Writing study: TransmissiveGS: Residual-Guided Disentangled Gaussian Splatting for Transmissive Scene Reconstruction and Rendering

NeurIPS overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← NeurIPS 2026 oral writing corpus

Zhenyu Liang, Xiao Zhang, Tianchao Li, Jack C Cheng, Chi-Keung Tang · Official NeurIPS 2026 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

15 physical pages

Before references / appendix

≈ 11.09 page-equivalents

Main figures / tables

10 / 4

Attached appendix figures / tables

0 / 0

Attached appendix span

No boundary detected; not assumed zero

Before references / appendix: ≈ 73.93%References / intervening material: ≈ 26.07%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.

Source-bound correction: Top-level heading and reference/appendix boundaries checked against source PDF text. Bibliography titles/author initials, numbered contribution/property bullets, diagram steps, equations and TOC entries are not substantive headings. Section spans are page-coordinate estimates, not ink area or prose-only length. This is an author manuscript mirror, not a verified accepted-version template. Source layout uses physical single-column y/page-height, including equation-heavy pages.

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	≈ 1.21 pp.	8.07%
2 Related Work	3	≈ 0.55 pp.	3.67%
3 Preliminary	3	≈ 0.67 pp.	4.47%
4 TransmissiveGS	4	≈ 3.07 pp.	20.47%
5 Experiments	7	≈ 2.16 pp.	14.4%
6 Conclusion	9	≈ 2.33 pp.	15.53%
References	12	≈ 3.91 pp.	26.07%

All detected figure and table locations

Main text

• Figure 1 · p. 1
• Figure 2 · p. 2
• Figure 3 · p. 5
• Figure 4 · p. 7
• Table 1 · p. 8
• Figure 5 · p. 8
• Table 2 · p. 9
• Table 3 · p. 9
• Figure 6 · p. 9
• Figure 7 · p. 10
• Table 4 · p. 10
• Figure 8 · p. 11
• Figure 9 · p. 11
• Figure 10 · p. 11

Attached appendix

No recognized caption labels.

Argument and organization

1. Problem and contribution

   Uses a difficult refractive/reflective dome example to separate appearance quality from surface reconstruction, then explains the entanglement between transmitted content and near-field reflection.

2. Argument sequence

   Physical decomposition motivates dual Gaussian sets; a three-stage training procedure is then followed by rendering, geometry and ablation evaluations rather than a single blended quality claim.

3. Evidence and verification

   Tables separate appearance metrics from angular surface error; the text explicitly excludes a geometry-prior baseline from one quantitative comparison and explains why it still appears qualitatively.

Problem, claim and contribution

Uses a difficult refractive/reflective dome example to separate appearance quality from surface reconstruction, then explains the entanglement between transmitted content and near-field reflection.

Figure 1 and Introduction, pp. 1–2 · Main paper

Argument and section order

Physical decomposition motivates dual Gaussian sets; a three-stage training procedure is then followed by rendering, geometry and ablation evaluations rather than a single blended quality claim.

Sections 3–5, pp. 3–9 · Main paper

Figures, captions and visual explanation

Figure 3 draws three vertically ordered training stages with distinct borders, a legend separating operation flow from gradient flow, explicit freeze/initialization labels and actual residual/normal/render images. The ablation grid aligns the same facade view and highlighted regions across component removals, the complete model and the reference, with an additional surface-related row. Adjacent tables distinguish angular reconstruction error from rendering metrics, and the discussion explains the information-access reason for excluding a prior-assisted method from one numeric comparison.

Figure 3, p. 5 (render inspected); Tables 2–3 and Figure 6, p. 9 (additional rendered inspection) · Main paper

Tables, comparisons and evidence

Tables separate appearance metrics from angular surface error; the text explicitly excludes a geometry-prior baseline from one quantitative comparison and explains why it still appears qualitatively.

Tables 1–3 and Section 5.2, pp. 8–9 · Main paper

Appendix and reproducibility

The reflection-field layer diagram is paired with architecture ablations; further images include a failure mode rather than only additional attractive successes.

Appendices A–D, Figures 7–10 and Table 4, pp. 10–11 · Main paper

Clarity and writing conventions

Consistent scattering/reflection notation connects physical interpretation, equations and image panels; fairness qualifications are placed beside the affected evaluation.

Sections 4–5, pp. 4–9 · Main paper

What to reuse—and what not to infer

Transferable lesson: For a staged rendering design, show intermediate physical quantities and distinguish forward computation, gradient flow, initialization and frozen components.

Boundary: Preserve the comparison's use or absence of geometry priors. Rendering improvements and geometry accuracy are different claims, and the appendix documents a difficult transmission-dominated regime.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-10-02

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Writing-oriented scan across manuscript framing, section sequence, figure/table captions, selected argument/evaluation prose and conclusion; not exhaustive technical or proof verification. Sections 1–6, pp. 1–9; main exposition includes physical motivation, representation, staged reconstruction and evaluation. Source is the exact public author version listed below, not a verified accepted/camera-ready revision.

Appendix / supplement

Writing-oriented appendix/documentation scan: Appendices A–D, pp. 10–11, precede references: reflection-field architecture, hyperparameter ablations, qualitative comparisons and a failure mode; references pp. 12–15. Source code, data, model outputs and external video assets were not executed or independently validated.

Visually inspected PDF pages

5, 9

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.

• Public author arXiv manuscript 2605.10705v1; not verified identical to the accepted or camera-ready OpenReview revision · 15 pages