let5see/AI-Video
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

修复视频尺寸传参未生效的问题

Browse Source
This commit is contained in:
puke
2025-11-20 20:09:43 +08:00
parent 04f0754335
commit 7f904f6b19
10 changed files with 123 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
api/routers/resources.py
View File

@@ -76,12 +76,12 @@ async def list_tts_workflows(pixelle_video: PixelleVideoDep):
raise HTTPException(status_code=500, detail=str(e))
@router.get("/workflows/image", response_model=WorkflowListResponse)
async def list_image_workflows(pixelle_video: PixelleVideoDep):
@router.get("/workflows/media", response_model=WorkflowListResponse)
async def list_media_workflows(pixelle_video: PixelleVideoDep):
"""
List available image generation workflows
List available media workflows (both image and video)
Returns list of image workflows from both RunningHub and self-hosted sources.
Returns list of all media workflows from both RunningHub and self-hosted sources.
Example response:
```json
@@ -94,13 +94,41 @@ async def list_image_workflows(pixelle_video: PixelleVideoDep):
"path": "workflows/runninghub/image_flux.json",
"key": "runninghub/image_flux.json",
"workflow_id": "123456"
},
{
"name": "video_wan2.1.json",
"display_name": "video_wan2.1.json - Runninghub",
"source": "runninghub",
"path": "workflows/runninghub/video_wan2.1.json",
"key": "runninghub/video_wan2.1.json",
"workflow_id": "123457"
}
]
}
```
"""
try:
# Get all workflows from media service (image generation is handled by media service)
# Get all workflows from media service (includes both image and video)
all_workflows = pixelle_video.media.list_workflows()
media_workflows = [WorkflowInfo(**wf) for wf in all_workflows]
return WorkflowListResponse(workflows=media_workflows)
except Exception as e:
logger.error(f"List media workflows error: {e}")
raise HTTPException(status_code=500, detail=str(e))
# Keep old endpoint for backward compatibility
@router.get("/workflows/image", response_model=WorkflowListResponse)
async def list_image_workflows(pixelle_video: PixelleVideoDep):
"""
List available image workflows (deprecated, use /workflows/media instead)
This endpoint is kept for backward compatibility but will filter to image_ workflows only.
"""
try:
all_workflows = pixelle_video.media.list_workflows()
# Filter to image workflows only (filename starts with "image_")

32
api/routers/video.py
View File

@@ -63,6 +63,17 @@ async def generate_video_sync(
try:
logger.info(f"Sync video generation: {request_body.text[:50]}...")
# Auto-determine media_width and media_height from template meta tags (required)
if not request_body.frame_template:
raise ValueError("frame_template is required to determine media size")
from pixelle_video.services.frame_html import HTMLFrameGenerator
from pixelle_video.utils.template_util import resolve_template_path
template_path = resolve_template_path(request_body.frame_template)
generator = HTMLFrameGenerator(template_path)
media_width, media_height = generator.get_media_size()
logger.debug(f"Auto-determined media size from template: {media_width}x{media_height}")
# Build video generation parameters
video_params = {
"text": request_body.text,
@@ -73,8 +84,9 @@ async def generate_video_sync(
"max_narration_words": request_body.max_narration_words,
"min_image_prompt_words": request_body.min_image_prompt_words,
"max_image_prompt_words": request_body.max_image_prompt_words,
# Note: image_width and image_height are now auto-determined from template
"image_workflow": request_body.image_workflow,
"media_width": media_width,
"media_height": media_height,
"media_workflow": request_body.media_workflow,
"video_fps": request_body.video_fps,
"frame_template": request_body.frame_template,
"prompt_prefix": request_body.prompt_prefix,
@@ -150,6 +162,17 @@ async def generate_video_async(
# Define async execution function
async def execute_video_generation():
"""Execute video generation in background"""
# Auto-determine media_width and media_height from template meta tags (required)
if not request_body.frame_template:
raise ValueError("frame_template is required to determine media size")
from pixelle_video.services.frame_html import HTMLFrameGenerator
from pixelle_video.utils.template_util import resolve_template_path
template_path = resolve_template_path(request_body.frame_template)
generator = HTMLFrameGenerator(template_path)
media_width, media_height = generator.get_media_size()
logger.debug(f"Auto-determined media size from template: {media_width}x{media_height}")
# Build video generation parameters
video_params = {
"text": request_body.text,
@@ -160,8 +183,9 @@ async def generate_video_async(
"max_narration_words": request_body.max_narration_words,
"min_image_prompt_words": request_body.min_image_prompt_words,
"max_image_prompt_words": request_body.max_image_prompt_words,
# Note: image_width and image_height are now auto-determined from template
"image_workflow": request_body.image_workflow,
"media_width": media_width,
"media_height": media_height,
"media_workflow": request_body.media_workflow,
"video_fps": request_body.video_fps,
"frame_template": request_body.frame_template,
"prompt_prefix": request_body.prompt_prefix,

