Rusty Shackleford wrote:or maybe a thread on using 3D programs - although with AI is there any space for us anymore?
I agree with ghostofmyeyes: we often hear the cliche that [insert innovation here] didn't replace [old art form] - if anything, it heightens the value of original artwork. AI makes a lot of things easier and allows creators to fill in gaps that they otherwise would pour a lot of wasted time into. But it doesn't completely replace original work, and people who make those creations are more valued because they can.
But it does pose a huge challenge: those who learn AI tools will GREATLY lift the skill floor for creations.
There's one category that my AI work almost entirely makes redundant: photomanipulations. That's because AI models basically are photo-manips but better - to an extent. What passed as a "good" manip was basically find a good source image, maybe blend the face with a different body, then blend it into a quicksand background. A lot of the QS community went mad with basic Photoshop tools back then. Now, looking at the most popular threads in the Photomanip forum and it's [celebrity] in quicksand, and AI can easily do that, either through raw image generation if the celeb is well known enough, or through photomanip techniques but run through AI. And a prolific AI creator can churn out hundreds of these faster than a photomanip creator. So if someone really wants a hundred unique images of Chloe Grace Moretz in quicksand, give me an hour. There's no room for mediocre photomanips with AI models doing the same thing.
Similar with QS stories. To be frank - and as a former QS writer - most QS stories are also generic and mediocre. Even the ones I really liked in the past by authors like Crypto and DerMudGecko aren't written in a unique style; they focused on shaping the content to be on the things that interested them as QS fetishists. The writing itself wasn't literary greatness nor was the content unique and original. These days it's mostly fanfics of characters in quicksand scenarios. AI does this better.
A wave of anti-AI protests were about AI replacing artists. But established artists who are proud of their unique styles (like AcidTester pointed out in another thread) have no reason to be afraid - AI doesn't do what they do. It was more the average tier of artists who are already making derivative works who felt threatened that other people with less training and skill can produce superior work. But the solution isn't to blacklist AI artists. It's a challenge to make THEIR own work stand out from what AI can do.
That's why I've gone on about sixgunzloaded doing a great job with this set. Firstly, not many people are still making extended quicksand sequences using 3D modelling software. Secondly, it's done really well. Thirdly, it's something that AI can't do. I generate ONE good scene. Do enough workarounds and I might even get a consistent character face. But I can't create a 3D scene. I can't weave together a cohesive story with AI frames because I can't generate the same thing twice. And AI has huge limitations on what it can generate, such as complex scenes, multiple characters.
If anything, learning both 3D modelling and AI models can generate extremely amazing still shots since you combine the best of both worlds - the 3D environment that you choose to create plus the detailed enhancement that the AI can process. Creators who are making single frames using Poser (or heck, Second Life) will see far better stylistic or photorealistic results if they use AI -- if that is the end result they want.
But as someone who is probably at the leading edge of AI QS art - I'll say bluntly that I cannot create The Rise of Esme with AI.