diff --git a/pixelle_video/pipelines/asset_based.py b/pixelle_video/pipelines/asset_based.py
index 413beed..5c85cac 100644
--- a/pixelle_video/pipelines/asset_based.py
+++ b/pixelle_video/pipelines/asset_based.py
@@ -472,7 +472,9 @@ Generate the video script now:"""
context.narrations = all_narrations
# Get template dimensions
- template_name = "1080x1920/image_pure.html"
+ # Use asset_default.html template which supports both image and video assets
+ # (conditionally shows background image or provides transparent overlay)
+ template_name = "1080x1920/asset_default.html"
# Extract dimensions from template name (e.g., "1080x1920")
try:
dims = template_name.split("/")[0].split("x")
@@ -524,9 +526,20 @@ Generate the video script now:"""
created_at=datetime.now()
)
- # Store matched asset path in the frame
- frame.image_path = scene["matched_asset"]
- frame.media_type = "image"
+ # Get asset path and determine actual media type from asset_index
+ asset_path = scene["matched_asset"]
+ asset_metadata = self.asset_index.get(asset_path, {})
+ asset_type = asset_metadata.get("type", "image") # Default to image if not found
+
+ # Set media type and path based on actual asset type
+ if asset_type == "video":
+ frame.media_type = "video"
+ frame.video_path = asset_path
+ logger.debug(f"Scene {i}: Using video asset: {Path(asset_path).name}")
+ else:
+ frame.media_type = "image"
+ frame.image_path = asset_path
+ logger.debug(f"Scene {i}: Using image asset: {Path(asset_path).name}")
# Store scene info for later audio generation
frame._scene_data = scene # Temporary storage for multi-narration
diff --git a/pixelle_video/services/frame_processor.py b/pixelle_video/services/frame_processor.py
index 30d16e0..7ec5507 100644
--- a/pixelle_video/services/frame_processor.py
+++ b/pixelle_video/services/frame_processor.py
@@ -73,8 +73,8 @@ class FrameProcessor:
frame_num = frame.index + 1
# Determine if this frame needs image generation
- # If image_path is already set (e.g. asset-based pipeline), we consider it "needs image" but skip generation
- has_existing_image = frame.image_path is not None
+ # If image_path or video_path is already set (e.g. asset-based pipeline), we consider it "has existing media" but skip generation
+ has_existing_media = frame.image_path is not None or frame.video_path is not None
needs_generation = frame.image_prompt is not None
try:
@@ -93,7 +93,6 @@ class FrameProcessor:
else:
logger.debug(f" 1/4: Using existing audio: {frame.audio_path}")
- # Step 2: Generate media (image or video, conditional)
# Step 2: Generate media (image or video, conditional)
if needs_generation:
if progress_callback:
@@ -106,8 +105,12 @@ class FrameProcessor:
action="media"
))
await self._step_generate_media(frame, config)
- elif has_existing_image:
- logger.debug(f" 2/4: Using existing image: {frame.image_path}")
+ elif has_existing_media:
+ # Log appropriate message based on media type
+ if frame.video_path:
+ logger.debug(f" 2/4: Using existing video: {frame.video_path}")
+ else:
+ logger.debug(f" 2/4: Using existing image: {frame.image_path}")
else:
frame.image_path = None
frame.media_type = None
@@ -117,7 +120,7 @@ class FrameProcessor:
if progress_callback:
progress_callback(ProgressEvent(
event_type="frame_step",
- progress=0.50 if (needs_generation or has_existing_image) else 0.33,
+ progress=0.50 if (needs_generation or has_existing_media) else 0.33,
frame_current=frame_num,
frame_total=total_frames,
step=3,
@@ -129,7 +132,7 @@ class FrameProcessor:
if progress_callback:
progress_callback(ProgressEvent(
event_type="frame_step",
- progress=0.75 if (needs_generation or has_existing_image) else 0.67,
+ progress=0.75 if (needs_generation or has_existing_media) else 0.67,
frame_current=frame_num,
frame_total=total_frames,
step=4,
@@ -313,12 +316,14 @@ class FrameProcessor:
# Generate frame using HTML (size is auto-parsed from template path)
generator = HTMLFrameGenerator(template_path)
- logger.debug(f"Generating frame with image: '{frame.image_path}' (type: {type(frame.image_path)})")
+ # Use video_path for video media, image_path for images
+ media_path = frame.video_path if frame.media_type == "video" else frame.image_path
+ logger.debug(f"Generating frame with media: '{media_path}' (type: {frame.media_type})")
composed_path = await generator.generate_frame(
title=storyboard.title,
text=frame.narration,
- image=frame.image_path,
+ image=media_path, # HTMLFrameGenerator handles both image and video paths
ext=ext,
output_path=output_path
)
@@ -372,7 +377,8 @@ class FrameProcessor:
os.unlink(temp_video_with_overlay)
elif frame.media_type == "image" or frame.media_type is None:
- # Image workflow: create video from image + audio
+ # Image workflow: Use composed image directly
+ # The asset_default.html template includes the image in the composition
logger.debug(f" → Using image-based composition")
segment_path = video_service.create_video_from_image(
diff --git a/pixelle_video/services/video_analysis.py b/pixelle_video/services/video_analysis.py
index 85ce5f0..56a41c5 100644
--- a/pixelle_video/services/video_analysis.py
+++ b/pixelle_video/services/video_analysis.py
@@ -32,9 +32,9 @@ class VideoAnalysisService(ComfyBaseService):
Uses ComfyKit to execute video understanding workflows.
