Wait… There IS an Animate Button!?

AI

Image created in Stable Diffusion

A very common situation for an animator when dealing with a client that doesn’t really know anything about what you do is that your job is easy. After all, it’s just a drawing, or a cartoon, or whatever. It’s pretty normal to have misconceptions about fields you aren’t well-versed in. It makes it difficult sometimes to convince a client that it required time and money, when they usually don’t have either.

A originally wrote about this situation here, which might be a little dated but may still be worth checking out. In it I explain that there is no such thing as the “Animate Button” which will do all the work for us. It takes a lot of time, knowledge, skill, technology, and even math to do it right. Doesn’t it?

With the advent of AI technologies like ChatGPT we’ve seen a large influx of Artificial Intelligence that can do just about anything you can imagine. It was only 6 or 7 years ago that Google was showing off its Deep Dream nightmare fuel technology that could tap into your darkest fears and generate a psychedelic drug-induced image of it. Ok, so it was probably worse than that. I’m pretty sure I’ve never been scared of seal-faced horsemen with a Doberman butt and a cat head hand riding a horse with a sloth for an ass. But now I do.

Well, in these few short years, we’ve come a LOOOONG way from that. Thanks to new AI technologies, anyone can access Stable Diffusion, or any other number of AI services, some free and some not. With a simple description of what you want, you too can get images like the one I posted above in mere seconds. Yes, I went to a free AI image generator and typed in that I wanted a picture of an anime girl smiling in the sunset. This is what it gave me. Impressive is an understatement.

Is it perfect? Of course not. These typically mess up hands, can’t always figure out arms, have a terrible time with eyes, and generate all sorts of artifacts like multiple limbs. As an AI, it has no actual understanding of art, style, or aesthetics. All it can do is use existing art to recreate objects, styles, and moods. It has to be trained to do this, which has caused a lot of controversy over how it is trained and with what. It has become a seriously dangerous tool if used improperly, capable of generating adult content of real people without them consenting, or even knowing about it, at all.

They also have other legal gray areas to sort out. Can they be copyrighted? Who owns any rights to the content? Is it plagiarism? How many fingers are too many on a hand?

As far as the impact on people’s livelihoods goes, it is likely to cut into the profits of artists. Many artists have even found their work used to train these AI in their style, which is upsetting, to say the least. Many of the big artist communities have outright banned posting AI-generated art on their platforms, but that will do little to hinder the development of the technology and the way people use it. Why pay a talented artist to draw you a character when an AI can do it well enough for free, and in the same style?

Of course, this is just for illustration, what about video?

So far, AI-generated videos have been infamously bad. Every frame has a different look, the lighting changes, the eyes morph and distort, and it just looks bad. Earlier this year, Linkin Park released a new music video after the 2017 passing of their Lead Vocalist, Chester Bennington. They decided to create a video with the same kind of gritty anime style they used for the music video to “Breaking the Habit.” Unfortunately, either due to an attempt to save money, appeal to the younger crowd, or push some ill-defined vision, they decided to render the entire thing with an AI. And… oh my god do I hate it.

I know plenty of people that loved it but try as I may, I can’t even tolerate it. It looks like they decided to give up. To me, it is just plain ugly. In order to hide the fact that AI-generated art isn’t really ready for the mainstream yet, especially for videos, they leaned into the flaws of the technology. Typically, an artist will use the tools at hand to define their art, not allow the flaws and limitations of that medium to do it for them. And that is what happened here. The wonky eyes, the misproportioned features, the displaced mouths, this video relishes in the failures of AI-generated art, rather than pushing the technology to show what it can do.

But that is where a tiny youtube channel called Corridor Crew stepped in. They decided to do something similar and create an entire anime using AI. But they didn’t leave it there. They took the technology, and tweaked it, adjusted it, manipulated how it worked, and used all sorts of tricks to fix the most blaring problems. And it (mostly) worked!

Well, that is honestly up for debate.

A lot of people hated it more than I hated the Linkin Park video. And I don’t blame them. There are ethical reasons on many levels to shy away from this. To start, they used screenshots of an existing movie, which was based on a specific style. While the legality is still being figured out, ethically it is theft, even if you don’t use the actual image itself in the final production. But another argument can be made that humans do the exact same thing. Art can not be created in a vacuum, we must be exposed to art, culture, style, and life, and take influence from those things. This is a more direct way of doing that, which is where the gray area rests. How far is too far?

Despite the criticisms and ethical quandaries, they were able to solve a lot of the problems AI-generated art has and produced something that was ridiculously good for the amount of effort put into it. No, it isn’t perfect. Again, there are a lot of flaws, like shifting eyes, flickering lights, and other artifacts. But it is a brilliant proof of concept that it can be done. They did have to film the entire thing, in costume, and they put a lot of work into the backgrounds and effects and every aspect of it. This made it more of an AI-Human collaboration than a straight-up AI-generated video.

And this is one of the ways AI just might work for the film and television production industry.

There will ALWAYS be artists that want to draw, want to animate, want to film, produce, edit, etc. That will never go away. And even if AI develops to a point where it can produce results as good as us, we will still be driven to keep creating. It is likely jobs will be lost due to this technology. Smaller companies may turn to AI to develop entire videos, or even channels of content because it is cheaper and easier and requires little to no skill. But, just like with the invention of the PC, the tool isn’t going to go away, and neither will the user. What we need to do is learn to adapt to use AI tools as part of our repertoire. It would be easier for teams to get more artistic options for previz work, or apply an effect on top of a film. It may speed up the process of rendering, or animating, or rigging. But it will never wake up one day and say, I want to create!

That’s what people do.

So it is time to wake up, stop dreading or ignoring the impending doom of AI, and embrace it. Learn how to use the technology. Learn how to fit it into your pipeline. While it may not be for everyone, someday people will look back on this time as we look back at the Industrial Revolution. We still have all the ethical and legal questions to sort through, but they will be sorted through. AI is not slowing down. If anything, these last 2 years have shown me it is speeding up. Fast.

There is still no “Animate Button”. Not really.

There are tools that can produce shoddy and unsatisfying work with a click of a button, but as they say, you either get something good, fast, or cheap, but not all three. If you want to press that animate button, it’s going to be fast, and cheap, but ugly. The only way, at least right now, that it can really be helpful in production is for more of a behind-the-scenes approach. Using it to generate character ideas or mood boards. At least for now. That is changing, and we must change with it.

But, AI technology is advancing. So keep your eyes and minds open, and advance with it.

Previous
Previous

Me, Myself, and AI

Next
Next

Where Does Virtual Reality Really Belong In Video Production?