Digital resources in the Social Sciences and Humanities OpenEdition Our platforms OpenEdition Books OpenEdition Journals Hypotheses Calenda Libraries OpenEdition Freemium Follow us
Selection of video game magazine covers that have been labelled as being likely to contain racy material.

Gaming Magazine Covers From Afar

by Aurelia Brandenburg and Adrian Demleitner

Gaming magazines can be simultaneously a challenging and an invaluable source for early histories of digital games and gaming. They were often produced and sold at low costs to make them affordable to their often young reader base. Since magazines like these could and would be shared with friends, it is relatively safe to assume that their actual reach exceeded the number of sold copies of an individual issue. Furthermore, they also played a crucial part in facilitating and influencing the often very homosocial nature of what was perceived as mainstream gaming culture for a long time. Gaming magazines provided readers with tips on what to buy, how to beat certain games, and who to admire among developers and journalists. These magazines offered the readers a sense of belonging within specialised communities. Understanding the homosocial aspects of the culture that, in our case, German gaming magazines represented therefore means understanding gaming magazines not only in their written contents, but also in their style.

Table of contents

  1. Gaming Magazines and their Covers
  2. Corpus, Data, and Approach
  3. Computational Approach
    1. Google Vision
    2. Automatic Clustering
  4. Reflecting the Computational Perspective
  5. Combining Qualitative and Computational Approaches
  6. Conclusion
  7. References
  8. Appendix
    1. Table 1: List of Magazines and Covers (Regular Issues Only)
    2. Table 2: Safe Search Annotation: Adult
    3. Table 3: Safe Search Annotation: Racy
    4. Table 4: Safe Search Annotation: Violence
    5. Table 5: Cluster Descriptions

Gaming Magazines and their Covers

One visual cornerstone of all magazines was  the cover. It was the selling point and the first argument to set oneself apart from competitors, especially in the 1990s when the German market for gaming magazines was truly booming.1 Covers were also an important tool of fandom, both as possible poster designs to be sold with an issue and as a way to connect magazines and their readers. For example, when Cybermedia started publishing PC Xtreme in 1996, their first issue not only tried to attract readers by promising them a report about the 100 best PC games of all times, but also advertised the magazine itself as done by veteran games journalists known to their possible reader base.

Cover of the first issue of PC Xtreme which was advertised as “The definitive game magazine by [Heinrich] Lenhardt, [Martin] Gaksch & [Winfried] Forster” on the upper left of the cover just below the magazine title (PC Xtreme 11/96)
Cover of PC Xtreme 11/96 advertising the magazine as the new magazine by Heinrich Lenhardt, Martin Gaksch and Winfried Forster with a photo of the three men below the magazine title.
This visual importance of covers led us to a couple of questions: What were readers and everyone else actually seeing on these covers? What were recurring motifs? What were the differences between individual magazines and publishers? And how did certain trends change over the course of the 1980s and 1990s?

Some of these questions can be answered through a qualitative analysis which also allows us to further develop our research hypothesis. For example, German gaming magazines of the 1980s tended to use commissioned illustrations with vague fantasy or sci-fi aesthetics as their cover images more often than they did in the 1990s—when promotional pictures became more popular and phenomena like Lara Croft-covers became more obvious. In a similar way, there is also a tendency for covers to depict images of naked or barely clothed women, often in bikini armor.This is a recurring motif that seems to have been particularly popular on covers showing fantasy-style illustrations, like Amiga Joker tended to use.

High fantasy aesthetics of half-naked women in sexualized poses were popular not only in gaming but also in other fantasy fandoms such as tabletop role playing games or fandoms of novels who were both professionally and informally connected to gaming. The illustrator responsible for many of Amiga Joker’s covers, for example, was Celâl Kandemiroğlu, a Turkish-German artist who worked as an illustrator not only for gaming and tech magazines but also for development studios such as Rainbow Arts and publishers of comics and books such as Bastei Lübbe.2 With this variety of clients, he effectively left a visual trail throughout all of these companies and the media they were selling.

