comfyui-huyl2-nodes
Custom utility nodes for ComfyUI by huyl2 (sort list v2, etc.)
Nodes (63)
Adaptive Resolution π
Audio File Scanner π΅
Wipe a directory clean from inside your graph
A scoped version of this pack's blunt rm -r *
A manual flush for heavy video pipelines
Stop Google from killing your session mid-render
Collect Files
A mask preview that never breaks your graph
Point it at a folder, get a sorted file list back
Turn a pasted list of paths into something loopable
Fix Dimensions (Div by 16)
Folder File Scanner π
Stamp your output filenames without leaving the graph
Stitch two images without picking left/right yourself
Image to SHA256
Read images straight off disk, no duplicate upload
_Loop Add (internal)
_Loop Less Than (internal)
A hardcoded VRAM budget, not a universal formula
Width and height instead of a single number
The missing shutil.move node for ComfyUI
String Embed In Image (RGB)
String Extract From Image (RGB)
MP3 Path β Audio
RAM-Cleanup
Strip a header row out of a text blob
Rename File
One fixed step-down, not a percentage scaler
Set Value For Dancing Mask
Set Value For Dancing None Mask
Set Value For MC
Set Value For MC I2V
A settings panel for someone else's image-to-video Colab
The license-gated settings node in huyl3-cpu's pack
Same settings panel, license_key dropped, resolution finally documented
A labeled config block for a batch anime-conversion job
Set Value With Path
Closing the loop you opened with Start
Giving ComfyUI's graph an actual loop
Where the loop actually decides whether to keep going
A real loop primitive, hiding in an otherwise very personal node pack
The node that gives this whole pack its name
Bare-bones, hardcoded 7-minute chunks, no report on what it did
Configurable chunk length, a real list output, and it leaves short files alone
Embed String in Alpha β hide text in a PNG's alpha channel
Extract String From Alpha β read back what steg_alpha_embed hid
Extract String From Colors (RGB) β no matching embed node in this pack
The display name and the actual class name disagree
Ten fixed string slots instead of a fiddly dynamic input widget
π Tunnel Auto Reconnect
Set Value (VF9)
Bridging Video Helper Suite into a plain string
A VHS_-prefixed node that has nothing to do with Video Helper Suite
Find your bottleneck clip before a batch operation
Not concatenation β it's muxing an audio track onto video
Video Cut to 8s Segments
Joins a folder of clips into one file, gated behind a license_key
Same job, louder name, no license key
Strip the soundtrack before you touch it
Cuts on visual changes, not fixed intervals β looks like a PySceneDetect wrapper
Stitching two clips without them drifting apart
The math behind chunking a long WanAnimate clip
Cropping off the padding Wan's 4n+1 rule leaves behind
ComfyUI-SortList
Custom nodes for ComfyUI with video processing utilities.
Installation
Google Colab
cd /content/ComfyUI/custom_nodes
git clone https://github.com/huyl3-cpu/comfyui-sortlist.git
Update to Latest Version
On Google Colab
Method 1: Quick Update (Recommended)
cd /content/ComfyUI/custom_nodes/comfyui-sortlist
git pull origin main
Method 2: Using Update Script
cd /content/ComfyUI/custom_nodes/comfyui-sortlist
bash update_colab.sh
After updating:
- Restart Runtime:
Runtime β Restart Runtime - Re-run ComfyUI startup cells
- Nodes will be updated!
Nodes
Video Cut to 8s Segments
Cuts large videos into smaller segments with customizable duration.
Features:
- β Auto FPS detection
- β Frame-accurate cutting
- β Guaranteed output: H.264, MP4, YUV420P
- β Auto-fallback: GPU β CPU encoding
- β Parallel processing
Parameters:
video_url: Path to input videosegment_duration: Segment length in seconds (default: 8)output_prefix: Output filename prefix (default: "a")use_gpu: Try GPU encoding first (default: true, auto-fallback to CPU)accurate_cut: Accurate seeking (default: true)parallel_workers: Number of parallel workers (default: 4)
Outputs:
videos_list: List of output video pathsoutput_directory: Path to output directory (videos_cut/)
Example:
Input: video.mp4 (268.2s, 25 FPS)
segment_duration: 8s
Output: 34 segments (33 Γ 8.0s + 1 Γ 4.2s)
Recent Updates
v1.1 (2026-01-26)
- β Added auto-fallback from GPU to CPU encoding
- β Enhanced error logging for better debugging
- β Fixed NVENC compatibility on Google Colab
- β Improved terminal output compatibility
- β Added diagnostic tools
License
MIT