Openness, AI, and the changing creative landscape

If you’ve been following the AI scene, you may have seen a video by Corridor Crew entitled Anime Rock, Paper, Scissors. The video is a short animation telling the story of twin brothers who want to gain control of an unnamed kingdom, and one challenges the other to the Holy Combat of Rock, Paper, Cross-blades. Hilarity ensues.

What makes this a remarkable exercise is that the animation was made with artificial intelligence tools, particularly Stable Diffusion. While the voices, music, and acting are all human, the animation was achieved by training an SD model using stills from Vampire Hunter D Bloodlust, and using other SD tools to superimpose the images to the actors.

If you were to follow only Twitter, you would be forgiven for thinking that the response has been largely negative, the replies to the announcement were often scathing, coming from a range of animators and artists; in general the critics and praisers on Twitter fell into the two camps that have been debating these issues for months, those already in favour of AI tools loved it, while AI detractors hated it. The response outside of the Hellsite has been more positive, the comments on YouTube have been largely praiseworthy (I don’t know if there’s some moderation going on). Also, at the time of writing the video has received 1.8 million views, with 206k likes and 21k dislikes, practically a ratio of 9 out of 10 likes v dislikes. As YouTuber Asmongold said, “I thought it was going to be cringe, but this is f*cking amazing!”

Personally I loved the concept, the story, the voices, the music, it’s all great for an amateur animation short, with great quality in almost every department, perhaps except a few details that feel off. The lip-syncing is not the best, there are a few inconsistencies in lighting, but for the most part the animation is quite decent.

Why is there opposition? Some people have argued that the training itself is stealing the artwork from Vampire Hunter D, while others object to the very use of SD, as it was trained on the works of artists without authorisation. There has been some opposition from people in the industry calling the use of these tools dangerous and irresponsible.

I won’t go into the legal issues, I think that this is a very straightforward case of fair use, but also there is the practical issue of a low likelihood of enforcement from Bloodlust’s producers. It would be counter-productive to sue over style and training using a relatively old anime, such litigation is not cheap, and the YouTubers aren’t a juicy target anyway, the damages would not be worth it in my view. But that is not the reason why I’m writing this blog post.

What is interesting is that Anime Rock, Paper Scissors is the precursor of possible changes in the media landscape, heralding a new age in user-generated content. For a long time, the rise of UGC has meant that there is a new class of creators that take advantage of technology to reach audiences in a way that was impossible in the analogue world. What has been interesting to witness from a copyright perspective, is that for a generation of content creators, copyright is not something that bothers them greatly, their only interaction with it has been through platform enforcement, so for them copyright law has become whatever YouTube, Instagram, Twitch and TikTok say it is. Infringement is seen through the lens of not falling foul of whatever rules the platforms have, while copyright ownership is also less important in a world of subscriptions, endorsements, and merchandise. Sure, there will be some people who will still benefit from copyright, but for the most part an entire ecosystem is growing without any thought towards copyright.

Anime Rock, Paper Scissors is interesting, because it could be used as a case study of copyright ownership, would Corridor Crew apply for copyright protection? Would they get it? If a 2 month production filled with writing, acting, special effects, sound effects, and music doesn’t deserve protection, what does? To concentrate on the AI element would be ridiculous. But the most important question is, why would Corridor even bother with copyright protection? It’s not their business model. Copyright belongs to a different era.

Here is where another very important part of the story should be highlighted. During the “Making Of” video, Niko Pueringer comments that all of the tools that they’ve used were open, which was precisely the strategy from Stability.ai when it released all of its models to the public under open source licences. He says that the tools are free to use, they can be downloaded by everyone, so this means a “democratization” of the creative process. As a result, they also give step-by-step instructions on how to replicate the video (to those who pay for a subscription of course). Share with the community. It’s refreshing to see a return to open source ideals, sharing, and openness, and while many will be skeptical of this approach, I think that it follows a tradition of openness, the remix culture that was in vogue during the 2000s.

Copyright will survive, but for many people it is irrelevant, and new business models will rise up to meet the demands of a different type of creative process. AI is just a part of a trend that has been changing creativity, at the centre of it is a new generation of content creators.

On the meantime, I’ll go and find my old DVDs of Vampire Hunter D, I had forgotten how much I love that style of anime.