Hugging Face Daily Papers · · 3 min read

OcclusionFormer: Arranging Z-Order for Layout-Grounded Image Generation

Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.

ICML 2026. We first construct SA-Z, a large-scale dataset enriched with explicit occlusion ordering and pixel-level annotations. Building upon our proposed dataset, we introduce OcclusionFormer, a novel occlusion-aware Diffusion Transformer framework that explicitly models Z-order priority by decoupling instances and compositing them via volume rendering.</p>\n","updatedAt":"2026-05-21T03:50:20.915Z","author":{"_id":"67ff29ecbf6889a333c69c7a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67ff29ecbf6889a333c69c7a/zilMQrxIgUKYvHBVCHaKL.jpeg","fullname":"Henghui Ding","name":"HenghuiDing","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8134542107582092},"editors":["HenghuiDing"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/67ff29ecbf6889a333c69c7a/zilMQrxIgUKYvHBVCHaKL.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.21343","authors":[{"_id":"6a0e808b164dbbc68a26c59c","name":"Ziye Li","hidden":false},{"_id":"6a0e808b164dbbc68a26c59d","name":"Henghui Ding","hidden":false}],"publishedAt":"2026-05-20T00:00:00.000Z","submittedOnDailyAt":"2026-05-21T00:00:00.000Z","title":"OcclusionFormer: Arranging Z-Order for Layout-Grounded Image Generation","submittedOnDailyBy":{"_id":"67ff29ecbf6889a333c69c7a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67ff29ecbf6889a333c69c7a/zilMQrxIgUKYvHBVCHaKL.jpeg","isPro":false,"fullname":"Henghui Ding","user":"HenghuiDing","type":"user","name":"HenghuiDing"},"summary":"Recent layout-to-image models have achieved remarkable progress in spatial controllability. However, they still struggle with inter-object occlusion. When bounding boxes overlap, most existing methods lack explicit occlusion information, which makes the generation in intersection regions inherently ambiguous and hinders the determination of complex occlusion relationships. As a result, they often produce entangled textures or physically inconsistent layering in the overlapped areas. To address this issue, we first construct SA-Z, a large-scale dataset enriched with explicit occlusion ordering and pixel-level annotations. Building upon our proposed dataset, we introduce OcclusionFormer, a novel occlusion-aware Diffusion Transformer framework that explicitly models Z-order priority by decoupling instances and compositing them via volume rendering. Furthermore, to ensure fine-grained spatial precision, we introduce a queried alignment loss that explicitly supervises individual instances and enhances semantic consistency. The proposed method effectively reduces ambiguity in overlapping regions, enforces correct occlusion dependencies, and preserves structural integrity, leading to substantial accuracy gains across diverse scenes.","upvotes":6,"discussionId":"6a0e808b164dbbc68a26c59e","projectPage":"https://henghuiding.com/OcclusionFormer/","githubRepo":"https://github.com/FudanCVL/OcclusionFormer","githubRepoAddedBy":"user","ai_summary":"OcclusionFormer addresses inter-object occlusion challenges in layout-to-image generation by modeling explicit Z-order priority through diffusion transformers and volume rendering techniques.","ai_keywords":["layout-to-image models","inter-object occlusion","bounding boxes","diffusion transformer","Z-order priority","volume rendering","queried alignment loss","spatial controllability","occlusion-aware","instance decoupling"],"githubStars":14,"organization":{"_id":"68942389bd697013fd0c2df8","name":"FudanCVL","fullname":"FudanCVL","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/67ff29ecbf6889a333c69c7a/w_oRCf4rMPmNy62G-sI9p.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"67ff29ecbf6889a333c69c7a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67ff29ecbf6889a333c69c7a/zilMQrxIgUKYvHBVCHaKL.jpeg","isPro":false,"fullname":"Henghui Ding","user":"HenghuiDing","type":"user"},{"_id":"66fb03d6b505f1a04c39d935","avatarUrl":"/avatars/e9b830c460ec02037758c9b3469bb8ad.svg","isPro":false,"fullname":"Xuanlang Dai","user":"XuanlangDai","type":"user"},{"_id":"66aef8691dd7d0a8c6584724","avatarUrl":"/avatars/df9c2a56f3d0746cf64a330137a105b4.svg","isPro":false,"fullname":"Ziye Li","user":"TribeRinb","type":"user"},{"_id":"687f0efc664c6265a6fa37ee","avatarUrl":"/avatars/493ce89756b350646107b10647b4d599.svg","isPro":false,"fullname":"Kehan Lan","user":"lannn2333","type":"user"},{"_id":"6656e60de50d7c4088186e41","avatarUrl":"/avatars/5a6efb5835ae36762d3b4065538c73bc.svg","isPro":false,"fullname":"Zhaoyan Gong","user":"kakakanina","type":"user"},{"_id":"69a2c4f1816fc48aea554904","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/aURgGWS_ezoY9zI_g_Tcs.png","isPro":false,"fullname":"李嘉豪","user":"evelyn-jones4","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"68942389bd697013fd0c2df8","name":"FudanCVL","fullname":"FudanCVL","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/67ff29ecbf6889a333c69c7a/w_oRCf4rMPmNy62G-sI9p.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.21343.md"}">
Papers
arxiv:2605.21343

OcclusionFormer: Arranging Z-Order for Layout-Grounded Image Generation

Published on May 20
· Submitted by
Henghui Ding
on May 21
Authors:
,

Abstract

OcclusionFormer addresses inter-object occlusion challenges in layout-to-image generation by modeling explicit Z-order priority through diffusion transformers and volume rendering techniques.

AI-generated summary

Recent layout-to-image models have achieved remarkable progress in spatial controllability. However, they still struggle with inter-object occlusion. When bounding boxes overlap, most existing methods lack explicit occlusion information, which makes the generation in intersection regions inherently ambiguous and hinders the determination of complex occlusion relationships. As a result, they often produce entangled textures or physically inconsistent layering in the overlapped areas. To address this issue, we first construct SA-Z, a large-scale dataset enriched with explicit occlusion ordering and pixel-level annotations. Building upon our proposed dataset, we introduce OcclusionFormer, a novel occlusion-aware Diffusion Transformer framework that explicitly models Z-order priority by decoupling instances and compositing them via volume rendering. Furthermore, to ensure fine-grained spatial precision, we introduce a queried alignment loss that explicitly supervises individual instances and enhances semantic consistency. The proposed method effectively reduces ambiguity in overlapping regions, enforces correct occlusion dependencies, and preserves structural integrity, leading to substantial accuracy gains across diverse scenes.

Community

Paper submitter about 9 hours ago

ICML 2026. We first construct SA-Z, a large-scale dataset enriched with explicit occlusion ordering and pixel-level annotations. Building upon our proposed dataset, we introduce OcclusionFormer, a novel occlusion-aware Diffusion Transformer framework that explicitly models Z-order priority by decoupling instances and compositing them via volume rendering.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images

· Sign up or log in to comment

Get this paper in your agent:

hf papers read 2605.21343
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2605.21343 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2605.21343 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2605.21343 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

Discussion (0)

Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.

Sign in →

No comments yet. Sign in and be the first to say something.

More from Hugging Face Daily Papers