Some observed aspects regarding the magazine covers are more difficult to explain or easier to overlook: While sexualized images of women are remarkable as instances of gaming magazines’ often blatant sexism at the time, other typical motifs might seem more inconspicuous. For example, video gaming magazines, in opposition to PC gaming magazines, tended to use pictures of cartoon characters quite often, which were less provocative. Knowing these magazines, this makes sense as well, since video gaming sometimes tended to be perceived more as something directed at children and magazines also had more licensed games using already popular cartoon characters to report about. With this blog post, we analysed magazine covers in a very broad sense by asking which motifs and topics can be found across a larger corpus of covers from a variety of German gaming magazines published in the 1980s and 1990s, what the resulting clusters might tell us about the style of these magazines and how it evolved.

Corpus, Data, and Approach

In the spirit of scalable viewing3, we decided to contrast our qualitative perspective with a computational approach. This allows us to observe gaming magazines and their covers from a more distanced perspective. Such an approach enables us to to sift through less prominent motifs and types of images. For that, we analysed 1089 covers of all issues of 17 different magazines published between 1983 and 2001.4 The magazines were chosen to be as representative as possible for the German gaming magazine market at the time: They include a variety of publishers, magazine types, runtimes, reach, as well as topics and target groups and span from the first commercial gaming magazine in Germany, TeleMatch in 1983, until the market’s peak with magazines such as GameStar und PC Games which are still sold and popular today.

Our computational method consisted of two different approaches, both of which make use of computer vision. We applied automatic image content labelling with Google Vision API, and a distant viewing approach through visual clusterings. Google Vision offers to produce metadata through a service that is also used for the Google Image search. That metadata consists of a rudimentary recognition of what is visible on an image, what possible topics the image covers, and automatic text recognition. We were especially interested in the Safe Search Annotation feature, which flags an image in terms of problematic content such as nudity or violence.

Visual clusterings are a simple method of organizing images by their similarity. If two images seem to depict similar subjects or cover similar topics they are placed in visual proximity. When scaled up to thousands or even millions of images, the resulting visualisation of image clouds and clusters can then be interpreted heuristically. This means that we necessarily have to ignore parts of the information a single image can hold, but it allows to speed up the process of analysis. Such a visualisation creates a mapping of proportions and relations between subjects and topics displayed on the magazine covers. Both approaches, the automatic tagging and the visual clustering, concentrates on the illustrative parts of the magazine covers. Other formal aspects such as typography, layout, and colors were left out in the analysis.

After going through the magazine covers manually, our goal with the computational approach was to have a rough categorisation on basic aspects, but also to get a fresh perspective on our material. We were interested in how computer vision would work with the magazine covers. How would it label the covers regarding adult material, and would it challenge our perception? Didit unveil some interesting trends that we haven’t observed in qualitative analysis?

Computational Approach

Selection of video game magazine covers that have been labelled as being likely to contain racy material.
Selection of video game magazine covers that have been labelled as being likely to contain racy material.

Google Vision

So far the result of the safe search annotation process5 didn’t unearth many surprises. The Google Vision API judged our video game covers slightly more conservative, but the overall trends were congruent with our observations. Of the 52 images that were “Likely” or “Very Likely” to contain “racy” material6, the magazines with the most counts are PC Action, ASM, Video Games, PC Games, and Man!ac. The rather large category of images that “Possibly” show “racy” material often contained covers in which we could clearly observe such content. This category also sometimes contains babies or other clearly non-racy depictions and we had to regard this category as very messy, to be treated with caution.

Automatic Clustering

We continued by feeding the Google Vision metadata into FiftyOne, an open source tool that helps to work with larger image datasets. Next to managing and exploring such image collections, FiftyOne also offers creating visual clusters.

View of the visual clustering of the video game magazine cover dataset with several sub-groups highlighted.
View of the visual clustering of the video game magazine cover dataset with several sub-groups highlighted.

Visual clusterings also leverage computer vision. Different approaches to setting up neural networks (AI) produce different computer vision models that then can be applied to “look” at the images. Depending on the model, different aspects are highlighted or different granularities and results are produced. This process then allows for an automatic comparison of images. If the model recognizes a cat on two images, it will deem the two images more similar to each other. Images with a cat and dog each will also be more similar than two images with a cat and a car.

