- Prepended ComfyUI model type folder (checkpoints/, clip/, vae/, etc.) to all dest paths
- Removed separate 'type' field from all model entries
- Consolidated SD3.5 duplicate entries (5 → 1)
- Simplified model configuration by embedding directory structure directly in destination paths
This change eliminates the need to parse the 'type' field separately in artifact_huggingface_download.sh,
making the configuration more explicit and easier to understand.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add the following models to models_huggingface.yaml:
- John6666/diving-illustrious-real-asian-v50-sdxl: SDXL fine-tune for photorealistic Asian subjects
- playgroundai/playground-v2.5-1024px-aesthetic: High-aesthetic 1024px SDXL-based model
- Lykon/dreamshaper-8: Versatile SD1.5 fine-tune with multi-style support
All models marked as non-essential and will download .safetensors files.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- CogVideoX-5b: Link 2 shards + index.json instead of single non-existent file
- CogVideoX-5b-I2V: Link 3 shards + index.json instead of single non-existent file - Fixes link command failures for these video generation models
- Shards are now properly symlinked to ComfyUI diffusion_models directory
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added comfyui-workspace-manager plugin to arty.yml for ComfyUI workflow and model management.
Repository: https://github.com/11cafe/comfyui-workspace-manager
Features:
- Workflow management with version history
- Model browser with one-click CivitAI downloads
- Image gallery per workflow
- Auto-save and keyboard shortcuts
Note: Plugin is marked as obsolete (April 2025) as ComfyUI now has built-in workspace features, but added per user request.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Removed acestep-simple-t2m-v1.json as the official Comfy-Org workflows provide better quality:
- acestep-official-t2m-v1.json - Advanced T2M with specialized nodes
- acestep-m2m-editing-v1.json - Music-to-music editing capability
🤖 Generated with Claude Code (https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Removed 3 empty placeholder workflows that only contained metadata:
- acestep-multilang-t2m-v1.json
- acestep-remix-m2m-v1.json
- acestep-chinese-rap-v1.json
Kept only the functional workflow:
- acestep-simple-t2m-v1.json (6 nodes, fully operational)
Users can use the simple workflow and modify the prompt for different use cases:
- Multi-language: prefix lyrics with language tags like [zh], [ja], [ko]
- Remixing: load audio input and adjust denoise strength (0.1-0.7)
- Chinese RAP: use Chinese RAP LoRA with strength 0.8-1.0
🤖 Generated with Claude Code (https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added ACE Step v1 3.5B model for state-of-the-art music generation:
- 15x faster than LLM baselines with superior structural coherence
- Supports 19 languages (en, zh, ja, ko, fr, es, de, it, pt, ru + 9 more)
- Voice cloning, lyric alignment, and multi-genre capabilities
Changes:
- Added ACE Step models to models_huggingface.yaml (checkpoint + Chinese RAP LoRA)
- Added ComfyUI_ACE-Step custom node to arty.yml with installation script
- Created 4 comprehensive workflows in comfyui/workflows/text-to-music/:
* acestep-simple-t2m-v1.json - Basic 60s text-to-music generation
* acestep-multilang-t2m-v1.json - 19-language music generation
* acestep-remix-m2m-v1.json - Music-to-music remixing with style transfer
* acestep-chinese-rap-v1.json - Chinese hip-hop with specialized LoRA
🤖 Generated with Claude Code (https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add vLLM embedding server for BAAI/bge-large-en-v1.5 (port 8002)
- Reorganize supervisor into two logical groups:
- comfyui-services: comfyui, webdav-sync
- vllm-services: vllm-qwen, vllm-llama, vllm-embedding
- Update arty.yml service management scripts for new group structure
- Add individual service control scripts for all vLLM models
Note: Embedding server currently uses placeholder implementation
For production use, switch to sentence-transformers or native vLLM embedding mode
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Removed diffrhythm-random-generation-v1.json as it's no longer needed.
