Computational Visual Media

About Computational Visual Media

Computational Visual Media — an open-access visual-computing journal with no author charge

Computational Visual Media (abbreviated Comput. Vis. Media, CVM) is a single-blind peer-reviewed, fully open-access journal published quarterly by Tsinghua University Press, co-published with Springer, and now presented on the SciOpen platform. Founded in 2015 by the Department of Computer Science and Technology at Tsinghua University, it publishes original high-quality research papers and significant review articles on novel ideas, methods and systems relevant to visual media — the creation, processing, analysis and understanding of images, video, geometry and other visual content. Two features set it apart from most high-impact computer-science titles, and both are verified in this profile from primary sources: it is listed in the Directory of Open Access Journals (DOAJ) under a Creative Commons Attribution (CC BY) licence, and it charges authors nothing — the DOAJ record states it has no article-processing charge, because publication costs are covered by Tsinghua University.

That combination — open access, no author fee, Q1 standing and a Journal Impact Factor displayed by the publisher's SciOpen platform as 8.3 for 2025, with a CiteScore displayed as 28.6 for 2025 — makes the journal intensely attractive, and therefore selective in practice despite its open door on charges. The record saved on this site lists a much higher Impact Factor of 17.3 and an H-index of 36. Those stored figures conflict with the publisher-platform figures in hand and with several aggregator values, and this profile treats that conflict openly in the metrics section below rather than repeating 17.3 as current fact. Authors evaluating this journal for a real submission should work from the publisher's current display and the underlying Journal Citation Reports or Scopus record for the year that matters to them.

Quick facts

History and institutional roots

A Tsinghua University Press account in hand records that Computational Visual Media is an English-language academic journal founded by Tsinghua's Department of Computer Science in 2015, co-published by Tsinghua University Press and Springer, with Professor Shi-Min Hu (Hu Shimin) of Tsinghua as chief editor and, at founding, Professor Ralph Martin of Cardiff University and Professor Ming C. Lin of the University of Maryland as associate editors. The founding editor's message, published in volume 1, number 1 (March 2015), launched the journal quarterly, with a first issue of eight papers devoted to the 3rd International Conference on Computational Visual Media, held at Tsinghua in April 2015, covering image decomposition, smoothing, completion, denoising and stylisation through to Poisson-disk sampling, cartoon structure-line extraction and sketch-based modelling — a table of contents that already announced the journal's graphics-plus-vision character.

At its establishment the journal received funding from a Chinese programme supporting the international influence of science and technology journals, and in 2019 it was selected for the China Science and Technology Journal Excellence Action Plan hosted by the China Association for Science and Technology, per the Tsinghua account. In April 2021 it was included in the Science Citation Index Expanded (SCIE) of Web of Science. By its sixth year, the same account reports inclusion in 24 well-known databases at home and abroad, including SCIE, EI, DBLP, INSPEC, Scopus, DOAJ and OCLC services. The journal's centre of gravity remains Beijing — bibliographic summaries give the publisher's address in the Xue Yan Building, Beijing 100084 — while its authorship, editorship and readership are international, as a Springer co-publication in English would suggest. The saved record's stored country (United States) and stored publisher (a numeric value, '43') both conflict with these primary sources; the publisher is Chinese, as the DOAJ record (country CN) independently confirms. See the conflict note in this profile's report.

Aims and scope in detail

The publisher's scope, reproduced consistently across Springer and Tsinghua descriptions in hand, is to publish original high-quality research papers and significant review articles on novel ideas, methods and systems relevant to visual media, including interdisciplinary work that combines computer graphics, computer vision, image and video processing, geometric computing and machine learning. The topic list given by the publisher includes:

• Editing and composition of visual media; interactive editing of visual media.
• Geometric computing for images and video; geometry modelling and processing.
• Machine learning for visual media.
• Physically based animation; realistic rendering.
• Recognition and understanding of visual media; cognition of visual media.
• Visual computing for robotics.
• Visualization and visual analytics.
• 3D visual media processing; classification and composition of visual media; content security for visual media; enhancement and re-rendering of visual media; visual media retrieval; social media computing.

Read that list as a map of a community, not a keyword net. The journal's home community is the graphics and visual-computing community centred on conferences such as the Computational Visual Media conference itself, Pacific Graphics and related venues, extended toward computer vision and machine learning where the work is about visual media rather than learning theory in the abstract. A pure machine-learning paper with an image dataset used only as a testbed, a robotics paper with no visual-computing contribution, or a visualisation paper with no computational novelty will each face a fit question. Conversely, work that crosses the graphics–vision boundary — neural rendering, 3D reconstruction and generation, geometry processing with learned methods, visual analytics that couples computation with human judgement — is exactly the interdisciplinary territory the scope claims. Recent issue contents visible on SciOpen confirm the living scope: surveys on self-supervised pre-training for 3D point clouds and AI-driven generation of 3D CAD models, alongside research on low-light image and video analysis, multi-view weakly supervised 3D human pose estimation and LiDAR data augmentation.