 

Next to choosing a model, clustering algorithms need to be considered as well, making the endeavour of visual clustering quite complex at times. An important aspect of this approach is to reflect on the heuristic task that follows afterwards. Visual clusterings, when done right, inherently seem meaningful. But this is a simple reaction to our ability to recognize patterns in the world. We need to ask questions on why the computational approach produced certain results, and understand the technical underbelly of this method.

Reflecting the Computational Perspective

Working with artificial intelligence such as large language models or, in our case with computer vision makes it obvious that AI can’t understand anything. AI can only calculate likelihoods. That makes sense if we remember that what we nowadays call AI are actually simply powerful statistics algorithms. We can see that in the results we get from Google Vision. The labeling of our images is not certain, as in ‘yes, this contains nudity’ and ‘no, this does not contain violence’. Instead it provides us with how likely it is that an image contains nudity or violence through the categories very unlikely, unlikely, possible, likely and very likely. It is hard to discern at what point a categories chabge. Google Visions categorisation does not provide explanations for its claims, it just gives us hints into some of the directions we couldlook at.

Google Vision has not found any likely or very likely indications of pornographic (adult) or violent content on the magazine covers, and only two handful have been categorised as possibly containing such material. This can be confirmed from manually checking the covers. While the percentage of a categorisation from possible to very likely is 0.9% for adult or violent content, it is at 11.9% for racy material. At least one in every ten magazine covers featuring lewd depictions is statistically speaking significant. At this point we could say that these magazine covers have never been scandalous, whatever that might mean, but that they certainly tried to explore what is safe to publish and what not. This insight isn’t particularly surprising but rather confirms the results of qualitative research on gaming magazines in general and mirrors existing research on both editorial content and print ads, especially in the 1990s and 2000s.7 Even more: In the case of German gaming magazines, there are also obvious parallels in editorial content in the very same magazines that tended to toy with sexualized content and misogyny through phenomena like the popularity of strip poker reviews and more8.

This bar chart shows which magazine covers possibly, likely, or very likely contained racy content. Magazines without any match for these categories are omitted from this graphic. The visualisation makes evident what we had already suspected, respectively, what we already partially categorised manually. This allows us to further our discussion by highlighting that racy content is not evenly shared among all analysed magazines, but concentrates on a few. Whereas some magazines place none or only little racy content on their covers, others worked extensively with that kind of aesthetic.

Some of the classifications are also downright wrong, which we attribute to the covers being very noisy in terms of their visual elements, mixing photography with illustration and different kinds of typography. These magazine covers are in their own genre of graphic layouts, and very close to what the boulevard press does. The covers are loud, visually screaming at their readers, feature unconventional layouts regarding text, are collaged, and use bright warning colors, making feature or information extraction for computer vision erroneous at times. This raises the question how these covers have been perceived by the magazines’ clients? What kind of effect did these loud and noisy cover had on readers and potential buyers?

Moving on and looking at the visual clustering of the covers, we can get a hold on some interesting takeaways. Most likely, the design and production of these covers followed an iterative and by-inspiration paradigm, and did not undergo extensive reader and buyer reception testing. A cover was made, maybe inspired by what others were doing already, the magazine published and then improved in the next issue, if necessary. In regard to this rather non-reflectory approach to the covers, the following can shine a light on some further topics of interest.

When a cover features real persons9 they are either featured in action, as in doing things that one would expect from an action film protagonist, or with very emotional expressions, such as surprise or fear. This insight is interesting insofar this kind of depiction is a driving force behind the Youtube recommendation algorithm.10Videos that feature a preview image with this type of imagery are evaluated as more likely to drive engagement, which is an aspect that Youtube favours. Furthermore, it also fits accounts such as the one of Anatol Locker, former editor-in-chief of Bravo Screenfun, who recalled in an interview in 2020 how he and his team learned very quickly through market research that putting people on their covers and especially pictures of popular game characters worked best with their target audience.11 This also contributed a nearly absurd amount of covers depicting Lara Croft at Bravo Screenfun which in turn also fits how as a last kind of expression that can be observed is people having a lascivious gaze towards the viewer of the cover, which links us back to racy content.