Keeping only the essential DiffRhythm workflows:
- simple text-to-music (95s)
- full-length generation (4m45s)
- reference-based style transfer
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Remove custom PivoineDiffRhythmRun wrapper node
- Add git patch file for ComfyUI_DiffRhythm __init__.py
- Patch adds LlamaConfig fix at import time
- Add arty script 'fix/diffrhythm-patch' to apply patch
- Revert all workflows to use original DiffRhythmRun
- Remove startup_patch.py and revert start.sh
This approach is cleaner and more maintainable than wrapping the node.
The patch directly fixes the tensor dimension mismatch (32 vs 64) in
DiffRhythm's rotary position embeddings by ensuring num_attention_heads
and num_key_value_heads are properly set based on hidden_size.
References:
- https://github.com/billwuhao/ComfyUI_DiffRhythm/issues/44
- https://github.com/billwuhao/ComfyUI_DiffRhythm/issues/48
Previous approach patched DiT.__init__ at runtime, but models were already
instantiated and cached. This version patches LlamaConfig globally BEFORE
any DiffRhythm imports, ensuring all model instances use the correct config.
Key changes:
- Created PatchedLlamaConfig subclass that auto-calculates attention heads
- Replaced LlamaConfig in transformers.models.llama module at import time
- Patch applies to all LlamaConfig instances, including pre-loaded models
This should finally fix the tensor dimension mismatch error.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Adds monkey-patch for DiT.__init__() to properly configure LlamaConfig with
num_attention_heads and num_key_value_heads parameters, which are missing
in the upstream DiffRhythm code.
Root cause: transformers 4.49.0+ requires these parameters but DiffRhythm's
dit.py only specifies hidden_size, causing the library to incorrectly infer
head_dim as 32 instead of 64, leading to tensor dimension mismatches.
Solution:
- Sets num_attention_heads = hidden_size // 64 (standard Llama architecture)
- Sets num_key_value_heads = num_attention_heads // 4 (GQA configuration)
- Ensures head_dim = 64, fixing the "tensor a (32) vs tensor b (64)" error
This is a proper fix rather than just downgrading transformers version.
References:
- https://github.com/billwuhao/ComfyUI_DiffRhythm/issues/44
- https://github.com/billwuhao/ComfyUI_DiffRhythm/issues/48🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Adds 'fix/diffrhythm-transformers' command to quickly downgrade
transformers library to 4.49.0 for DiffRhythm compatibility.
Usage: arty fix/diffrhythm-transformers
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
The correct function to patch is decode_audio from infer_utils module,
which is where chunked VAE decoding actually happens. This intercepts
the call at the right level to force chunked=False.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
The previous approach of overriding diffrhythmgen wasn't working because
ComfyUI doesn't pass the chunked parameter when it's not in INPUT_TYPES.
This fix monkey-patches the infer() function at module level to always
force chunked=False, preventing the tensor dimension mismatch error.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
The parameters must match the diffrhythmgen() function signature order,
not the INPUT_TYPES order. The function has 'edit' as the first parameter.
Correct widgets_values order (11 parameters):
0: edit (boolean)
1: model (string)
2: style_prompt (string)
3: lyrics_or_edit_lyrics (string)
4: edit_segments (string)
5: odeint_method (enum)
6: steps (int)
7: cfg (int)
8: quality_or_speed (enum)
9: unload_model (boolean)
10: seed (int)
Note: style_audio_or_edit_song comes from input connection (not in widgets)
Note: chunked parameter is hidden (not in widgets)
Updated workflows:
- diffrhythm-simple-t2m-v1.json
- diffrhythm-random-generation-v1.json
- diffrhythm-reference-based-v1.json
- diffrhythm-full-length-t2m-v1.json
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
Fix "edit song, edit lyrics, edit segments must be provided" error by adding
the two missing parameters to all three DiffRhythm workflow files:
- diffrhythm-random-generation-v1.json
- diffrhythm-reference-based-v1.json
- diffrhythm-full-length-t2m-v1.json
Added empty string parameters at positions 9 and 10 in widgets_values array:
- edit_song: "" (empty when edit=false)
- edit_lyrics: "" (empty when edit=false)
The DiffRhythmRun node requires 12 parameters total, not 10. These workflows
use edit=false (no editing), so the edit parameters should be empty strings.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add FFmpeg and its development libraries to setup/system-packages script:
- ffmpeg: Main FFmpeg executable
- libavcodec-dev: Audio/video codec library
- libavformat-dev: Audio/video format library
- libavutil-dev: Utility library for FFmpeg
- libswscale-dev: Video scaling library
These libraries are required for torchcodec to function properly with
DiffRhythm audio generation. Also added FFmpeg version verification
after installation.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add torchcodec to ComfyUI requirements.txt to fix audio tensor caching
error in DiffRhythm. This package is required for save_with_torchcodec
function used by DiffRhythm audio nodes.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add arty script to download eval.yaml and eval.safetensors files from
HuggingFace space for DiffRhythm node support. These files are required
for DiffRhythm evaluation model functionality.
