VAST-AI TripoSG integration notice Portions of the rectified-flow transformer and shape-VAE decoder architecture, image preprocessing, sampling equations, and hierarchical surface extraction in this family are modified from: VAST-AI-Research/TripoSG https://github.com/VAST-AI-Research/TripoSG revision fc5c40990181e2a756c4e0b1c2f4d6b5202faf8c Those portions are provided under the accompanying MIT License: MIT License Copyright (c) 2025 VAST-AI-Research and contributors. All ported source files carry a prominent modification notice. Independently written diffusers-3d package glue remains under the package Apache-2.0 license. The upstream transformer file states that it is based on Tencent HunyuanDiT. The upstream NOTICE reads: This software contains code derived from HunyuanDiT (https://github.com/Tencent/HunyuanDiT). Notice: Tencent Hunyuan is licensed under the Tencent Hunyuan Community License Agreement, Copyright © 2024 Tencent. All Rights Reserved. The trademark rights of “Tencent Hunyuan” are owned by Tencent or its affiliate. This family implements the TripoSG transformer architecture with Diffusers building blocks (Apache-2.0). Whether that architecture's HunyuanDiT derivation carries Tencent Hunyuan Community License obligations for this code has not been determined. Not included: - the FlashVDM decoder, its attention processor, and flash_extract_geometry, which upstream distributes under the Tencent Hunyuan FlashVDM Community License Agreement (triposg/LICENSE upstream); - diso dual marching cubes (CC BY-NC 4.0); - the briaai/RMBG-1.4 background remover (bria-rmbg-1.4 license). Weights are not redistributed. The VAST-AI/TripoSG model card declares the MIT license; its DINOv2-L image encoder derives from facebook/dinov2-large (Apache-2.0). Converted checkpoints keep their own licenses. No affiliation, sponsorship, or endorsement by VAST-AI-Research, Tripo, Tencent, Meta, or BRIA AI is claimed.