Research map / 3D vision / 3D generation
DiffTF++: 3D-Aware Diffusion Transformer for Large-Vocabulary 3D Generation
Ziang Cao, Fangzhou Hong, Tong Wu, Liang Pan, Ziwei Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add triplane regularization (TV and L2) (ablation) | diffusion on fitted triplanes · OmniObject3D | FID: 125.21 -> 113.01 | Improved |
| add triplane normalization (ablation) | diffusion on fitted triplanes · OmniObject3D | FID: 113.01 -> 109.81 | Improved |
| add cross-plane attention 3D-aware encoder/decoder (ablation) | DiffTF · OmniObject3D | FID: 109.81 -> 39.13 | Improved |
| replace CP-TF with original 2D transformer (ablation) | CP encoder model · OmniObject3D | FID: 39.13 -> 52.35 | Hurt |
| add 3D-aware transformer (CP-TF) (ablation) | CP encoder model · OmniObject3D | FID: 39.13 -> 25.36 | Improved |
| replace 3D-aware convolution with the proposed 3D-aware modules (ablation) | DiffTF · OmniObject3D | FID: 84.55 -> 25.36 | Improved |
| add multi-view reconstruction loss (ablation) | DiffTF · OmniObject3D | FID: 25.36 -> 23.30 | Improved |
| add 3D-aware refinement (ablation) | DiffTF + reconstruction loss · OmniObject3D | FID: 23.30 -> 20.97 | Improved |
| DiffTF to DiffTF++ (loss + refinement) | DiffTF · OmniObject3D | FID: 25.36 -> 20.97 | Improved |
| DiffTF to DiffTF++ | DiffTF · ShapeNet Car | FID: 36.68 -> 32.74 | Improved |
| DiffTF to DiffTF++ | DiffTF · ShapeNet Plane | FID: 14.46 -> 11.32 | Improved |
| DiffTF to DiffTF++ | DiffTF · ShapeNet Chair | FID: 35.16 -> 31.52 | Improved |
Later papers that compared with this method
1 later accepted paper reports DiffTF as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- UVGS: Reimagining Unstructured 3D Gaussian Splatting using UV Mapping (CVPR 2025 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GET3D — GET3D: A Generative Model of High Quality 3D Textured Shapes Learned from Images (NeurIPS 2022) · better on 5 of 5 shared results
- DiffRF — DiffRF: Rendering-Guided 3D Radiance Field Diffusion (CVPR 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Large-Vocabulary 3D Diffusion Model with Transformer (ICLR 2024 · similarity 0.94)
- PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers (NeurIPS 2025 · similarity 0.91)
- One-2-3-45++: Fast Single Image to 3D Objects with Consistent Multi-View Generation and 3D Diffusion (CVPR 2024 · similarity 0.90)
- Cycle3D: High-quality and Consistent Image-to-3D Generation via Generation-Reconstruction Cycle (AAAI 2025 · similarity 0.90)
- Diffusion 3D Features (Diff3F): Decorating Untextured Shapes with Distilled Semantic Features (CVPR 2024 · similarity 0.90)
- DMV3D: Denoising Multi-view Diffusion Using 3D Large Reconstruction Model (ICLR 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
