See-through: Marigold Depth
Copyright 2026 The See-through Authors

This model is part of See-through (https://github.com/shitagaki-lab/see-through).
The See-through code is licensed under the Apache License, Version 2.0.
The licence terms for these model weights are in LICENSE.

==============================================================================
Third-party models
==============================================================================

This model is derived from, or includes components of, the following:

* Marigold Depth v1.1 by PRS, ETH Zürich. Copyright (c) 2023-2025 PRS, ETH Zürich.
  https://huggingface.co/prs-eth/marigold-depth-v1-1
  Open RAIL++-M License
  The UNet was fine-tuned from this model. The VAE, text encoder and tokenizer
  are taken from it.

* Stable Diffusion 2 by Stability AI. Copyright (c) 2022 Stability AI and contributors.
  CreativeML Open RAIL++-M License
  Base model of Marigold. The use-based restrictions in its Attachment A apply
  to this model.

==============================================================================
Changes made by The See-through Authors
==============================================================================

* The UNet was fine-tuned from Marigold Depth v1.1 for pseudo-depth estimation
  of anime characters, and converted into the multi-frame UNet
  (UNetFrameConditionModel) used by the See-through pipeline.
* All components were saved in the diffusers format.

==============================================================================
Citation
==============================================================================

Jian Lin, Chengze Li, Haoyun Qin, Kwun Wang Chan, Yanghua Jin, Hanyuan Liu,
Stephen Chun Wang Choy, and Xueting Liu. 2026. See-through: Single-image Layer
Decomposition for Anime Characters. In SIGGRAPH Conference Papers '26. ACM.
https://doi.org/10.1145/3799902.3811209