Can 3D scenes be represented by and rendered from a set of compressed tokens?
It turns out they can and it pairs very well with generative rendering to handle uncertainty!
Make sure to check out @Mohamma68780050's recent work Scenetok, accepted at #CVPR2026.
Links below.
๐ขSuper excited to announce that our work "๐ฆ๐ฐ๐ฒ๐ป๐ฒ๐ง๐ผ๐ธ: ๐ ๐๐ผ๐บ๐ฝ๐ฟ๐ฒ๐๐๐ฒ๐ฑ, ๐๐ถ๐ณ๐ณ๐๐๐ฎ๐ฏ๐น๐ฒ ๐ง๐ผ๐ธ๐ฒ๐ป ๐ฆ๐ฝ๐ฎ๐ฐ๐ฒ ๐ณ๐ผ๐ฟ ๐ฏ๐ ๐ฆ๐ฐ๐ฒ๐ป๐ฒ๐" has been accepted at #CVPR2026๐ข
You can bring our Sudoku solving diffusion models to other domains!
If you are at interested and at #ICML2025, come see @bartek_pog and @ChrisWewer's ๐ Spatial Reasoners package โ now released in beta!
Here are some examples for images and videos.
Links below.
Can diffusion models solve visual Sudoku?
If you are at #ICML2025, come to our poster in the Wednesday morning poster session (Poster Session 3 East, Poster 3412) and find out!
@ChrisWewer@bartek_pog Bernt Schiele @janericlenssen
Can image generators solve visual Sudoku?
Naively, no - with sequentialization and the correct order, they can!
Check out @ChrisWewer's and @bartek_pog's work, SRM, for details.
Project: https://t.co/o8hitFISSG
Paper: https://t.co/DOp4Ixyh3v
Code: https://t.co/5eQbajbhk2
Happy to announce that our work latentSplat is accepted at #ECCV2024.
latentSplat achieves high-quality 360ยฐ novel view synthesis with real-time rendering, purely trained on videos!
For more information, check out the project page:
https://t.co/6AQvk8fnaT
Excited to present our newest work latentSplat, achieving high quality in 360ยฐ novel view synthesis with real-time rendering, purely trained on videos!
For more information, check out the project page:
https://t.co/9yVpWr5aqy