There are two more minor observations of interest. Fictive characters, from cartoons or games, are divided into animal or not-animal by the visual clustering. We didn’t think much of this aspect beforehand. But the computer vision algorithm creating separate clusters for each points to a statistical significance in different displays of human or animal characters. This can be taken as a specific genre of cover design. These depictions of fictive characters are often rather cute or harmless and share a perspective on cover design and gaming culture that is diametral on the opposite of the racy content. We got aware of how much these covers are trying to appeal to boys when manually labeling the visual clusters that the automatic process created. The topics and subjects covered are straight from a teenager’s pop cultural interests.

A last observation concerns how new consoles are often staged like racing cars. They are not calm and pristine devices, like Apple would advertise them, for example. They’re also not highlighted as the complex technological devices they are. Video game magazine covers exclusively frame game console announcements through speed, which is expressed through motion blur, skewed perspectives, and flashes. They are depicted as fast and in action, similar to how race games would be advertised. This clustering wasn’t obvious to our eye, but the computer vision algorithm bundled those magazine covers together with others that display sports or action scenes, or car racing games. It wasn’t obvious to use how this trend originated, but it would certainly be an interesting lead to follow up that could tell us more about how magazines and readers perceived each other.

Combining Qualitative and Computational Approaches

Computational approaches are best taken with a grain of salt. Algorithms can’t be reasonably critical about the material they have to work with. They simply do as they’re told. A good example is the labeling of images as potential racy, when human figures show skin. Most often this works, but there are very likely times this approach fails, for example when an image of a baby is labeled as such. Computational methods, such as the mass-labeling of images, can be supportive in scalable viewing approaches. Algorithms can form part of a more holistic research methodology, in lack of a better term, in which better access to the material is created through the help of the machine. Be it researching material in languages one isn’t familiar with or having very large datasets that would escape our capabilities of working through.

In this case we were interested in how a computational approach enhances or challenges qualitative research on the magazine covers. To that end we haven’t been surprised in our application of distant viewing. The dataset is relatively small for a computational approach and we have been already working on it extensively, getting a good understanding of the magazine covers’ contents. Applying our computational approach on top of that helped in a clearer vision of the content’s distribution of categories and magazine brands, as well as creating a traceable source for our analysis. Quantitative approaches are always built on top of qualitative research or contain certain aspects of it. The main difference is that parts of the data collection in computational research is reproducible, whereas in qualitative research it is often found in the eye of the beholder.

Getting a clearer vision on the distribution of racy content in relation to the magazine brands, one observation stood out. In most cases where a woman is depicted on a cover, she is either placed in a fantasy or cyborg setting, or is carrying weapons. This is a type of othering, meaning being treated as fundamentally different as the dominant or “normal” group. The goal of depicting women this way is to strip them of their right for decency, and allows the magazines to show them in a sexualised fashion. Drawing from fantasy and scifi settings allowed the magazines to build upon bordering subcultures such as fantasy fandom, TTRPGs, board games and others, and connect to the sexism that was already common in those domains.

Simultaneously, when looking at the clusters resulting from our corpus, it is also interesting to note how the covers mirror different types of magazines and styles. The differences between video game magazines and computer game magazines seem striking, which in turn matches the overall trend of the market during the 1980s and 1990s: Video game magazines only became financially stable or even successful during the 1990s when consoles became more common and due to phenomena such as licensed games of cartoon characters, magazines with cartoon covers became relatively common as well. Regarding their contents, neither computer gaming nor video gaming magazines did differ much and in the end, both types of magazines often had a large readerbase of people under the age of 18, but their covers mirror different perceptions of which magazines were imagined to rather address children and which were imagined to rather address adults. And yet, both types or rather styles of cover design still reflect upon a larger trend to appeal to boys and men in particular.

