GrobusDobus
Newbie
- Dec 11, 2019
- 15
- 73
- 112
The thing is, once all the ethical concerns surrounding AI usage (if they ever do) get ironed out, and the people who deserve to get paid for their art being stolen to train the models get paid, I've got zero problems with a hypothetical future in which these tools advance to the point where they're invisible, and merely exist to aid in the actual creative process. I don't want AI doing the coloring, and animating, and voice-over, but I'm fine with it assisting the people who are actually doing those things.I see some misunderstandings of what the purpose is of AI with regards to animation, let me help clarify.
In traditional 2D/3D animation, animators set "golden poses" and animate the keyframes in between. For example, when animating a person standing up from a chair, the golden poses would be the character sitting down and then standing up. The computer helps interpret the motion in between the sitting down and standing up pose, but the animator has to do the bulk of the work to make the motion look seamless and natural. A tedious process. Now, with the way AI is progressing, AI can take care of a lot of this grunt work, so artists are free to focus on the emphasized parts of the movements, saving a lot of time and effort, and they can focus on the parts that are more important in the animation. This technique is being done in 2D animations like that new Sekiro show that's coming out soon, and major studios like Pixar are experimenting with it, you can pretty much guarantee they will use AI in their workflow in some capacity for their future movies. Its purpose right now in industry is to speed up an animator's workflow, that's it.
Now Miro, being the lazy bugger that he is, wants the AI to do all the work for him. In fact looking at that trailer, it is just hashed together AI generated video segments. He's probably using WAN 2.1 txtToImg by how how generic the characters look, doesn't even look pretrained on anything. For those unaware, this AI model can generate only a few seconds of video at a time, and you can splice the end-frame of each video to the start of the new one to give the appearance that it is one long video, which is what Miro is doing. It's pretty obvious when you pay attention to the constant background changes.
You can also train this AI model to generate certain characters, called Loras (e.g.You must be registered to see the linksis trained on Witcher 4 trailer data), so you can prompt the AI to generate a character with that likeness, but Miro's characters does not look pretrained on anything. It literally looks like he downloaded default WAN 2.1 AI model, typed something, spliced some things together, and called it a day. How this guy still has patreons is beyond me
We just aren't there yet.
There are plenty of other artists out there besides Miro who have done brief little experiments using AI to crank out comics or stills or whatever to supplement their actual work, and just like Miro's AI garbage, it looked awful. It was completely devoid of charm, and I thank fuck that these creators listened to the feedback of their fans, which is that a slower workflow (with breaks as needed to avoid burnout) is preferable to wasting time on AI art that sucks.