Sign in

Slava Elizarov

@selizarov.bsky.social
264 followers 857 following 15 posts

Senior research scientist at Unity | Generative models, Computer Graphics

PostsRepliesMedia
Slava Elizarov @selizarov.bsky.social · 23/01/2025
Thrilled to share that Geometry Image Diffusion has been accepted to #ICLR2025! 🚀 Paper: openreview.net/forum?id=Glm...
openreview.net
Geometry Image Diffusion: Fast and Data-Efficient Text-to-3D with...
Generating high-quality 3D objects from textual descriptions remains a challenging problem due to high computational costs, the scarcity of 3D data, and the complexity of 3D representations. We...
191
Slava Elizarov @selizarov.bsky.social · 06/12/2024
P.P.S. We recommend you check out Omages (omages.github.io), an awesome concurrent work that also explores geometry images (called "Omages") for 3D generation. We believe GIMs have a bright future in deep learning — let’s bring it forward together 🚀
omages.github.io
Object Images 64x
An Object is Worth 64x64 Pixels: Generating 3D Object via Image Diffusion
020
Slava Elizarov @selizarov.bsky.social · 06/12/2024
P.S. Thanks to Simon Donné, @ciararowles.bsky.social, Shimon Vainer, Dante De Nigris, and Konstantin Kutsy for being such an awesome team! Additional thanks to Alexander Demidko and Dr. Lev Melnikovsky from the Weizmann Institute for all the insightful discussions we had during this project
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
So whether you’re looking for speed, flexibility, or eco-friendly workflows, Geometry Image Diffusion has you covered. Got curious? Dive into our paper to learn more! Paper: arxiv.org/abs/2409.03718 Code: coming soon Site: unity-research.github.io/Geometry-Ima... (10/10)
arxiv.org
Geometry Image Diffusion: Fast and Data-Efficient Text-to-3D with Image-Based Surface Representation
Generating high-quality 3D objects from textual descriptions remains a challenging problem due to computational cost, the scarcity of 3D data, and complex 3D representations. We introduce Geometry Ima...
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
And we’re not just saving forests. The assets you generate with Geometry Image Diffusion are free from baked-in lighting. Re-light them in any environment to fit your scene and save some energy while you’re at it! 💡 (9/10)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
(But I must admit that it’s hard to resist generating thousand barrels because they’re all so different)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
Why produce a thousand barrels? Let’s save the forest! 🌳 Just edit the one you’ve already generated (8/10)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
Want an unexpected twist? The generated 3D objects come with meaningful, separable parts, making them easy to edit and manipulate (7/10)
120
Slava Elizarov @selizarov.bsky.social · 06/12/2024
Our assets can be easily triangulated by connecting neighboring pixels, and come unwrapped with textures included — no waste here ♻️ (6/10)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
(Prompts: Lovecraftian teacup with a tentacle instead of the handle; A steampunk airplane; An avocado-shaped chair)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
Our model is trained on a 100k subset of Objaverse — smaller than what’s typically used for 3D generation. Yet, it generalizes well across a wide range of prompts (5/10)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
With a frozen Stable Diffusion model for textures and its trainable copy for geometry, the geometry model can tap into SD’s powerful natural image prior (4/10)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
At the heart of our method is Collaborative Control. It allows two models to work together — one for generating the geometry image and another for creating textures — all while sharing information to ensure everything lines up perfectly 🤝(3/10)
110
Slava Elizarov @selizarov.bsky.social · 06/12/2024
The secret? We use geometry images, which are essentially 2D representations of 3D surfaces 🖼️ (think of GIMs as UV maps’ close cousins) This lets us recycle existing Text-to-Image models like Stable Diffusion, instead of building complex 3D architectures from scratch (2/10)
120
Slava Elizarov @selizarov.bsky.social · 06/12/2024
Does 3D generation always have to be either slow or complex and data-hungry?🤔 We don’t think so! With Geometry Image Diffusion, we’re all about reusing (and recycling ♻️) what already works — making it faster and easier by reducing complexity and data needs 🚀(1/10)
171