Returns detailed textual descriptions of video content.
- Convention: workflows follow {source}/video_understanding.json pattern
- - runninghub/video_understanding.json (default, cloud-based)
- - selfhost/video_understanding.json (local ComfyUI, future)
+ Convention: workflows follow {source}/analyse_video.json pattern
+ - runninghub/analyse_video.json (default, cloud-based)
+ - selfhost/analyse_video.json (local ComfyUI, future)
Usage:
# Use default (runninghub cloud)
@@ -50,7 +50,7 @@ class VideoAnalysisService(ComfyBaseService):
workflows = pixelle_video.video_analysis.list_workflows()
"""
- WORKFLOW_PREFIX = "video_understanding"
+ WORKFLOW_PREFIX = "analyse_video"
WORKFLOWS_DIR = "workflows"
def __init__(self, config: dict, core=None):
@@ -114,8 +114,8 @@ class VideoAnalysisService(ComfyBaseService):
# 2. Resolve workflow path using convention
if workflow is None:
- # Use standardized naming: {source}/video_understanding.json
- workflow = resolve_workflow_path("video_understanding", source)
+ # Use standardized naming: {source}/analyse_video.json
+ workflow = resolve_workflow_path("analyse_video", source)
logger.info(f"Using {source} workflow: {workflow}")
# 3. Resolve workflow (returns structured info)
diff --git a/templates/1080x1920/image_pure.html b/templates/1080x1920/asset_default.html
similarity index 62%
rename from templates/1080x1920/image_pure.html
rename to templates/1080x1920/asset_default.html
index 880d42f..f38c226 100644
--- a/templates/1080x1920/image_pure.html
+++ b/templates/1080x1920/asset_default.html
@@ -1,5 +1,6 @@
+
@@ -9,17 +10,16 @@
margin: 0;
padding: 0;
}
-
+
body {
margin: 0;
padding: 0;
width: 1080px;
height: 1920px;
font-family: 'PingFang SC', 'Source Han Sans', 'Microsoft YaHei', sans-serif;
- background: #000;
overflow: hidden;
}
-
+
.page-container {
width: 1080px;
height: 1920px;
@@ -27,7 +27,10 @@
overflow: hidden;
}
- /* 1. Background Image Layer (垫底图片) */
+ /* 1. Background Media Layer (背景媒体层)
+ - For image assets: displays the image
+ - For video assets: hidden (video is composited in later step)
+ */
.background-layer {
position: absolute;
top: 0;
@@ -44,10 +47,15 @@
display: block;
}
+ /* Hide background layer when no image (video mode) */
+ .background-layer:empty {
+ display: none;
+ }
+
/* 2. Gradient Overlay (渐变遮罩)
- Ensures text readability regardless of image brightness
+ Ensures text readability regardless of background brightness
Top: Darker for Title
- Middle: Transparent for Image visibility
+ Middle: Transparent for Media visibility
Bottom: Darker for Subtitles
*/
.gradient-overlay {
@@ -57,13 +65,11 @@
width: 100%;
height: 100%;
z-index: 1;
- background: linear-gradient(
- to bottom,
- rgba(0,0,0,0.6) 0%,
- rgba(0,0,0,0.1) 25%,
- rgba(0,0,0,0.1) 60%,
- rgba(0,0,0,0.8) 100%
- );
+ background: linear-gradient(to bottom,
+ rgba(0, 0, 0, 0.6) 0%,
+ rgba(0, 0, 0, 0.1) 25%,
+ rgba(0, 0, 0, 0.1) 60%,
+ rgba(0, 0, 0, 0.8) 100%);
}
/* 3. Content Layer (内容层) */
@@ -72,7 +78,8 @@
z-index: 2;
width: 100%;
height: 100%;
- padding: 120px 80px 0px 80px; /* Top, Right, Bottom, Left */
+ padding: 120px 80px 0px 80px;
+ /* Top, Right, Bottom, Left */
box-sizing: border-box;
display: flex;
flex-direction: column;
@@ -85,7 +92,7 @@
font-size: 80px;
font-weight: 700;
line-height: 1.2;
- text-shadow: 0 4px 12px rgba(0,0,0,0.5);
+ text-shadow: 0 4px 12px rgba(0, 0, 0, 0.5);
margin-bottom: 40px;
text-align: center;
}
@@ -110,16 +117,20 @@
font-weight: 500;
line-height: 1.6;
text-align: center;
- text-shadow: 0 2px 8px rgba(0,0,0,0.6);
+ text-shadow: 0 2px 8px rgba(0, 0, 0, 0.6);
backdrop-filter: blur(4px);
}
+
-
-
-
+
+
+
@@ -141,5 +152,23 @@