- Add models/diffrhythm-eval script to download eval-model files
- Update setup/comfyui-nodes to create eval-model directory
- Files downloaded from ASLP-lab/DiffRhythm HuggingFace space
- Script includes file verification and size reporting
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add DiffRhythm dependencies to requirements.txt (19 packages)
- Add reference audio placeholder for style transfer workflow
- DiffRhythm nodes now loading in ComfyUI
- All four workflows ready for music generation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added all required packages for ComfyUI_DiffRhythm extension:
- torchdiffeq: ODE solvers for diffusion models
- x-transformers: Transformer architecture components
- librosa: Audio analysis and feature extraction
- pandas, pyarrow: Data handling
- ema-pytorch, prefigure: Training utilities
- muq: Music quality model
- mutagen: Audio metadata handling
- pykakasi, jieba, cn2an, pypinyin: Chinese/Japanese text processing
- Unidecode, phonemizer, inflect: Text normalization and phonetic conversion
- py3langid: Language identification
These dependencies enable the DiffRhythm node to load and function properly in ComfyUI, fixing the "ModuleNotFoundError: No module named 'infer_utils'" error.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Use Gitea container registry instead of Docker Hub
- Update workflow to use gitea.actor and REGISTRY_TOKEN
- Update documentation to reflect correct registry URL
- Match supervisor-ui workflow configuration
- Add Dockerfile with minimal setup (supervisor, tailscale)
- Add start.sh bootstrap script for container initialization
- Add Gitea workflow for automated Docker image builds
- Add comprehensive RUNPOD_TEMPLATE.md documentation
- Add bootstrap-venvs.sh for Python venv health checks
This enables deployment of the AI orchestrator on RunPod using:
- Minimal Docker image (~2-3GB) for fast deployment
- Network volume for models and data persistence (~80-200GB)
- Automated builds on push to main or version tags
- Full Tailscale VPN integration
- Supervisor process management
- Changed checkpoint from waiIllustriousSDXL_v150.safetensors to ponyDiffusionV6XL_v6StartWithThisOne.safetensors
- Fixed metadata model reference (was incorrectly referencing LoRA)
- Added files field to models_civitai.yaml for explicit filename mapping
- Aligns workflow with actual Pony Diffusion V6 XL model
- Add files field to badx-sdxl, pony-pdxl-hq-v3, pony-pdxl-xxx
- Specifies actual downloaded filenames (BadX-neg.pt, zPDXL3.safetensors, zPDXLxxx.pt)
- Allows script to properly link embeddings where YAML name != filename
Changed checkpoint from 'add-detail-xl.safetensors' (which is a LoRA) to
'waiIllustriousSDXL_v150.safetensors' which is the downloaded anime NSFW model
Updated arty.yml workflow linking script to include NSFW workflows:
- Added nsfw_ prefix for NSFW workflow category
- Links 4 NSFW workflows (LUSTIFY, Pony, RealVisXL, Ultimate Upscale)
- Updated workflow count from 20 to 25 total production workflows
- Updated documentation to list all 7 workflow categories
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
