Research map / Image generation / Text-to-image diffusion
TPA3D: Triplane Attention for Fast Text-to-3D Generation
Bin-Shih Wu, Hong-En Chen, Sheng-Yu Huang, Yu-Chiang Frank Wang
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 word-level refinement (TPA) to sentence-only conditioning vs TAPS3D | TPA3D · ShapeNet Car | CLIP R-Precision@5: 12.55 -> 80.94 | Improved |
| TPA3D vs unconditional GET3D on FID for ShapeNet | TPA3D · ShapeNet Car | FID: 11.50 -> 18.50 | Hurt |
| TPA3D vs GET3D on multi-class OmniObject3D | TPA3D · OmniObject3D Vehicle | FID: 98.15 -> 68.80 | Improved |
| add cross-word attention (ablation) | TPA block · ShapeNet Car | CLIP R-Precision@5: 70.94 -> 80.94 | Improved |
| add cross-plane attention (ablation) | TPA block · ShapeNet Car | FID: 30.0 -> 18.5 | Improved |
| add TPA blocks (all) (ablation) | TPA3D · ShapeNet Car | FID: 32.6 -> 18.5 | Improved |
| add texture TPA branch (ablation) | TPA3D · ShapeNet Car | CLIP R-Precision@5: 76.85 -> 80.94 | Improved |
| add geometry TPA branch (ablation) | TPA3D · ShapeNet Car | FID: 20.4 -> 18.5 | Improved |
| use 6 TPA blocks instead of 3 (ablation) | TPA3D · ShapeNet Car | FID: 20.1 -> 18.5 | Improved |
| add CLIP similarity loss L_clip (ablation) | TPA3D · ShapeNet Car | FID: 56.9 -> 18.5 | Improved |
| add mismatching loss L_mis (ablation) | TPA3D · ShapeNet Car | FID: 68.6 -> 18.5 | Improved |
| use TPA3D instead of TAPS3D with complex captions | TPA3D · ShapeNet Car FID gap | FID gap vs TAPS3D: 14.1% -> 29.9% | Improved |
Closest papers by meaning
- IT3D: Improved Text-to-3D Generation with Explicit View Synthesis (AAAI 2024 · similarity 0.89)
- PI3D: Efficient Text-to-3D Generation with Pseudo-Image Diffusion (CVPR 2024 · similarity 0.88)
- DiffusionGAN3D: Boosting Text-guided 3D Generation and Domain Adaptation by Combining 3D GANs and Diffusion Priors (CVPR 2024 · similarity 0.88)
- DIRECT-3D: Learning Direct Text-to-3D Generation on Massive Noisy 3D Data (CVPR 2024 · similarity 0.86)
- MARVEL-40M+: Multi-Level Visual Elaboration for High-Fidelity Text-to-3D Content Creation (CVPR 2025 · similarity 0.86)
- VP3D: Unleashing 2D Visual Prompt for Text-to-3D Generation (CVPR 2024 · similarity 0.86)
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.
