-260628", True, "720p"): 0.0064,
-}
-
-
-def seedance2_price_per_1k_tokens(model_id: str, has_video_input: bool, resolution: str) -> float | None:
-    return SEEDANCE2_PRICE_PER_1K_TOKENS.get((model_id, has_video_input, resolution))
-
-
 RECOMMENDED_PRESETS = [
     ("1024x1024 (1:1)", 1024, 1024),
     ("864x1152 (3:4)", 864, 1152),
@@ -329,6 +302,7 @@ def seedance2_price_per_1k_tokens(model_id: str, has_video_input: bool, resoluti
     "dreamina-seedance-2-5-260628": {
         "480p": {"min": 409_600, "max": 8_295_044},
         "720p": {"min": 409_600, "max": 8_295_044},
+        "1080p": {"min": 409_600, "max": 8_295_044},
     },
 }
 
@@ -412,3 +386,45 @@ class SeedAudioResponse(BaseModel):
     original_duration: float | None = Field(default=None)
     code: int | None = Field(default=None)
     message: str | None = Field(default=None)
+
+
+class MediaKitVideoEnhanceRequest(BaseModel):
+    video_url: str = Field(...)
+    tool_version: str = Field(...)
+    scene: str | None = Field(None)
+    enhance_style: str | None = Field(None)
+    resolution: str | None = Field(None)
+    resolution_limit: int | None = Field(None)
+    fps: float | None = Field(None)
+    bitrate_level: str = Field(...)
+
+
+class MediaKitError(BaseModel):
+    code: str | None = Field(None)
+    type: str | None = Field(None)
+    message: str | None = Field(None)
+    param: str | None = Field(None)
+
+
+class MediaKitTaskCreateResponse(BaseModel):
+    success: bool = Field(...)
+    task_id: str | None = Field(None)
+    request_id: str | None = Field(None)
+    error: MediaKitError | None = Field(None)
+
+
+class MediaKitTaskResult(BaseModel):
+    video_url: str = Field(...)
+    duration: float | None = Field(None)
+    fps: float | None = Field(None)
+    resolution: str | None = Field(None)
+    tool_version: str | None = Field(None)
+
+
+class MediaKitTaskResponse(BaseModel):
+    success: bool = Field(...)
+    task_id: str | None = Field(None)
+    task_type: str | None = Field(None)
+    status: str | None = Field(None)
+    result: MediaKitTaskResult | None = Field(None)
+    error: MediaKitError | None = Field(None)

comfy_api_nodes/apis/fishaudio.py

@@ -0,0 +1,49 @@
+from pydantic import BaseModel, Field
+
+
+class FishAudioProsody(BaseModel):
+    speed: float = Field(1.0, description="Speaking rate multiplier, 0.5-2.0")
+    volume: float = Field(0.0, description="Volume adjustment in decibels")
+
+
+class FishAudioTTSRequest(BaseModel):
+    text: str = Field(..., description="Text to synthesize")
+    reference_id: str | list[str] | None = Field(None, description="Voice model ID or list of IDs")
+    temperature: float = Field(0.7, description="Expressiveness, 0-1")
+    top_p: float = Field(0.7, description="Nucleus sampling diversity, (0, 1]")
+    prosody: FishAudioProsody = Field(..., description="Speed and volume adjustments")
+    normalize: bool = Field(True, description="Normalize numbers and text for English and Chinese")
+    format: str = Field("wav", description="Output audio format")
+
+
+class FishAudioASRRequest(BaseModel):
+    language: str | None = Field(None, description="Optional ISO 639-1 language hint")
+    ignore_timestamps: bool = Field(True, description="Skip precise timestamp computation")
+
+
+class FishAudioASRSegment(BaseModel):
+    text: str | None = Field(None, description="Segment text")
+    start: float | None = Field(None, description="Segment start time in seconds")
+    end: float | None = Field(None, description="Segment end time in seconds")
+
+
+class FishAudioASRResponse(BaseModel):
+    text: str | None = Field(None, description="Transcribed text")
+    duration: float | None = Field(None, description="Audio duration in seconds")
+    segments: list[FishAudioASRSegment] | None = Field(None, description="Timestamped transcript segments")
+    language_code: str | None = Field(None, description="Detected language as ISO 639-1 code")
+    language: str | None = Field(None, description="Detected language display name")
+
+
+class FishAudioCreateModelRequest(BaseModel):
+    type: str = Field("tts", description="Model type")
+    title: str = Field(..., description="Voice model name")
+    train_mode: str = Field("fast", description="Training mode; fast is instantly available")
+    visibility: str = Field("private", description="Model visibility")
+    enhance_audio_quality: bool = Field(..., description="Enhance reference audio quality")
+
+
+class FishAudioCreateModelResponse(BaseModel):
+    id: str = Field(..., alias="_id", description="Voice model ID for use as reference_id")
+    state: str | None = Field(None, description="Training state")
+    visibility: str | None = Field(None, description="Model visibility")

