Writing study: LaRender: Training-Free Occlusion Control in Image Generation via Latent Rendering

ICCV overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← ICCV 2025 oral writing corpus

Xiaohang Zhan, Dingming Liu · Official ICCV 2025 oral record

Observed writing choices, not acceptance rules. This study examines the inspected manuscript version. It does not establish why the paper was accepted or that its presentation choices should be copied unchanged.

Figure, table and section ledger

Editorial contribution type: Architecture & representation (primary).

Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.

Canonical PDF

10 physical pages

Before references / appendix

≈ 8.05 page-equivalents

Main figures / tables

6 / 3

Attached appendix figures / tables

0 / 0

Attached appendix span

No boundary detected; not assumed zero

Before references / appendix: ≈ 80.5%References / intervening material: ≈ 19.5%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.

Section-by-section lengths and source pages

The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.

Top-level heading	Starts on PDF page	Estimated span	Share of entire PDF
1. Introduction	2	≈ 0.67 pp.	6.7%
2. Related Work	2	≈ 0.57 pp.	5.7%
3. Methodology	3	≈ 1.99 pp.	19.9%
4. Experiments	5	≈ 3.09 pp.	30.9%
5. Conclusion, Limitations and Social Impact	8	≈ 0.68 pp.	6.8%
References	9	≈ 1.95 pp.	19.5%

All detected figure and table locations

Main text

• Figure 1 · p. 1
• Figure 2 · p. 4
• Table 1 · p. 6
• Table 2 · p. 6
• Figure 3 · p. 6
• Table 3 · p. 6
• Figure 4 · p. 7
• Figure 5 · p. 8
• Figure 6 · p. 8

Attached appendix

No recognized caption labels.

Argument and organization

1. 1 Introduction; 2 Related Work

   Distinguishes photographic occlusion from bounding-box position and linguistic front/behind relations.

2. 3 Methodology

   Adapts volume-rendering composition to object latents, then schedules semantic density and defines user/LLM input modes.

3. 4 Experiments; 5 Conclusion, Limitations and Social Impact

   Combines occlusion metrics, human rankings and ablations before opacity applications and labeled failures.

Problem, claim and contribution

An explicit example explains that a person standing in front of another need not visually occlude them, making the target relation operationally distinct from scene-graph position.

Section 2.1, physical p.2 · Main paper

Argument and section order

The classical rendering formula precedes its latent analogue, followed by the denoising schedule needed to avoid premature feature mixing.

Sections 3.1–3.3, pp.3–5 · Main paper

Figures, captions and visual explanation

The overview links graph sorting to layered features, attention/box masks and accumulated transmittance; the final failure strip labels wrong position, wrong occlusion, lost concepts and mixed concepts.

Figure 2, p.4; Figure 6, p.8 · Main paper

Tables, comparisons and evidence

Human ranking and perceived success include interval definitions and denoising-step differences; the evaluation explains why FID was omitted for these small/no-real-image datasets.

Section 4.2, p.5; Table 1, p.6 · Main paper

Appendix and reproducibility

Additional FLUX examples, attention-map ablations and a two-element opacity grid are routed to supplements, not inspected.

Sections 4.3–4.5 and Figure 5, pp.7–8 · Main paper

Clarity and writing conventions

The conclusion acknowledges that unreasonable layouts and feature erasure/mixing can defeat occlusion control, qualifying the physics-grounded terminology.

Section 5, p.8 · Main paper

What to reuse—and what not to infer

Transferable lesson: Define the visual relationship being controlled and pair a physical analogy with concrete failure categories and metric limitations.

Boundary: Latent rendering is not a verified physical renderer or unconditional occlusion guarantee. Semantic opacity is an artistic control, not calibrated material transmittance; user/LLM graphs and boxes remain inputs, and training-free still depends on pretrained generators and optional language models.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-10-07

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Writing-oriented scan across the main manuscript, physical PDF pp.1–8, covering framing, section progression, method explanations, captions, comparison protocols and concluding scope; reference boundary checked. Selected full-page rasters were inspected. Not exhaustive line-by-line reading, proof verification or experimental reproduction.

Appendix / supplement

Main-paper appendix/supplement references were checked. A separate supplementary source is linked in the inventory but was not retrieved or inspected; its contents are not included in the reading or visual-page counts.

Visually inspected PDF pages

4, 8

This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.

• Official CVF open-access ICCV 2025 accepted-paper PDF, retrieved 7 October 2026; CVF states identical to accepted version except watermark; not assumed identical to initial submission · 10 pages