Editorial leadership

The SciOpen journal page lists Shi-Min Hu of Tsinghua University as Editor-in-Chief, consistent with the founding account. Hu is a professor in Tsinghua's Department of Computer Science and Technology whose own research covers digital geometry processing, video processing, rendering, computer animation and computer-aided geometric design — a profile that matches the journal's centre of gravity. Historical associate editors named at founding (Ralph Martin, Cardiff; Ming C. Lin, Maryland) illustrate the journal's international editorial reach from its first volume; check the current editorial board on SciOpen or the Springer page before relying on any individual associate-editor name, because boards evolve over a decade of publication. The publisher also states that the journal uses a rapid review policy — quantified in the metrics section from the SciOpen display — which authors should treat as a service aim reported by the publisher's platform, not a guarantee for any individual manuscript.

Metrics — a genuine conflict, reported honestly

This journal is the clearest case in this batch of why stored and secondary figures must never be repeated without checking. The record saved on this site lists an Impact Factor of 17.3 and an H-index of 36. The publisher's own SciOpen journal page, retrieved for this profile, displays an Impact Factor of 8.3 (2025) and a CiteScore of 28.6 (2025). Aggregator summaries in hand give yet other values for other years and definitions, including an Impact Factor-type value of 7.10 in one summary, 8.3 in another (attributed to the June 2026 Journal Citation Reports release), SJR values around 2.965–3.166, and H-index values in the high 20s to mid-30s depending on source and date. The 17.3 figure does appear in one aggregator's 2024 metrics table as a Journal Impact Factor, which suggests the stored value preserves an earlier — and unusually high — year's figure rather than being invented, but it is not the figure the publisher currently displays, and it must not be quoted as current.

The SciOpen page also displays service figures: 28 days to first decision, 93 days to acceptance and 60 days to publication, as platform-displayed values in hand. Treat those as the publisher platform's reported medians or aims at the time of checking, useful for planning, and confirm them on the live page when you submit. What is not in dispute across any source in hand is the qualitative standing: a Q1 journal in its Scopus categories (Artificial Intelligence; Computer Graphics and Computer-Aided Design; Computer Vision and Pattern Recognition, per the saved record's subject areas), indexed in SCIE since 2021, with citation performance far above the median computer-science title. If a tenure, graduation or funding rule depends on a threshold impact factor or quartile, verify the exact figure in Journal Citation Reports or Scopus for the specified year — the spread of values above shows how badly a directory page, including this one, can mislead if it pretends to a single settled number.

Indexing and discoverability

Per the Tsinghua account and bibliographic summaries in hand, the journal is indexed in SCIE (Web of Science, from April 2021), Scopus, EI Compendex, DBLP, INSPEC and DOAJ, among a wider set of 24 databases reported by the publisher's account, including CNKI and OCLC WorldCat services. For computer-science authors, DBLP and EI coverage matter alongside the citation indexes, because the field discovers work through those channels as much as through Web of Science. DOAJ listing was verified directly for this profile: the DOAJ search API, queried by both ISSNs, returned the journal with publisher Tsinghua University Press, country China, CC BY licence and no article-processing charge, with a link to the journal's SciOpen about page. Confirm any single database claim in that database's own source list if it carries formal consequences.

Open access and charges — the verified position

This is the journal's standout practical feature, and it is verified from two independent primary-type sources in hand. The DOAJ record states the journal has no article-processing charge and is published under CC BY. Publisher and Springer descriptions explain why: publication costs are covered by Tsinghua University, so authors do not need to pay an article-processing charge. For authors without grant funding for publication fees — and for funders who mandate CC BY open access — that removes two of the most common barriers to publishing in a high-standing venue. It also predictably increases competition for space: a free, open, Q1 journal with a rapid-review reputation receives more good submissions than it can publish, so the absence of a fee should never be mistaken for an easy acceptance. The saved record's APC entry ('N/A') is consistent with the no-charge position, though for the wrong-looking reason; the accurate statement is that there is no APC because the sponsor covers costs, as the sources above state. Licence, charge and sponsor arrangements can change, so confirm them on the SciOpen / DOAJ record on the day you submit.

Who should submit — and how to think about fit

Submit here if your work offers a novel idea, method or system for visual media with evidence at the standard the graphics and visual-computing community expects: comparisons against strong current methods on the field's standard datasets and test suites, ablations that show which component carries the gain, and — for systems work — a demonstration that the system does something previously impractical. Review articles should be significant syntheses by authors with standing in the area, not annotated bibliographies. The quarterly cadence and rapid-review aim suit authors who need a journal version of conference-grade work without a multi-year queue, and the CC BY, no-fee model suits authors with open-access mandates and no publication budget. Aim elsewhere if the contribution is a straightforward application of an existing network to a new dataset with marginal gains, if the visual-media element is incidental to a general machine-learning claim (a machine-learning venue fits better), or if the work is a graphics technique with no evaluation the community can inspect. As with every selective journal, a desk decision on fit is a real possibility; a short, honest pre-submission reading of recent CVM issues is the cheapest way to predict it.

