DeathAngel00/Photogen-checkpoints

Photogen Models

All model weights for Photogen — a StoryMaker-based character-consistent image generation pipeline.

Files

StoryMaker Adapter (mask.bin)

The main character adapter weights, originally from RED-AIGC/StoryMaker.

• File: mask.bin (3.32 GB)
• Usage: Load via pipe.load_character_adapter()

InsightFace Face Encoder (buffalo_l/)

Face detection and recognition models for identity preservation.

File	Size	Description
det_10g.onnx	16.9 MB	Face detection model
w600k_r50.onnx	174 MB	Face recognition model
2d106det.onnx	5.0 MB	2D landmark detection
1k3d68.onnx	144 MB	3D face landmarks
genderage.onnx	1.3 MB	Age and gender estimation

Extract into models/buffalo_l/ for use with InsightFace.

Trained Checkpoints

Custom character adapter weights trained on custom portrait data.

Checkpoint	Size	Description
checkpoint-50/mask.bin	641 MB	Early training (50 steps)
checkpoint-100/mask.bin	641 MB	Medium training (100 steps)
checkpoint-150/mask.bin	641 MB	Recommended (150 steps)
checkpoint-200/mask.bin	641 MB	Extended training (200 steps)

Download

# Install huggingface-cli if needed
pip install huggingface-hub

# Download everything
huggingface-cli download DeathAngel00/Photogen-checkpoints --local-dir . --local-dir-use-symlinks False

# Or download individual files
huggingface-cli download DeathAngel00/Photogen-checkpoints mask.bin --local-dir checkpoints
huggingface-cli download DeathAngel00/Photogen-checkpoints checkpoint-150/mask.bin --local-dir output_sd15