6
api/schemas/video.py
View File

@@ -56,9 +56,9 @@ class VideoGenerateRequest(BaseModel):
min_image_prompt_words: int = Field(30, ge=10, le=100, description="Min image prompt words")
max_image_prompt_words: int = Field(60, ge=10, le=200, description="Max image prompt words")
# === Image Parameters ===
# Note: image_width and image_height are now auto-determined from template meta tags
image_workflow: Optional[str] = Field(None, description="Custom image workflow")
# === Media Parameters ===
# Note: media_width and media_height are auto-determined from template meta tags
media_workflow: Optional[str] = Field(None, description="Custom media workflow (image or video)")
# === Video Parameters ===
video_fps: int = Field(30, ge=15, le=60, description="Video FPS")

10
pixelle_video/models/storyboard.py
View File

@@ -23,6 +23,10 @@ from typing import List, Optional, Dict, Any
class StoryboardConfig:
"""Storyboard configuration parameters"""
# Required parameters (must come first in dataclass)
media_width: int # Media width (image or video, required)
media_height: int # Media height (image or video, required)
# Task isolation
task_id: Optional[str] = None # Task ID for file isolation (auto-generated if None)
@@ -42,10 +46,8 @@ class StoryboardConfig:
tts_speed: Optional[float] = None # TTS speed multiplier (0.5-2.0, 1.0 = normal)
ref_audio: Optional[str] = None # Reference audio for voice cloning (ComfyUI mode only)
# Image parameters
image_width: int = 1024
image_height: int = 1024
image_workflow: Optional[str] = None # Image workflow filename (None = use default)
# Media workflow
media_workflow: Optional[str] = None # Media workflow filename (image or video, None = use default)
# Frame template (includes size information in path)
frame_template: str = "1080x1920/default.html" # Template path with size (e.g., "1080x1920/default.html")

16
pixelle_video/pipelines/custom.py
View File

@@ -93,8 +93,8 @@ class CustomPipeline(BasePipeline):
tts_speed: float = 1.2,
ref_audio: Optional[str] = None,
image_workflow: Optional[str] = None,
# Note: image_width and image_height are now auto-determined from template
media_workflow: Optional[str] = None,
# Note: media_width and media_height are auto-determined from template
frame_template: Optional[str] = None,
video_fps: int = 30,
@@ -189,8 +189,8 @@ class CustomPipeline(BasePipeline):
# Read media size from template meta tags
template_path = resolve_template_path(frame_template)
generator = HTMLFrameGenerator(template_path)
image_width, image_height = generator.get_media_size()
logger.info(f"📐 Media size from template: {image_width}x{image_height}")
media_width, media_height = generator.get_media_size()
logger.info(f"📐 Media size from template: {media_width}x{media_height}")
if template_type == "image":
logger.info(f"📸 Template requires image generation")
@@ -270,9 +270,9 @@ class CustomPipeline(BasePipeline):
tts_workflow=final_tts_workflow, # Use processed workflow
tts_speed=tts_speed,
ref_audio=ref_audio,
image_width=image_width,
image_height=image_height,
image_workflow=image_workflow,
media_width=media_width,
media_height=media_height,
media_workflow=media_workflow,
frame_template=frame_template
)
@@ -387,7 +387,7 @@ class CustomPipeline(BasePipeline):
"tts_workflow": tts_workflow,
"tts_speed": tts_speed,
"ref_audio": ref_audio,
"image_workflow": image_workflow,
"media_workflow": media_workflow,
"frame_template": frame_template,
"bgm_path": bgm_path,
"bgm_volume": bgm_volume,