Conclusion

Our computational approach overall confirmed what we knew based on qualitative analysis, and yet, it also broadened our understanding of the source corpus evaluated here. For one, confirming the perception that certain strands such as covers of women in sexualised poses were truly common might seem like a small insight, but given that our analysis is based on a corpus that can be considered representative of German gaming magazines until 2000 in general, it still remains an important insight. Secondly, the visual clustering highlighted a few groups that we weren’t necessarily aware of before. The cluster around gaming consoles stylized as fast cars, or airplanes and helicopters always shown mid-action point to possibly insightful further research on aesthetics and reader preferences. And thirdly, while our analysis here can merely count as a relatively small test, it still also hints at some interesting questions to pursue with similar corpora. Would there be significant differences in styles when using a larger, international corpus of gaming magazine covers? For example, are there remarkable differences between British, German, and French magazines that can be examined that way? Are there parallels to be detected when, for example, using the same method on digital gaming magazine covers and board game magazine covers? What about differences and parallels in print ads? Questions like these would all have exceeded our scope here, but they could offer some interesting answers to reexamine, connect and enrich qualitative approaches to the same sources.

References

Brandenburg, Aurelia. “‘On the other hand, the girls are quite discreet’: Strip-Poker-Reviews in early German Gaming Magazines”. Confoederatio Ludens, 18.11.2024, https://doi.org/10.58079/12p9o.

Brandenburg, Aurelia. “Print runs of German Gaming Magazines 1980-2000”. Confoederatio Ludens, 23.10.2023, https://doi.org/10.58079/mr91.

Condis, Megan and Jess Morrissette. “Dudes, Boobs, and GameCubes: Video Game Advertising Enters Adolescence”. Media, Culture & Society 45, no. 6 (2023), 1285–1302.

GamesWirtschaft. “Celâl Kandemiroğlu: Games-Illustrator verstorben”. Gameswirtschaft.de, 18.03.2022, https://www.gameswirtschaft.de/nachrichten/celal-kandemiroglu-verstorben/.

Kirkpatrick, Graeme. “How Gaming Became Sexist: A Study of UK Gaming Magazines 1981–1995”, Media, Culture & Society 39, no. 4 (2017), 453–468.

Pause, Johannes and Niels-Oliver Walkowski. “Scalable Viewing”. Open-Media-Studies-Blog, 11.03.2019, https://zfmedienwissenschaft.de/online/scalable-viewing.

Plass-Fleßenkämper, Benedikt and Sönke Siemens. “Making Mags: Bravo Screenfun”. Stay Forever, 07.06.2020, https://steadyhq.com/de/stayforever/posts/b07198ec-0a84-410e-9b9f-7bde2c897bc7 / https://www.spielejournalist.de/2020/06/07/bravo-screenfun-anatol-locker/.

Appendix

Table 1: List of Magazines and Covers (Regular Issues Only)

Magazine Issues Total number of covers included Comment
Amiga Games 10/92-11/96 48 Complete
Amiga Joker 11/89-10/96 70 Complete
ASM 03/86-02/95 94 Complete
Gamers 01/92-12/95 31 Complete
GameStar 09/97-12/00 33 Complete within timeframe
Joystick 09/88-09/89 11 Complete
Man!ac 11/93-12/00 86 Complete within timeframe
PC Games 10/92-12/01 111 Complete within timeframe
PC Joker 06/91-01/01 90 Complete
PC Spiel 03/95-12/96 22 Complete
PC Xtreme 11/96-04/97 6 Complete
PC Action 02/96-12/00 59 Complete within timeframe
PC Player 01/93-07/01 102 Complete
Play Time 05/91-08/95 50 Complete
Power Play 12/87-04/00 145 Complete
TeleMatch 01/83-02/85 19 Complete
Video Games 01/91-03/01 112 Complete

Table 2: Safe Search Annotation: Adult

Adult Count
VeryUnlikely 890
Unlikely 189
Possible 10

Represents the adult content likelihood for the image. Adult content may contain elements such as nudity, pornographic images or cartoons, or sexual activities.

Table 3: Safe Search Annotation: Racy

Racy Count
VeryUnlikely 627
Unlikely 332
Possible 78
Likely 35
VeryLikely 17

Likelihood that the request image contains racy content. Racy content may include (but is not limited to) skimpy or sheer clothing, strategically covered nudity, lewd or provocative poses, or close-ups of sensitive body areas.

