Rejected from ACM Transactions on Graphics? Where to Submit Next

Archive source code, dependencies, environment, training and test splits, pretrained weights, meshes, textures, scenes, camera paths, rendering settings, hardware, seeds, optimization traces, failed runs, videos, user-study protocol, exclusions, questionnaires, statistics, licenses, and repository state.

Write the contribution as graphics problem -> representation or algorithm -> assumptions -> artifact -> technical test -> visual or human evaluation -> efficiency -> failure boundary -> reusable insight. Mark each link as demonstrated, measured, perceived, inferred, or missing.

Diagnose the ACM TOG rejection before rerouting.

A desk rejection from TOG can reflect archival significance, scope, obvious completeness, visual evidence, or broad graphics interest. It may arrive through a regular journal route or a conference-linked journal track, and the governing rules differ.

A post-review rejection after external review gives you an artifact audit. Reviewer comments about missing ablations, unstable geometry, temporal artifacts, cherry-picked scenes, weak perceptual study design, unavailable assets, or excessive compute will follow the paper.

A transfer offer is not a standard path from TOG into an unrelated journal. A conference route adds deadline, anonymity, dual-submission, prior-publication, rebuttal, presentation, and proceedings rules. Check the current call directly.

Computer Graphics Forum

Best for: a revised broad graphics contribution with an archival account of method, comparisons, visual evidence, and limitations. Its journal and Eurographics ecosystem makes it a close audience match without making it an automatic fallback.

Think twice if: reviewers exposed a correctness or completeness defect. CGF still expects in-depth technical work, not a shortened TOG submission.

IEEE Transactions on Visualization and Computer Graphics

Best for: visualization, visual analytics, VR/AR, human-computer interaction for visual systems, graphics hardware, or an application where user tasks matter.

Think twice if: the paper uses a visualization dataset but contributes only a generic model. State the visual-computing question and evaluate the intended task.

Computers & Graphics

Best for: interactive techniques, practical visual systems, innovative graphics applications, reusable software, or a result whose use case is as important as its algorithm.

Think twice if: the application is a demo wrapped around an under-evaluated method. Reusable software still needs technical and empirical credibility.

The Visual Computer

Best for: work spanning graphics, animation, visualization, image processing, vision, or virtual environments that benefits from a broad visual-computing readership.

Think twice if: the paper's contribution is specialized enough that a geometry, visualization, vision, or HCI audience would evaluate it more accurately.

Graphical Models

Best for: a revised manuscript centered on geometry, shape representation, physical simulation, animation, deformation, or mathematical modeling.

Think twice if: geometry is only an intermediate format for a perception or application claim.

International Journal of Computer Vision

Best for: perception, reconstruction, recognition, inverse graphics, or scene understanding, with graphics output serving as evidence rather than the scientific endpoint.

Think twice if: the manuscript's contribution is how to synthesize, render, animate, edit, or author visual content. That remains a graphics question.

Write a one-paragraph editor test naming the visual-computing question, representation or algorithm, assumptions, decisive experiment, human or perceptual evidence, compute, failure regime, and reader benefit. If it fits every venue unchanged, the routing decision is not finished.

For a graphics-method route, identify what representation, algorithm, formulation, or physical model changes. Show correctness, ablations, matched baselines, visual evidence, and complexity.

For a visualization or interaction route, define user, task, decision, interface, comparator, study design, and outcome. Attractive screenshots are not task evidence.

For a vision route, state the perception problem, data distribution, supervision, generalization test, calibration, and failure modes. Do not sell rendering quality as recognition accuracy or vice versa.

For an application or software route, specify the workflow, reusable artifact, performance constraints, integration boundary, and what a practitioner can now do that was previously impractical.

Rewrite the title, teaser, first two abstract sentences, contribution bullets, and evaluation plan. If the destination story needs a user study, dataset, baseline, or failure analysis that does not exist, do the work or choose another route.

For each headline result, record dataset provenance, split, seed, hardware, comparator version, tuning budget, metric direction, statistical unit, selected examples, excluded failures, and artifact availability.

Revise the title, teaser, abstract, related work, contribution bullets, method diagram, assumptions, pseudocode, datasets, baselines, metrics, visual comparisons, video, user study, runtime, memory, ablations, failures, ethics, artifact statement, and conclusion together.

