firtoz/trellis 🔢🖼️✓ → ❓

▶️ 438.3K runs 📅 Dec 2024 ⚙️ Cog 0.9.8 🔗 GitHub 📄 Paper ⚖️ License

3d-gaussians game-asset-generation image-to-3d

About

A powerful 3D asset generation model

Example Output

Output

{"model_file":"https://replicate.delivery/yhqm/5xOmxKPXDTpnIdxRRvs91WKWHTYNGmdBjuE7DbBEigZf0WCKA/output.glb","color_video":"https://replicate.delivery/yhqm/wJejQmfsfyNf3T4soquMhYLR34F07LESybQQsy8iYmy2n2SQB/output_color.mp4","gaussian_ply":"https://replicate.delivery/yhqm/tjyjYewq6fj2xUkco7MnifmmL2CeHG3Wfl7Zw5vdxwu0PtlgC/output_gaussian.ply","normal_video":null,"combined_video":null,"no_background_images":null}

Performance Metrics

36.77s Prediction Time

36.78s Total Time

All Input Parameters

{
  "seed": 0,
  "images": [
    "https://replicate.delivery/pbxt/MJaYRxQMgIzPsALScNadsZFCXR2h1n97xBzhRinmUQw9aw25/ephemeros_a_dune_sandworm_with_black_background_de398ce7-2276-4634-8f1d-c4ed2423cda4.png"
  ],
  "texture_size": 2048,
  "mesh_simplify": 0.9,
  "generate_color": true,
  "generate_model": true,
  "randomize_seed": true,
  "generate_normal": false,
  "save_gaussian_ply": true,
  "ss_sampling_steps": 38,
  "slat_sampling_steps": 12,
  "return_no_background": false,
  "ss_guidance_strength": 7.5,
  "slat_guidance_strength": 3
}

Input Parameters

seed Type: integerDefault: 0: Random seed for generation
images (required) Type: array: List of input images to generate 3D asset from
texture_size Type: integerDefault: 1024Range: 512 - 2048: GLB Extraction - Texture Size (only used if generate_model=True)
mesh_simplify Type: numberDefault: 0.95Range: 0.9 - 0.98: GLB Extraction - Mesh Simplification (only used if generate_model=True)
generate_color Type: booleanDefault: true: Generate color video render
generate_model Type: booleanDefault: false: Generate 3D model file (GLB)
randomize_seed Type: booleanDefault: true: Randomize seed
generate_normal Type: booleanDefault: false: Generate normal video render
save_gaussian_ply Type: booleanDefault: false: Save Gaussian point cloud as PLY file
ss_sampling_steps Type: integerDefault: 12Range: 1 - 50: Stage 1: Sparse Structure Generation - Sampling Steps
slat_sampling_steps Type: integerDefault: 12Range: 1 - 50: Stage 2: Structured Latent Generation - Sampling Steps
return_no_background Type: booleanDefault: false: Return the preprocessed images without background
ss_guidance_strength Type: numberDefault: 7.5Range: 0 - 10: Stage 1: Sparse Structure Generation - Guidance Strength
slat_guidance_strength Type: numberDefault: 3Range: 0 - 10: Stage 2: Structured Latent Generation - Guidance Strength

