Writing study: LOTS of Fashion! Multi-Conditioning for Image Generation via Sketch-Text Pairing

ICCV overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← ICCV 2025 oral writing corpus

Federico Girella, Davide Talon, Ziyue Liu, Zanxi Ruan, Yiming Wang, Marco Cristani · 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

≈ 8.21 page-equivalents

Main figures / tables

4 / 2

Attached appendix figures / tables

0 / 0

Attached appendix span

No boundary detected; not assumed zero

Before references / appendix: ≈ 82.1%References / intervening material: ≈ 17.9%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	≈ 0.74 pp.	7.4%
2. Related Works	2	≈ 0.61 pp.	6.1%
3. Method	3	≈ 1.8 pp.	18%
4. Experiments	5	≈ 3.23 pp.	32.3%
5. Conclusions	8	≈ 0.62 pp.	6.2%
6. Acknowledgment	9	≈ 0.16 pp.	1.6%
References	9	≈ 1.79 pp.	17.9%

All detected figure and table locations

Main text

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

Attached appendix

No recognized caption labels.

Argument and organization

1. 1 Introduction; 2 Related Works

   Defines garment-specific sketch/text binding as the missing control beyond a global prompt.

2. 3 Method

   Encodes each pair independently and defers merging to adapter guidance during diffusion.

3. 4 Experiments; 5 Conclusions

   Introduces the derived Sketchy data, localization-aware evaluation and conditioned/fine-tuned baseline groups before design ablations.

Problem, claim and contribution

Sketch outlines and local textual garment attributes are described as complementary inputs whose incorrect binding causes attribute leakage.

Section 1, physical p.2 · Main paper

Argument and section order

The method separates within-pair fusion from between-pair merging, and the evaluation specifically measures whether attributes appear on the intended garment.

Sections 3.1–3.2 and 4.2, pp.3–6 · Main paper

Figures, captions and visual explanation

The pipeline keeps local pair tokens separate from global text and marks frozen versus trainable modules, explaining when shared weights do not imply cross-pair interaction.

Figure 2, p.4 · Main paper

Tables, comparisons and evidence

The table exposes local/global visual/text conditioning and distinguishes zero-shot from Sketchy-fine-tuned baselines; user precision, recall and F1 reveal localization tradeoffs.

Table 1, p.7 · Main paper

Appendix and reproducibility

Additional qualitative examples are routed to supplementary material, which was not inspected.

Section 4.3, p.7 · Main paper

Clarity and writing conventions

The conclusion gives an explicit non-visually grounded example, summer-vibe, to delimit what local attribute control does not solve.

Section 5, p.8 · Main paper

What to reuse—and what not to infer

Transferable lesson: For compositional control, report both desired-attribute presence and leakage, and show each baseline’s conditioning privileges.

Boundary: The task-specific study involves 14 participants and derived sketches/descriptions, not a professional-design productivity trial. LOTS does not win every FID, SSIM or recall cell; inter-annotator agreement is not statistical significance of a method difference.

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 · 10 pages