SeeThrough3D: Occlusion Aware 3D Control in Text-to-Image Generation
Vaibhav Agrawal 1 Rishubh Parihar 2 Pradhaan S Bhat 2 Ravi Kiran Sarvadevabhatla 1 R. Venkatesh Babu 2 and 1 IIIT Hyderabad 2 IISc Bengaluru · 2026
Read SeeThrough3D: Occlusion Aware 3D Control in Text-to-Image Generation by Vaibhav Agrawal 1 Rishubh Parihar 2 Pradhaan S Bhat 2 Ravi Kiran Sarvadevabhatla 1 R. Venkatesh Babu 2 and 1 IIIT Hyderabad 2 IISc Bengaluru (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-23359.txt.
Description
Vaibhav Agrawal and four co-authors tackle occlusion reasoning in 3D layout-based image making. Occlusion reasoning is basic yet missed: partly hidden objects need depth-steady shapes and scale. Older methods make lifelike scenes that follow input layouts but often fail to model precisely which objects hide which.
The key is OSCR, an occlusion-aware 3D scene representation in which objects appear as see-through 3D boxes in a virtual setting, rendered from the desired camera viewpoint. See-through sides reveal hidden object regions, letting the model reason about what hides what, while the viewpoint gives camera control. Visual tokens from that picture condition a ready flow-based text-to-image model, and masked self-attention binds each object box to its caption, avoiding trait mixing. Trained on made-up multi-object scenes with much hiding, SeeThrough3D extends to unseen kinds with precise layout and camera control.
Catalog record
| CREATOR | Vaibhav Agrawal 1 Rishubh Parihar 2 Pradhaan S Bhat 2 Ravi Kiran Sarvadevabhatla 1 R. Venkatesh Babu 2 and 1 IIIT Hyderabad 2 IISc Bengaluru |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-23359