COPYRIGHT AND ATTRIBUTION NOTES FOR diffuseR

diffuseR is licensed under the Apache License (>= 2). Portions of the
package are R ports of, or derived from, third-party work:

1. HuggingFace diffusers (Apache License 2.0)
   Copyright The HuggingFace Team.
   https://github.com/huggingface/diffusers
   The LTX video pipeline components (transformer, video/audio VAE,
   text-embedding connectors, vocoder, latent upsampler, schedulers, and
   pipeline logic including the AdaIN latent filter and tone mapping) are
   R ports written from the diffusers reference implementation.
   The FlowMatch Euler scheduler is ported from
   diffusers/schedulers/scheduling_flow_match_euler_discrete.py.

2. HuggingFace transformers (Apache License 2.0)
   Copyright The HuggingFace Team.
   https://github.com/huggingface/transformers
   The Gemma3 text encoder (R/gemma3_text_encoder.R) is an R port of the
   Gemma3 implementation in transformers.

3. Lightricks Ltd.
   https://github.com/Lightricks/LTX-2
   Numeric constants (e.g. distilled sigma schedules), the single-file
   checkpoint key layout, and pipeline parameter defaults are facts taken
   from the official LTX release. No code from the Lightricks repository
   (LTX-2 Community License) is included in this package.

   LTX model weights are downloaded by the user directly from Lightricks'
   HuggingFace repositories under the LTX-2 Community License and are not
   part of, nor redistributed with, this package.

4. Stability AI / Stable Diffusion (SD 2.1, SDXL)
   Model weights are downloaded by the user under their respective
   licenses (CreativeML Open RAIL++-M and similar) and are not part of,
   nor redistributed with, this package.

See inst/REFERENCES.md for a technique-by-technique account of where
each implementation idea comes from.