Table 4: Safe Search Annotation: Violence

Violence Count
VeryUnlikely 421
Unlikely 658
Possible 10

Likelihood that this image contains violent content. Violent content may include death, serious harm, or injury to individuals or groups of individuals.

Table 5: Cluster Descriptions

ID Description (manual) Count % of Total
0 Various (unclear cluster) 95 8.72
1 Androids, Cyborgs, People in Space Suits or Scifi-Armor, Battle-Mechs 83 7.62
2 Portraits showing Skin, Eroticised Gazes 73 6.70
3 Fantasy, Middle Age, Swords, People in Armor, Horses 91 8.36
4 Closeup Portraits, Youtube Face 80 7.35
5 Racing, Racing Cars 47 4.32
6 Characters with Guns and Weapons 81 7.44
7 Helicopters, Tanks, Spaceships, Heavy Artillery 50 4.59
8 Various (probably clustered through large round visual object on cover) 39 3.58
9 People Cartoon (and Video Game) Characters 57 5.23
10 Airborne Vehicles mid-flight 27 2.48
11 Football 21 1.93
12 Monsters, Dragons, Dinosaurs, Demons, Aliens 109 10.01
13 Space, Spaceships, Scifi 77 7.07
14 Animal Cartoon (and Video Game) Characters 55 5.05
15 (Video Game) Characters in unspecified Action Scene 104 9.55

  1. Brandenburg, Aurelia. “Print runs of German Gaming Magazines 1980-2000”. Confoederatio Ludens, 23.10.2023, https://doi.org/10.58079/mr91. []
  2. GamesWirtschaft. “Celâl Kandemiroğlu: Games-Illustrator verstorben”. Gameswirtschaft.de, 18.03.2022, https://www.gameswirtschaft.de/nachrichten/celal-kandemiroglu-verstorben/. []
  3. Pause, Johannes and Niels-Oliver Walkowski. “Scalable Viewing”. Open-Media-Studies-Blog, 11.03.2019, https://zfmedienwissenschaft.de/online/scalable-viewing. []
  4. Namely Amiga Games, Amiga Joker, ASM, Gamers, GameStar, Joystick, Man!ac, PC Games, PC Joker, PC Spiel, PC Xtreme, PC Action, PC Player, Play Time, Power Play, TeleMatch and Video Games. See also Table 1 of the appendix. []
  5. The technical setup to access the Google Vision API is relatively simple, and we leveraged the open source program Memespector-GUI. It allowed us to upload our video game cover images and in turn received a data sheet in return, containing all the metadata and the safe search annotation labels. An extract of this data can be found in the appendix. []
  6. Google Vision defines racy as ‘content [that] may include (but is not limited to) skimpy or sheer clothing, strategically covered nudity, lewd or provocative poses, or close-ups of sensitive body areas.’ []
  7. See, for example Kirkpatrick, Graeme. “How Gaming Became Sexist: A Study of UK Gaming Magazines 1981–1995”, Media, Culture & Society 39, no. 4 (2017), 453–468, Condis, Megan and Jess Morrissette. “Dudes, Boobs, and GameCubes: Video Game Advertising Enters Adolescence”. Media, Culture & Society 45, no. 6 (2023), 1285–1302. []
  8. See, for example, Brandenburg, Aurelia. “‘On the other hand, the girls are quite discreet’: Strip-Poker-Reviews in early German Gaming Magazines”. Confoederatio Ludens, 18.11.2024, https://doi.org/10.58079/12p9o. []
  9. See appendix for cluster sizes. []
  10. See https://allscience.substack.com/p/on-the-grim-reality-of-youtube-face. []
  11. Plass-Fleßenkämper, Benedikt and Sönke Siemens. “Making Mags: Bravo Screenfun”. Stay Forever, 07.06.2020, 00:36:55-00:37:10, https://www.spielejournalist.de/2020/06/07/bravo-screenfun-anatol-locker/. []

OpenEdition suggests that you cite this post as follows:
Aurelia Brandenburg, Adrian Demleitner (December 8, 2025). Gaming Magazine Covers From Afar. Confoederatio Ludens. Retrieved January 22, 2026 from https://doi.org/10.58079/15aok


2 thoughts on “Gaming Magazine Covers From Afar”

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.