Aggression in Pornography is a very interesting research paper that aims to add empirical evidence to the divisive conversation about the harmfulness of pornography by analyzing a collection of freely available pornography on a popular website.
The authors claim that most previous research on the matter comes from a place of predisposition and also that many fail to acknowledge important factors like viewer reception and non-hetero pairings. Also, that previous research done in a somewhat similar fashion worked with a now outdated media landscape like analyzing rental tapes.

The authors combine quantitative information based on analyzed media with qualitative information from interviews with pornography consumers. Although the sample size of 409 videos (taken from Pornhub's "most popular" and "random" categories) and 122 interviewed people isn't very large, it still made for some interesting findings that often contradicted popular myths and beliefs about the subject.
For example, the quantitative results showcase a lack of evidence that popular pornography has gotten "harder" over time, data that same-sex pornography had relatively higher amounts of aggression than hetero pornography, and that pornography starring certain combinations of ethnic performers are coded with more aggression than others. (I also find it interesting to note here that this research was done before Pornhub started only allowing verified professionals to upload to their site. At the time of this research, Pornhub was probably the biggest intersection of popularity and a lack of moderation on the market, which probably made this a prime source of this.)
The qualitative information based on the interviews showed for example that women were statistically a lot more likely to enjoy aggressive pornography than men, that regular viewers are very much able to distinguish fantasy and their actual sex life so that for most one doesn't influence the other, and that most consumers value consent and pleasure a lot more than aggression, even in the context of BDSM acts.

While this research was not comprehensive enough to be able to confidently generalize the findings on a media-wide scale, it does give some very interesting insight and makes some good points or raises some good questions. It would be great to see similar research done on a larger scale.