nodes/Baverne-comfyUI-TiledWan
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Overwritting 5 Debug outputs

Browse Source
This commit is contained in:
Baverne
2025-08-05 17:10:27 +02:00
parent bcccc542db
commit 6da5985f5c
1 changed files with 17 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
nodes.py
+17 -5
View File
@@ -1448,10 +1448,12 @@ class TiledWanVideoVACEpipe:
},
}
RETURN_TYPES = ("IMAGE", "STRING")
RETURN_NAMES = ("processed_video", "processing_info")
RETURN_TYPES = ("IMAGE", "STRING", "IMAGE", "IMAGE")
RETURN_NAMES = ("processed_video", "processing_info", "first_temporal_chunk", "second_temporal_chunk")
OUTPUT_TOOLTIPS = ("Processed video with WanVideo VACE pipeline applied through tiled processing. Maintains original dimensions and quality.",
"Detailed information about the processing including tile counts, timing, and processing summary.")
"Detailed information about the processing including tile counts, timing, and processing summary.",
"First temporal chunk for debugging - shows the stitched content of the first temporal chunk",
"Second temporal chunk for debugging - shows the stitched content of the second temporal chunk (or first chunk if only one exists)")
FUNCTION = "process_tiled_wanvideo"
OUTPUT_NODE = True
CATEGORY = "TiledWan"
@@ -1562,6 +1564,16 @@ class TiledWanVideoVACEpipe:
print(f"\n🔄 STEP 5: Temporal stitching...")
stitched_video = self._stitch_temporal_chunks_new(temporal_chunks, temporal_tiles, frame_overlap)
# STEP 5.5: Extract debug temporal chunks
print(f"\n🔍 STEP 5.5: Extracting temporal chunks for debugging...")
first_chunk = temporal_chunks[0]['content'] if len(temporal_chunks) > 0 else video
second_chunk = temporal_chunks[1]['content'] if len(temporal_chunks) > 1 else first_chunk
print(f"🔍 Debug chunks extracted:")
print(f" • First chunk shape: {first_chunk.shape}")
print(f" • Second chunk shape: {second_chunk.shape}")
print(f" • Total temporal chunks: {len(temporal_chunks)}")
# STEP 6: Crop to original dimensions (ensure exact input size)
print(f"\n✂️ STEP 6: Cropping to original dimensions...")
print(f"📐 Stitched video shape: {stitched_video.shape}")
@@ -1581,7 +1593,7 @@ class TiledWanVideoVACEpipe:
print(f"✅ Successful tiles: {successful_tiles}/{total_tiles}")
print("="*80 + "\n")
return (final_video, processing_summary)
return (final_video, processing_summary, first_chunk, second_chunk)
except Exception as e:
print(f"❌ Error in tiled WanVideo VACE pipeline: {str(e)}")
@@ -1592,7 +1604,7 @@ class TiledWanVideoVACEpipe:
# Return original video in case of error
error_info = f"Error during tiled WanVideo processing: {str(e)}"
return (video, error_info)
return (video, error_info, video, video)
def _extract_and_process_wanvideo_tiles(self, video, mask, temporal_tiles, spatial_tiles_h, spatial_tiles_w,