28
pixelle_video/pipelines/standard.py
View File

@@ -68,8 +68,10 @@ class StandardPipeline(BasePipeline):
async def __call__(
self,
# === Input ===
# === Input (Required) ===
text: str,
media_width: int, # Required: Media width (from template)
media_height: int, # Required: Media height (from template)
# === Processing Mode ===
mode: Literal["generate", "fixed"] = "generate",
@@ -95,10 +97,8 @@ class StandardPipeline(BasePipeline):
min_image_prompt_words: int = 30,
max_image_prompt_words: int = 60,
# === Image Parameters ===
image_width: int = 1024,
image_height: int = 1024,
image_workflow: Optional[str] = None,
# === Media Workflow ===
media_workflow: Optional[str] = None,
# === Video Parameters ===
video_fps: int = 30,
@@ -155,9 +155,9 @@ class StandardPipeline(BasePipeline):
min_image_prompt_words: Min image prompt length
max_image_prompt_words: Max image prompt length
image_width: Generated image width (default 1024)
image_height: Generated image height (default 1024)
image_workflow: Image workflow filename (e.g., "image_flux.json", None = use default)
media_width: Media width (image or video, required)
media_height: Media height (image or video, required)
media_workflow: Media workflow filename (image or video, e.g., "image_flux.json", "video_wan.json", None = use default)
video_fps: Video frame rate (default 30)
@@ -254,9 +254,9 @@ class StandardPipeline(BasePipeline):
tts_workflow=final_tts_workflow, # Use processed workflow
tts_speed=tts_speed,
ref_audio=ref_audio,
image_width=image_width,
image_height=image_height,
image_workflow=image_workflow,
media_width=media_width,
media_height=media_height,
media_workflow=media_workflow,
frame_template=frame_template or "1080x1920/default.html",
template_params=template_params # Custom template parameters
)
@@ -374,13 +374,13 @@ class StandardPipeline(BasePipeline):
# Enable parallel if either TTS or Image uses RunningHub (most time-consuming parts)
is_runninghub = (
(config.tts_workflow and config.tts_workflow.startswith("runninghub/")) or
(config.image_workflow and config.image_workflow.startswith("runninghub/"))
(config.media_workflow and config.media_workflow.startswith("runninghub/"))
)
if is_runninghub and RUNNING_HUB_PARALLEL_LIMIT > 1:
logger.info(f"🚀 Using parallel processing for RunningHub workflows (max {RUNNING_HUB_PARALLEL_LIMIT} concurrent)")
logger.info(f" TTS: {'runninghub' if config.tts_workflow and config.tts_workflow.startswith('runninghub/') else 'local'}")
logger.info(f" Image: {'runninghub' if config.image_workflow and config.image_workflow.startswith('runninghub/') else 'local'}")
logger.info(f" Media: {'runninghub' if config.media_workflow and config.media_workflow.startswith('runninghub/') else 'local'}")
semaphore = asyncio.Semaphore(RUNNING_HUB_PARALLEL_LIMIT)
completed_count = 0
@@ -541,7 +541,7 @@ class StandardPipeline(BasePipeline):
"tts_workflow": tts_workflow,
"tts_speed": tts_speed,
"ref_audio": ref_audio,
"image_workflow": image_workflow,
"media_workflow": media_workflow,
"prompt_prefix": prompt_prefix,
"frame_template": frame_template,
"template_params": template_params,

8
pixelle_video/services/frame_processor.py
View File

@@ -187,7 +187,7 @@ class FrameProcessor:
# Determine media type based on workflow
# video_ prefix in workflow name indicates video generation
workflow_name = config.image_workflow or ""
workflow_name = config.media_workflow or ""
is_video_workflow = "video_" in workflow_name.lower()
media_type = "video" if is_video_workflow else "image"
@@ -196,10 +196,10 @@ class FrameProcessor:
# Call Media generation (with optional preset)
media_result = await self.core.media(
prompt=frame.image_prompt,
workflow=config.image_workflow, # Pass workflow from config (None = use default)
workflow=config.media_workflow, # Pass workflow from config (None = use default)
media_type=media_type,
width=config.image_width,
height=config.image_height
width=config.media_width,
height=config.media_height
)
# Store media type

12
pixelle_video/services/persistence.py
View File

@@ -380,9 +380,9 @@ class PersistenceService:
"tts_workflow": config.tts_workflow,
"tts_speed": config.tts_speed,
"ref_audio": config.ref_audio,
"image_width": config.image_width,
"image_height": config.image_height,
"image_workflow": config.image_workflow,
"media_width": config.media_width,
"media_height": config.media_height,
"media_workflow": config.media_workflow,
"frame_template": config.frame_template,
"template_params": config.template_params,
}
@@ -402,9 +402,9 @@ class PersistenceService:
tts_workflow=data.get("tts_workflow"),
tts_speed=data.get("tts_speed"),
ref_audio=data.get("ref_audio"),
image_width=data.get("image_width", 1024),
image_height=data.get("image_height", 1024),
image_workflow=data.get("image_workflow"),
media_width=data.get("media_width", data.get("image_width", 1024)), # Backward compatibility
media_height=data.get("media_height", data.get("image_height", 1024)), # Backward compatibility
media_workflow=data.get("media_workflow", data.get("image_workflow")), # Backward compatibility
frame_template=data.get("frame_template", "1080x1920/default.html"),
template_params=data.get("template_params"),
)

