Writing study: FixTalk: Taming Identity Leakage for High-Quality Talking Head Generation in Extreme Cases

ICCV overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← ICCV 2025 oral writing corpus

Shuai Tan, Bill Gong, Bin Ji, Ye Pan · 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

13 physical pages

Before references / appendix

≈ 8.1 page-equivalents

Main figures / tables

8 / 4

Attached appendix figures / tables

0 / 0

Attached appendix span

No boundary detected; not assumed zero

Before references / appendix: ≈ 62.31%References / intervening material: ≈ 37.69%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	≈ 0.88 pp.	6.77%
2. Related Work	2	≈ 0.95 pp.	7.31%
3. Investigate Identity Leakage in GAN	3	≈ 0.9 pp.	6.92%
4. Method	4	≈ 2.01 pp.	15.46%
5. Experiments	6	≈ 2.22 pp.	17.08%
6. Conclusion	8	≈ 0.16 pp.	1.23%
Acknowledgements	9	≈ 0.05 pp.	0.38%
References	9	≈ 4.9 pp.	37.69%

All detected figure and table locations

Main text

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

Attached appendix

No recognized caption labels.

Argument and organization

1. 1 Introduction; 2 Related Work

   Introduces efficiency, decoupled control and rendering quality as simultaneous requirements, then narrows the design to GAN-based animation.

2. 3 Investigate Identity Leakage; 4 Method

   Uses self- versus cross-driven feature substitutions to motivate separate motion and detail modules.

3. 5 Experiments; 6 Conclusion

   Separates video-driven and audio-driven protocols, component ablation, system efficiency and transfer to a second baseline.

Problem, claim and contribution

Names two visible defects, identity leakage and rendering artifacts, instead of treating talking-head quality as one undifferentiated score.

Sections 1 and 3.2, physical pp.2–4 · Main paper

Argument and section order

A diagnostic section comes before the proposed method, making the same leaked identity feature harmful for control but useful for missing-detail recovery.

Sections 3–4, pp.3–6 · Main paper

Figures, captions and visual explanation

The overview fades the inherited baseline, colors EMI and EDI separately, and distinguishes training-only from training/inference paths with a line-style legend.

Figure 4, p.5 · Main paper

Tables, comparisons and evidence

Separate HDTF video and MEAD audio tables prevent task mixing; the MEAD table retains the baseline’s stronger synchronization score. The user study specifies 20 participants and 10 sampled videos per method.

Tables 1–3 and Section 5.2, p.7 · Main paper

Appendix and reproducibility

Baseline FDM, audio extension, full baseline comparisons, quantitative ablations and ethics are explicitly assigned to supplementary material; only these main-text pointers were inspected.

Sections 3.1, 4.2, 5 and 6, pp.3,6–8 · Main paper

Clarity and writing conventions

The diagnostic question forms a clear bridge into the two-module method, but absolute phrases such as completely disentangled and free of any artifacts exceed what finite comparison tables establish.

Sections 3.2–4.2, pp.4–6 · Main paper

What to reuse—and what not to infer

Transferable lesson: Let a controlled failure diagnosis explain why each new module exists, and visually separate inherited, training-only and inference-time computation.

Boundary: A feature-substitution case and average identity metric do not prove complete disentanglement. Real-time speed is tied to the reported A100 and image setup; ethical details are not verified by this main-only scan.

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

5, 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 · 13 pages