[NeurIPS 2025] Direct3D‑S2: Gigascale 3D Generation Made Easy with Spatial Sparse Attention
Python · active 2025-05-23 → 2026-06-13 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 387 days from 2025-05-23 to 2026-06-13. Pushes: 15 total, peak 4 in a day. Pull requests: 3 total, peak 1 in a day. Issues: 45 total, peak 7 in a day. Comments: 136 total, peak 31 in a day. Stars: 712 total, peak 60 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| Linyou | 18 | 10 | 0 | 8 |
| Deathdadev | 14 | 0 | 0 | 13 |
| y4my4my4m | 10 | 0 | 0 | 8 |
| shuangw98 | 10 | 5 | 0 | 5 |
| SNAKEIX | 9 | 0 | 0 | 9 |
| niksad8 | 9 | 0 | 0 | 9 |
| cleverestx | 8 | 0 | 0 | 8 |
| SoftologyPro | 7 | 0 | 0 | 6 |
| KelestZ | 6 | 0 | 0 | 2 |
| Dariusz1989 | 4 | 0 | 0 | 3 |
| loktar00 | 4 | 0 | 0 | 4 |
| wsbagnsv1 | 4 | 0 | 0 | 4 |
| chenerg | 4 | 0 | 0 | 1 |
| bearstonem | 3 | 0 | 0 | 2 |
| yuedajiong | 3 | 0 | 0 | 2 |
| rwn17 | 3 | 0 | 0 | 1 |
| deman311 | 3 | 0 | 0 | 2 |
| Sephi87 | 3 | 0 | 0 | 2 |
| rfeinman | 2 | 0 | 1 | 1 |
| Aikon404 | 2 | 0 | 0 | 2 |
Recent activity
Latest issues, pull requests and releases
- Pull request#67Pigumalio2025-11-30 06:37
- Issue#66Pigumalio2025-11-30 06:36Full support for RTX 5090 (and Blackwell) fork.
- Issue#65LeeCASC2025-11-27 06:40SDF data
- Issue comment#61jjjkkyz2025-11-17 06:50How to use sparse vae?
- Issue#64NeoTech2025-11-14 21:52Error in buildig udf_ext - this fixes it.
- Issue#63MrLihj2025-10-23 03:56Does the conditioning effect of the input image weaken—or even vanish—when training at high-resolution voxel grids?
- Issue comment#62shuangw982025-10-16 07:38Asking about the normalization of both mesh and voxel grid
- Issue#62DavidYan20012025-10-12 16:14Asking about the normalization of both mesh and voxel grid
- Issue comment#12pio570192025-10-03 12:14Windows support
- Issue#58SerdarHelli2025-09-05 13:25Training Code Availability
- Issue#57Vxtzq2025-09-01 16:01VRAM-Optimized Version of Direct3D-S2
- Issue#56palindar2025-08-31 11:35Project update
- Issue comment#12cleverestx2025-08-16 04:49Windows support
- Issue#55jjjkkyz2025-08-15 08:56Remove inner mesh without refiner
- Issue comment#47LDYang6942025-08-01 06:29SSA runs very slow on NVIDIA H20 GPU
- Issue comment#30MathamPollard2025-07-24 07:37windows跑不起来 卡在torchsparse模块
- Issue#53MathamPollard2025-07-24 03:46一直卡在sparse sampling不动
- Issue comment#52MathamPollard2025-07-24 03:43Pointer argument (at 0) cannot be accessed from Triton (cpu tensor?)
- Issue#52MathamPollard2025-07-22 12:22Pointer argument (at 0) cannot be accessed from Triton (cpu tensor?)
- Issue comment#51Lefzeus2025-07-22 07:34ModuleNotFoundError: No module named 'udf_ext'
- Issue#50rfeinman2025-07-21 01:51Details/reference on dense stage?
- Issue#49O2Mania2025-07-15 09:27Different results from the project page
- Issue comment#41Leko72025-07-10 11:29Do you have any plans to release the training code and processed data?
- Issue comment#480shixu2025-07-08 12:12设备配置运行要求是什么？
- Issue comment#46chenerg2025-07-07 23:59关于sort_block 函数的效果
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 712 stars here means stars gained during the window, not the repo's star count.