Рет қаралды 8,395
3D generation (text-to-3D, single-image to 3D) has significantly progressed this year. This video tries to summarize several ideas behind this technology and recent trends.
The papers discussed in the video:
• DreamFusion dreamfusion3d.github.io/
• Magic3D research.nvidia.com/labs/dir/...
• Fantasia3D fantasia3d.github.io/
• DreamGaussian dreamgaussian.github.io/
• RealFusion lukemelas.github.io/realfusion/
• NeRDi arxiv.org/abs/2212.03267
• NeuralLift-360 vita-group.github.io/NeuralLi...
• NFSD orenkatzir.github.io/nfsd/
• ProlificDreamer ml.cs.tsinghua.edu.cn/prolifi...
• Zero-1-to-3 zero123.cs.columbia.edu/
• Magic123 guochengqian.github.io/projec...
• DreamCraft3D mrtornado24.github.io/DreamCr...
• SyncedDreamer liuyuan-pal.github.io/SyncDre...
• MVDreamer mv-dream.github.io/
• Consistent-1-to-3 jianglongye.com/consistent123/
• Zero123++ arxiv.org/abs/2310.15110
• Wonder3D www.xxlong.site/Wonder3D/
• SweetDreamer sweetdreamer3d.github.io/
• LRM scalei3d.github.io/LRM/
• Instant3D instant-3d.github.io/
• DMV3D openreview.net/forum?id=H4yQe...