1. Name the graphics insight: distinguish the new representation, algorithm, formulation, interaction, or artifact from module assembly.
2. Map prior work fairly: compare against the closest technical lineage, not only convenient recent baselines.
3. Align the benchmark: equalize data, resolution, supervision, compute, tuning, preprocessing, and output constraints.
4. Test correctness: add proofs, invariants, synthetic controls, conservation, reconstruction checks, or unit tests where appropriate.
5. Expand visual evidence: use matched views, zooms, motion, difficult scenes, and uncurated failures.
6. Use claim-aligned metrics: connect image, geometry, temporal, perceptual, or task measures to the promised benefit.
7. Design human evaluation properly: preregister tasks where feasible, define units, randomization, exclusions, power, and analysis.
8. Report efficiency: include training, inference, memory, preprocessing, interaction latency, and hardware.
9. Make the artifact reproducible: document code, environment, data, models, seeds, licenses, and a minimal verification path.
10. State failure boundaries: name scenes, motions, materials, topology, lighting, users, or distributions where the method breaks.

Audit the graphics evidence chain before resubmission.

Appeal only when a documented factual or process error could change the decision. A reviewer preferring another aesthetic, significance threshold, or problem formulation is not usually an appeal case.

Use a conference route only after confirming the current cycle's originality, dual-submission, anonymity, prior-review, supplementary-material, patent, and presentation rules. A prior rejected manuscript may still require substantial revision and fresh anonymization.

Submit fresh when the contribution center moves to visualization, interaction, vision, geometry, or application software. Close the prior process. Never submit the same manuscript to another venue simultaneously.

In our review of ACM Transactions on Graphics manuscripts, we inspect the problem statement, technical lineage, assumptions, implementation, datasets, splits, baselines, metrics, visual comparisons, video, perceptual and user studies, runtime, memory, artifacts, licenses, failures, and claims. These are specific rejection patterns grounded in reviewable artifacts, not private ACM outcomes. Manusights internal analysis maps where the teaser, method, benchmark, video, and artifact support different versions of the claim. We observe a pattern only when a concrete mismatch can be reproduced or located in the submitted package.

Pattern 1: the teaser shows the best case, not the scientific case

For ACM Transactions on Graphics candidates, the teaser often combines scenes where the method succeeds, while baseline alignment and failure conditions appear later or nowhere. We build a matched evidence board with identical cameras, crops, resolution, inputs, and difficult cases.

We then align the teaser, Figure 1 caption, Results, video, and limitations. A strong teaser can remain, but it must not imply performance the full evaluation contradicts.

Pattern 2: baseline fairness changes the winner

In our ACM Transactions on Graphics review work, the proposed method may use higher resolution, more supervision, a newer backbone, extra optimization, or hand-selected parameters. We create an information-and-compute ledger, rerun feasible baselines, and disclose irreducible differences.

For graphics papers, the corrected table often narrows the numeric gain but clarifies the real contribution: quality at lower memory, stability under motion, editable representation, or a capability unavailable to the comparator.

Pattern 3: image metrics stand in for perception

PSNR, SSIM, LPIPS, Chamfer distance, or task-specific errors improve, but the paper claims perceptual quality, preference, usability, or realism. We add a study or narrow the claim, then separate participants, scenes, judgments, and repeated measures correctly.

We check the abstract, metric table, study protocol, statistics, and conclusion so a proxy is not described as human evidence.

Pattern 4: the artifact cannot reproduce the paper

The code depends on private preprocessing, missing assets, undocumented environment versions, unavailable weights, or a split assembled manually. We create a clean setup, minimal test case, deterministic seed path, license inventory, and expected output.

We revise the Methods, supplement, artifact statement, repository instructions, and limitations together. Reproducibility is part of the technical contribution, not a post-acceptance chore.

Choose the next venue only when the revised abstract can name the graphics question, representation or algorithm, assumptions, artifact, decisive technical test, visual or human evidence, efficiency, reproducibility path, and failure boundary. Verify current scope and submission rules immediately before uploading.

We checked current ACM author policy, TOG and SIGGRAPH-linked guidance, current destination scopes, the local Manusights owner inventory, and live exact-query results on July 13, 2026. We compared those boundaries with the manuscript artifacts we inspect in graphics reviews. Official sources establish scope and policy; the matrices, ledger, stress test, and review patterns are Manusights analysis.

Read final Search Console data after 14 complete days. At 21 complete days, keep, revise, consolidate, or stop based on indexation, exact-owner impressions, clicks, query fit, and qualified /ai-review starts. The source journal cluster had 11,625 impressions and no preview start; that does not prove exact rejection-query demand.