comfy_api_nodes/apis/meshy.py

@@ -14,6 +14,7 @@ class InputShouldRemesh(TypedDict):
 class InputShouldTexture(TypedDict):
     should_texture: str
     enable_pbr: bool
+    texture_resolution: str
     texture_prompt: str
     texture_image: Input.Image | None
 
@@ -25,7 +26,7 @@ class MeshyTaskResponse(BaseModel):
 class MeshyTextToModelRequest(BaseModel):
     mode: str = Field("preview")
     prompt: str = Field(..., max_length=600)
-    art_style: str = Field(..., description="'realistic' or 'sculpture'")
+    art_style: str = Field(...)
     ai_model: str = Field(...)
     topology: str | None = Field(..., description="'quad' or 'triangle'")
     target_polycount: int | None = Field(..., ge=100, le=300000)
@@ -35,6 +36,7 @@ class MeshyTextToModelRequest(BaseModel):
     )
     symmetry_mode: str = Field(..., description="'auto', 'off' or 'on'")
     pose_mode: str = Field(...)
+    ultra_mode: bool = Field(False)
     seed: int = Field(...)
     moderation: bool = Field(False)
 
@@ -43,6 +45,7 @@ class MeshyRefineTask(BaseModel):
     mode: str = Field("refine")
     preview_task_id: str = Field(...)
     enable_pbr: bool | None = Field(...)
+    texture_resolution: str = Field(...)
     texture_prompt: str | None = Field(...)
     texture_image_url: str | None = Field(...)
     ai_model: str = Field(...)
@@ -61,7 +64,9 @@ class MeshyImageToModelRequest(BaseModel):
     )
     should_texture: bool = Field(...)
     enable_pbr: bool | None = Field(...)
+    texture_resolution: str | None = Field(None)
     pose_mode: str = Field(...)
+    ultra_mode: bool = Field(False)
     texture_prompt: str | None = Field(None, max_length=600)
     texture_image_url: str | None = Field(None)
     seed: int = Field(...)
@@ -80,6 +85,7 @@ class MeshyMultiImageToModelRequest(BaseModel):
     )
     should_texture: bool = Field(...)
     enable_pbr: bool | None = Field(...)
+    texture_resolution: str | None = Field(None)
     pose_mode: str = Field(...)
     texture_prompt: str | None = Field(None, max_length=600)
     texture_image_url: str | None = Field(None)
@@ -103,8 +109,10 @@ class MeshyTextureRequest(BaseModel):
     ai_model: str = Field(...)
     enable_original_uv: bool = Field(...)
     enable_pbr: bool = Field(...)
-    text_style_prompt: str | None = Field(...)
-    image_style_url: str | None = Field(...)
+    texture_resolution: str = Field(...)
+    text_style_prompt: str | None = Field(None)
+    image_style_url: str | None = Field(None)
+    multiview_image_urls: list[str] | None = Field(None)
 
 
 class MeshyModelsUrls(BaseModel):

comfy_api_nodes/apis/wan.py

@@ -184,6 +184,32 @@ class Wan27Text2VideoTaskCreationRequest(BaseModel):
     parameters: Wan27Text2VideoParametersField = Field(...)
 
 
+class Wan3MediaItem(BaseModel):
+    type: str = Field(...)
+    url: str = Field(...)
+
+
+class Wan3InputField(BaseModel):
+    prompt: str | None = Field(None)
+    media: list[Wan3MediaItem] | None = Field(None)
+
+
+class Wan3ParametersField(BaseModel):
+    resolution: str = Field(...)
+    ratio: str = Field(...)
+    duration: int = Field(..., ge=-1, le=30)
+    seed: int = Field(..., ge=0, le=2147483647)
+    audio: bool = Field(True)
+    prompt_extend: bool = Field(True)
+    watermark: bool = Field(False)
+
+
+class Wan3TaskCreationRequest(BaseModel):
+    model: str = Field(...)
+    input: Wan3InputField = Field(...)
+    parameters: Wan3ParametersField = Field(...)
+
+
 class TaskCreationOutputField(BaseModel):
     task_id: str = Field(...)
     task_status: str = Field(...)