Required Files (v3)
To use the workflows, you need to have the following models installed in ComfyUI.
Place them in the corresponding directories as specified below.
<your_comfyui_dir> / models / checkpoints :
pixart_sigma-FP16.safetensors (1.2GB)
photon_refiner-FP16.safetensors (2.1GB)
<your_comfyui_dir> / models / clip :
t5_xxl_encoder-FP8.safetensors (4.9GB)
<your_comfyui_dir> / models / vae :
pixart_vae.safetensors (0.1GB)
Required Nodes (v3)
IMPORTANT: Ensure that your ComfyUI is updated to the latest version.
Additionally, the workflows require the following custom nodes to be installed:
ComfyUI_ExtraModels: provides support for PixArt-Sigma.
ComfyUI-Crystools: used for some simple string operations.
-- under construction --
Version History
v3
- Models in FP16 & FP8 for min VRAM usage (~6GB).
- Unified PixArt+refiner prompts (no more two separate prompts).
- 1536 x 1024 native resolution.
- Option to select portrait/landscape image orientation.
- Option to choose between fast/quality samplers.
- Option for the refiner to ignore the prompt.
- Variations achieved by changing the refiner seed.
- Six preconfigured workflows.
- CivitAI prompt extractor compatibility.
v2
- Simplified steps configuration.
- Added nodes to easily modify CFG and SEED parameters.
- Aesthetic node color/position changes for better usability.
- Updated install instructions with corrections and clarity.
- Improved final image organization into date-based folders.
v1
- The initial release providing core functionality.Description
Models in FP16 & FP8 for min VRAM usage (~6GB).
Unified PixArt+refiner prompt (no longer requiring two separate prompts).
1536 x 1024 native resolution.
Option to select portrait or landscape image orientation.
Option to choose between two samplers (fast/quality).
Option for the refiner to ignore the prompt and focus on other elements.
Variations achieved by changing the refiner seed.
Six preconfigured workflows (PHOTO, DARKFAN80, PIXEL, INK, 1GIRL, MILO).
CivitAI prompt extractor compatibility.
FAQ
Comments (4)
Very impressive for such a small model! Really gives base SDXL a run for its money. Do you plan to make a Photon finetune of PixArt? The just released SD 3.5 medium also seems to be a good candidate, and it doesn't seem to be a lot more resource intensive than SDXL.
I like the look of this honestly.
But does anyone know why there are so few Loras for pixart? people are not interested in working with it or harder to train successfully?
The main methods for Lora training are the official code and Onetrainer, but ExtraModels has long faced compatibility issues with Onetrainer’s Lora. This might make it difficult for general users to easily utilize the system.I’d like to create my own Lora models, but this compatibility problem is a hurdle.
Since this extension is crucial for Pixart, I plan to test it and report the issue to the developers. I truly wish for an environment that’s easier for everyone to participate in.
I think some update made this workflow unusable.
When I click "run" it makes nothing. Doesnt happen with other workflows