Writing study: Generating Physically Stable and Buildable Brick Structures from Text

ICCV overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← ICCV 2025 oral writing corpus

Ava Pun, Kangle Deng, Ruixuan Liu, Deva Ramanan, Changliu Liu, Jun-Yan Zhu · 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: Domain-grounded methods & physical systems (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

12 physical pages

Before references / appendix

≈ 8.14 page-equivalents

Main figures / tables

7 / 1

Attached appendix figures / tables

0 / 0

Attached appendix span

No boundary detected; not assumed zero

Before references / appendix: ≈ 67.83%References / intervening material: ≈ 32.17%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	1	≈ 0.66 pp.	5.5%
2. Related Work	2	≈ 0.86 pp.	7.17%
3. Dataset	3	≈ 1 pp.	8.33%
4. Method	4	≈ 2.46 pp.	20.5%
5. Experiments	6	≈ 2.01 pp.	16.75%
6. Conclusion	8	≈ 0.16 pp.	1.33%
Acknowledgments	9	≈ 0.1 pp.	0.83%
References	9	≈ 3.86 pp.	32.17%

All detected figure and table locations

Main text

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

Attached appendix

No recognized caption labels.

Argument and organization

1. 1 Introduction; 2 Related Work; 3 Dataset

   Separates visual plausibility from physical assembly and builds captioned stable brick sequences from meshes.

2. 4 Method

   Explains tokenization and fine-tuning before validity rejection, stability rollback and optional color/texture.

3. 5 Experiments; 6 Conclusion

   Contrasts mesh-to-brick baselines, sampling ablations and real assembly demonstrations.

Problem, claim and contribution

Physical realization is split into standard-component assembly and structural stability, rather than treated as an automatic consequence of attractive 3D generation.

Sections 1 and 3, physical pp.1–3 · Main paper

Argument and section order

The dataset stability filter is introduced before the same physical model is reused for inference verification; Algorithm 1 makes rollback termination explicit.

Sections 3–4.2 and Algorithm 1, pp.3–6 · Main paper

Figures, captions and visual explanation

The method overview separates tokenization, training and inference; red rejection marks and a rollback arrow show two distinct failure corrections.

Figure 3, p.4 · Main paper

Tables, comparisons and evidence

The main table defines validity and restricts stability/CLIP/DINO metrics to valid outputs; baseline rows with added stability analysis are kept separate from original methods.

Table 1 and Figure 5, p.7 · Main paper

Appendix and reproducibility

Dataset/training details are routed to Appendix A, robotic assembly to B and manual assembly to C. Those demonstrations and supplementary contents were not inspected.

Sections 3,5.1,5.3, pp.3–4,6,8 · Main paper

Clarity and writing conventions

Uniform brick coloring is distinguished from UV texturing, making clear which appearance outputs use ordinary commercially available bricks.

Section 4.3, p.6 · Main paper

What to reuse—and what not to infer

Transferable lesson: For constrained generation, report constraint satisfaction separately from similarity, state conditional metric denominators, and expose retry budgets.

Boundary: The full method reaches 98.8% stable outputs in the reported sample, not an unconditional guarantee. Axis-aligned library/grid and modeled static forces limit the physical claim; Algorithm 1 can return when the rollback budget is exceeded, and selected assembly demonstrations do not validate every generated design.

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, 7

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 · 12 pages