12
web/components/output_preview.py
View File

@@ -58,7 +58,7 @@ def render_single_output(pixelle_video, video_params):
frame_template = video_params.get("frame_template")
custom_values_for_video = video_params.get("template_params", {})
workflow_key = video_params.get("image_workflow")
workflow_key = video_params.get("media_workflow")
prompt_prefix = video_params.get("prompt_prefix", "")
with st.container(border=True):
@@ -123,18 +123,20 @@ def render_single_output(pixelle_video, video_params):
progress_bar.progress(min(int(event.progress * 100), 99)) # Cap at 99% until complete
# Generate video (directly pass parameters)
# Note: image_width and image_height are now auto-determined from template
# Note: media_width and media_height are auto-determined from template
video_params = {
"text": text,
"mode": mode,
"title": title if title else None,
"n_scenes": n_scenes,
"image_workflow": workflow_key,
"media_workflow": workflow_key,
"frame_template": frame_template,
"prompt_prefix": prompt_prefix,
"bgm_path": bgm_path,
"bgm_volume": bgm_volume if bgm_path else 0.2,
"progress_callback": update_progress,
"media_width": st.session_state.get('template_media_width'),
"media_height": st.session_state.get('template_media_height'),
}
# Add TTS parameters based on mode
@@ -245,12 +247,14 @@ def render_batch_output(pixelle_video, video_params):
shared_config = {
"title_prefix": video_params.get("title_prefix"),
"n_scenes": video_params.get("n_scenes") or 5,
"image_workflow": video_params.get("image_workflow"),
"media_workflow": video_params.get("media_workflow"),
"frame_template": video_params.get("frame_template"),
"prompt_prefix": video_params.get("prompt_prefix") or "",
"bgm_path": video_params.get("bgm_path"),
"bgm_volume": video_params.get("bgm_volume") or 0.2,
"tts_inference_mode": video_params.get("tts_inference_mode") or "local",
"media_width": video_params.get("media_width"),
"media_height": video_params.get("media_height"),
}
# Add TTS parameters based on mode (only add non-None values)

24
web/components/style_config.py
View File

@@ -610,7 +610,7 @@ def render_style_config(pixelle_video):
workflow_options if workflow_options else ["No workflows found"],
index=default_workflow_index,
label_visibility="collapsed",
key="image_workflow_select"
key="media_workflow_select"
)
# Get the actual workflow key (e.g., "runninghub/image_flux.json")
@@ -621,14 +621,14 @@ def render_style_config(pixelle_video):
workflow_key = "runninghub/image_flux.json" # fallback
# Get media size from template
image_width = st.session_state.get('template_media_width', 1024)
image_height = st.session_state.get('template_media_height', 1024)
media_width = st.session_state.get('template_media_width')
media_height = st.session_state.get('template_media_height')
# Display media size info (read-only)
if template_media_type == "video":
size_info_text = tr('style.video_size_info', width=image_width, height=image_height)
size_info_text = tr('style.video_size_info', width=media_width, height=media_height)
else:
size_info_text = tr('style.image_size_info', width=image_width, height=image_height)
size_info_text = tr('style.image_size_info', width=media_width, height=media_height)
st.info(f"📐 {size_info_text}")
# Prompt prefix input
@@ -679,8 +679,8 @@ def render_style_config(pixelle_video):
prompt=final_prompt,
workflow=workflow_key,
media_type=template_media_type,
width=int(image_width),
height=int(image_height)
width=int(media_width),
height=int(media_height)
))
preview_media_path = media_result.url
@@ -725,8 +725,8 @@ def render_style_config(pixelle_video):
st.caption(tr("image.not_required_hint"))
# Get media size from template (even though not used, for consistency)
image_width = st.session_state.get('template_media_width', 1024)
image_height = st.session_state.get('template_media_height', 1024)
media_width = st.session_state.get('template_media_width')
media_height = st.session_state.get('template_media_height')
# Set default values for later use
workflow_key = None
@@ -741,6 +741,8 @@ def render_style_config(pixelle_video):
"ref_audio": str(ref_audio_path) if ref_audio_path else None,
"frame_template": frame_template,
"template_params": custom_values_for_video if custom_values_for_video else None,
"image_workflow": workflow_key,
"prompt_prefix": prompt_prefix if prompt_prefix else ""
"media_workflow": workflow_key,
"prompt_prefix": prompt_prefix if prompt_prefix else "",
"media_width": media_width,
"media_height": media_height
}