Output Schema

Example Execution Logs

INFO:predict:Loading and preprocessing input images...
INFO:predict:Using randomized seed: 1256691986
INFO:predict:Running TRELLIS pipeline...
Sampling:   0%|          | 0/38 [00:00<?, ?it/s]
Sampling:   5%|▌         | 2/38 [00:00<00:02, 12.02it/s]
Sampling:  11%|█         | 4/38 [00:00<00:02, 11.55it/s]
Sampling:  16%|█▌        | 6/38 [00:00<00:02, 11.42it/s]
Sampling:  21%|██        | 8/38 [00:00<00:02, 11.36it/s]
Sampling:  26%|██▋       | 10/38 [00:00<00:02, 11.32it/s]
Sampling:  32%|███▏      | 12/38 [00:01<00:02, 11.30it/s]
Sampling:  37%|███▋      | 14/38 [00:01<00:02, 11.29it/s]
Sampling:  42%|████▏     | 16/38 [00:01<00:01, 11.28it/s]
Sampling:  47%|████▋     | 18/38 [00:01<00:01, 11.28it/s]
Sampling:  53%|█████▎    | 20/38 [00:01<00:01, 11.27it/s]
Sampling:  58%|█████▊    | 22/38 [00:01<00:01, 11.25it/s]
Sampling:  63%|██████▎   | 24/38 [00:02<00:01, 11.26it/s]
Sampling:  68%|██████▊   | 26/38 [00:02<00:01, 11.27it/s]
Sampling:  74%|███████▎  | 28/38 [00:02<00:00, 11.27it/s]
Sampling:  79%|███████▉  | 30/38 [00:02<00:00, 12.19it/s]
Sampling:  87%|████████▋ | 33/38 [00:02<00:00, 14.86it/s]
Sampling:  95%|█████████▍| 36/38 [00:02<00:00, 16.93it/s]
Sampling: 100%|██████████| 38/38 [00:02<00:00, 12.82it/s]
Sampling:   0%|          | 0/12 [00:00<?, ?it/s]
Sampling:   8%|▊         | 1/12 [00:00<00:02,  5.13it/s]
Sampling:  17%|█▋        | 2/12 [00:00<00:01,  5.18it/s]
Sampling:  25%|██▌       | 3/12 [00:00<00:01,  5.17it/s]
Sampling:  33%|███▎      | 4/12 [00:00<00:01,  5.18it/s]
Sampling:  42%|████▏     | 5/12 [00:00<00:01,  5.18it/s]
Sampling:  50%|█████     | 6/12 [00:01<00:01,  5.18it/s]
Sampling:  58%|█████▊    | 7/12 [00:01<00:00,  5.18it/s]
Sampling:  67%|██████▋   | 8/12 [00:01<00:00,  5.19it/s]
Sampling:  75%|███████▌  | 9/12 [00:01<00:00,  5.19it/s]
Sampling:  83%|████████▎ | 10/12 [00:01<00:00,  5.19it/s]
Sampling: 100%|██████████| 12/12 [00:02<00:00,  6.78it/s]
Sampling: 100%|██████████| 12/12 [00:02<00:00,  5.65it/s]
INFO:predict:TRELLIS pipeline complete!
INFO:predict:Available output formats: dict_keys(['mesh', 'gaussian'])
INFO:predict:Starting video rendering...
INFO:predict:Generating color video from gaussian output...
Rendering: 0it [00:00, ?it/s]
Rendering: 20it [00:00, 191.48it/s]
Rendering: 42it [00:00, 206.36it/s]
Rendering: 63it [00:00, 201.27it/s]
Rendering: 84it [00:00, 203.23it/s]
Rendering: 107it [00:00, 211.09it/s]
Rendering: 120it [00:00, 205.80it/s]
INFO:predict:Available gaussian render types: ['color', 'depth']
INFO:predict:Generated color video successfully
INFO:predict:Video rendering complete!
INFO:predict:Generating GLB model...
INFO- Loaded 11494 vertices and 22961 faces.
0% done
25% done
50% done
75% done
100% done
Rendering: 0it [00:00, ?it/s]
Rendering: 11it [00:00, 108.20it/s]
Rendering: 22it [00:00, 106.84it/s]
Rendering: 34it [00:00, 109.84it/s]
Rendering: 45it [00:00, 109.48it/s]
Rendering: 57it [00:00, 111.09it/s]
Rendering: 69it [00:00, 110.84it/s]
Rendering: 81it [00:00, 111.59it/s]
Rendering: 93it [00:00, 111.27it/s]
Rendering: 100it [00:00, 110.63it/s]
INFO:predict:GLB model generation complete!
INFO:predict:Saving Gaussian point cloud as PLY...
INFO:predict:Gaussian PLY file saved successfully!
INFO:predict:Prediction complete! Returning results...

Version Details

Version ID: e8f6c45206993f297372f5436b90350817bd9b4a0d52d2a76df50c1c8afa2b3c
Version Created: May 16, 2025

Run on Replicate →