Risks, limits and verification checklist

• Never quote the stored Impact Factor of 17.3 as current — the publisher's SciOpen display in hand is 8.3 (2025). Cite year and source for any figure.
• Confirm the no-APC and CC BY position in DOAJ / SciOpen at submission; sponsor arrangements can change.
• Single-blind review: author identities are visible to reviewers, per publisher descriptions — prepare the manuscript accordingly.
• Check the current editorial board and author instructions on SciOpen / Springer before submission; this profile summarises scope, not formatting rules.
• The stored record's country (United States) and numeric publisher field are wrong for this journal — the publisher is Tsinghua University Press, China, confirmed by DOAJ.

Frequently asked questions

Does the journal charge authors? No. The DOAJ record states it has no article-processing charge, and publisher descriptions state publication costs are covered by Tsinghua University. Verify at submission, as arrangements can change.

Is it fully open access? Yes — a fully open-access journal under CC BY, listed in DOAJ (verified by ISSN query for this profile).

Who publishes it? Tsinghua University Press, co-published with Springer, presented on SciOpen. It was founded in 2015 by Tsinghua's Department of Computer Science and Technology.

Who is the Editor-in-Chief? Shi-Min Hu, Tsinghua University, per the SciOpen journal page.

What is its impact factor? The publisher's SciOpen page displays 8.3 (2025) with CiteScore 28.6 (2025). The 17.3 stored on this record appears to preserve an earlier year's aggregator figure and should not be cited as current.

How often does it publish? Quarterly — four issues per year.

Sources and method

Anchored to the saved record on this site (post 1272, ISSNs 2096-0433 / 2096-0662, H-index 36, Impact Factor 17.3 as stored, publisher stored as numeric '43' and country as United States — both conflicts flagged) and verified against primary sources retrieved for this profile: the DOAJ record via the DOAJ search API (publisher Tsinghua University Press, country CN, CC BY, no APC), the SciOpen journal page (Editor-in-Chief Shi-Min Hu, Impact Factor 8.3 (2025), CiteScore 28.6 (2025), decision / acceptance / publication day figures, recent contents) and the Tsinghua University Press account of the journal's founding, editorship, SCIE inclusion in April 2021 and database coverage. Springer / publisher scope wording supplied the topic list, single-blind model, quarterly cadence and the sponsor-covers-costs explanation. Conflicting metric values across sources are reported side by side in the text, not resolved by assertion.

Publishing here alongside the conference culture

Visual computing is a conference-led field: much of its fastest work appears first at venues such as the Computational Visual Media conference, SIGGRAPH and SIGGRAPH Asia, Pacific Graphics, Eurographics events and the major computer-vision conferences. A journal like Computational Visual Media occupies a specific place in that ecology, and authors will use it best if they understand the division of labour. The journal version is where a method receives its full treatment — complete derivations, broader comparisons, ablations, failure cases and implementation detail that a conference page limit excludes — and where review articles can consolidate a fast-moving area, such as 3D point-cloud pre-training or AI-driven CAD generation, for readers who cannot follow every conference cycle. The founding issue's origin as a conference-linked special issue, and the continuing presence of special-issue activity around the conference community in related graphics titles, show that the journal was designed to complement, not replace, that conference circuit.

That positioning creates two practical duties for authors. The first is honest extension: a journal submission that grows out of the authors' own conference paper must add substantial new material and must cite and clearly delineate the earlier version, following the publisher's and the field's norms on prior publication — editors in this community will know the conference literature, often personally. The second is archival completeness: because the journal article may become the version of record that later researchers implement from, code, data and evaluation protocols should be described to a standard a conference paper rarely reaches. A result that cannot be re-implemented from the journal version wastes the very advantage the format offers.

For readers, especially those entering visual computing from adjacent fields such as machine learning, robotics or design, the journal offers a curated, open-access doorway that conference proceedings — scattered, paywalled in places, and uneven in exposition — do not. Because every article is freely readable under CC BY at no cost to author or reader, a supervisor can assign a CVM survey to a whole class, a practitioner outside a university can read the primary source rather than a summary, and a researcher in a low-income institution faces no access wall. That openness is a deliberate institutional choice, paid for by Tsinghua University's sponsorship as the DOAJ record and publisher descriptions confirm, and it is the feature most likely to compound the journal's influence over the next decade: in a field that moves this fast, the paper that can be read is the paper that gets built on.