TRELLIS

TRELLIS is Microsoft's open image-to-3D model. It builds the object as a sparse voxel structure first, then 32 Gaussians per occupied voxel. Upstream needs CUDA-only extensions; LibreYOLO runs the same weights in plain PyTorch through the LibreGen factory.

Task
3d generation
Dimensioni
l at 518 px
Installa
pip install "libreyolo[gen]"
Livello di supporto
Livello parallelo, dalla v1.8.0. Un'interfaccia di prodotto separata, con una propria factory e un proprio contratto.
Origine
TRELLIS di Microsoft Research, MIT. Articolo, sorgente
Licenze
Codice MIT, pesi MIT. Uso commerciale

Install

bash
pip install "libreyolo[gen]"

The extra brings transformers, which loads the DINOv2 image encoder.

Predict

Nothing is gated and no login is needed. Weights download on first use at pinned revisions.

Python
from libreyolo import LibreGen, SAMPLE_IMAGE model = LibreGen("trellis")         # weights download on first useresult = model(SAMPLE_IMAGE, seed=0) result.save("object.ply")           # standard 3DGS layoutresult.save("object.splat")         # compact web formatprint(len(result.gaussians))        # follows the object, not a knob
CLI
# Saves the object as .ply under runs/generate/predict.libreyolo generate source=https://raw.githubusercontent.com/LibreYOLO/libreyolo/release/libreyolo/assets/guggenheim-bilbao.jpg model=trellis seed=0
Settings
from libreyolo import LibreGen, SAMPLE_IMAGE model = LibreGen("trellis")result = model(SAMPLE_IMAGE, steps=12)          # faster samplingresult = model(SAMPLE_IMAGE, guidance=3.0, margin=1.3)

A dense flow transformer denoises the object's occupancy at 16^3, a 3D CNN decodes it to the active voxels of a 64^3 grid, a sparse flow transformer denoises a structured latent on those voxels, and a sparse transformer turns each voxel into 32 Gaussians. Both flow stages read DINOv2 tokens of the 518 px input. The number of Gaussians therefore follows the object (often 200,000 to 600,000).

Upstream's sparse convolutions (spconv) and attention (xformers or flash-attn) run on CUDA only. LibreYOLO replaces them with PyTorch operations and keeps the parameter names, so the released weights load unchanged on NVIDIA GPUs, Apple silicon and CPU. Noise is drawn on the CPU as upstream does, so a seed gives the same object on every device.

The output is result.gaussians, the same payload as Speridlabs Idol: xyz, scales, rotations, opacities, sh, colors, seed and params, z-up in [-0.5, 0.5]^3. Background removal runs on LibreYOLO's BiRefNet port.

Settings

SettingDefaultMeaning
steps25Sampling steps, for both flow stages.
guidance5.0Classifier-free guidance, for both flow stages.
shift3.0Flow time shift.
margin1.2Crop side over object size, as upstream.

Variants

One size, l (TRELLIS-image-large): 1.3B parameters in four networks, about 2.7 GB, plus 1.2 GB for the DINOv2 ViT-L/14 encoder.

Export

TaskONNXTorchScriptExecuTorchTensorRTOpenVINOPaddleMNNRKNNAxelerancnnTFLiteCoreMLCore AI
3D generation3D generation to ONNX: non supportato3D generation to TorchScript: non supportato3D generation to ExecuTorch: non supportato3D generation to TensorRT: non supportato3D generation to OpenVINO: non supportato3D generation to Paddle: non supportato3D generation to MNN: non supportato3D generation to RKNN: non supportato3D generation to Axelera: non supportato3D generation to ncnn: non supportato3D generation to TFLite: non supportato3D generation to CoreML: non supportato3D generation to Core AI: non supportato

Export is not implemented for LibreGen 3D families. export(), train() and val() raise NotImplementedError (ADR 0031 in the library).

Licensing

Controlla la licenza nel repository Hugging Face degli specifici pesi che scarichi. Ogni checkpoint nell<link>organizzazione LibreYOLO</link> ne include una e non è sempre la stessa per tutta la famiglia. Quel repository è la fonte autorevole; il riepilogo seguente descrive le condizioni applicabili al momento dellultima verifica di questa pagina.

Questa è una descrizione delle licenze coinvolte, non una consulenza legale. Se la risposta è importante a fini commerciali, leggi personalmente le licenze e chiedi una consulenza indipendente.

Lavoro originale
TRELLIS, Microsoft Research
Licenza upstream
MIT
Codice LibreYOLO
MIT
Pesi
MIT, ripubblicati su huggingface.co/LibreYOLO
Interpretazione
MIT is a permissive license, so the code and weights can be used in commercial and closed-source products; the only obligation is to keep the copyright and license notice with copies. LibreYOLO's adapter is MIT, and the flow models, decoders, sparse modules and sampler it adapts from TRELLIS keep MIT, noted in libreyolo/models/gen/trellis/NOTICE. LibreYOLO replaces upstream's CUDA-only spconv, xformers and flash-attn calls with PyTorch, and does not use TRELLIS's Gaussian holder, which carries Inria's non-commercial Gaussian Splatting license. The microsoft/TRELLIS-image-large weights (MIT) are redistributed unchanged as LibreYOLO/LibreTRELLISl together with the BiRefNet weights (MIT); the DINOv2 image encoder, facebook/dinov2-with-registers-large (Apache-2.0), downloads from Meta. All are pinned and none is gated.

TRELLIS is by Microsoft Research (microsoft/TRELLIS). Code and the microsoft/TRELLIS-image-large weights are MIT, so commercial use is fine with the notice kept. LibreYOLO redistributes the weights unchanged, with the BiRefNet weights (MIT), as LibreYOLO/LibreTRELLISl. The DINOv2 encoder, facebook/dinov2-with-registers-large, is Apache-2.0 and downloads from Meta. Upstream's Gaussian holder, under Inria's non-commercial Gaussian Splatting license, is not used.

Citation

@article{xiang2024structured,
    title   = {Structured 3D Latents for Scalable and Versatile 3D Generation},
    author  = {Xiang, Jianfeng and Lv, Zelong and Xu, Sicheng and Deng, Yu and Wang, Ruicheng and Zhang, Bowen and Chen, Dong and Tong, Xin and Yang, Jiaolong},
    journal = {arXiv preprint arXiv:2412.01506},
    year    = {2024}
}

Copiato dal blocco di citazione degli autori disponibile su github.com/microsoft/TRELLIS#-citation.

Verificato con LibreYOLO v1.8.0. Le tabelle di supporto, i checkpoint e i dati dei benchmark in questa pagina vengono generati dalla libreria rilasciata e dai pesi pubblicati, non scritti a mano.