Blokjutsu
Copyright 2026 Bob Carpenter

This product is licensed under the Apache License, Version 2.0 (see LICENSE).

This product is a desktop front end for MOSS-Image, developed by the OpenMOSS Team / MOSI.AI:
  https://github.com/OpenMOSS/MOSS-Image
  MOSS-Image Technical Report, https://arxiv.org/abs/2610.23374
MOSS-Image's pipeline code and model weights are licensed under the Apache License, Version 2.0.

blokjutsu/moss_image.py is MOSS-Image's pipeline code (pipeline_moss_image.py from OpenMOSS-Team/MOSS-Image-SFT),
copied unchanged; Copyright the OpenMOSS Team / MOSI.AI. The model weights are not included; the app downloads
them from Hugging Face when the user asks it to.

The captioning instructions in blokjutsu/describe.py are adapted from MOSS-Image's USERGUIDE.md (Apache License,
Version 2.0); the changes are described in that file.

blokjutsu/layout.py follows MOSS-Image's layout compiler (skills/moss-image-prompt-enhancer/scripts/
compile_prompt.py, Apache License, Version 2.0); the changes are described at the top of that file.

blokjutsu/moss_forward.py is adapted from the MOSS-Image transformer's forward pass (Apache License, Version
2.0); the changes are described at the top of that file.

Models used at run time (downloaded on request, not included):
  MOSS-Image-SFT and MOSS-Image-Base, OpenMOSS Team (Apache-2.0)
  Qwen3.5-9B, Qwen Team, Alibaba Cloud (Apache-2.0): MOSS-Image's text encoder; Blokjutsu also downloads its
    output layer (lm_head) from Qwen/Qwen3.5-9B for Describe image and Enhance prompt
  FLUX.2 VAE, Black Forest Labs (Apache-2.0)

Third-party Python libraries are installed separately and keep their own licenses; see README.md.
