3 ms·

And that's why laymen with prompts will never replace trained artists/designers for anything but trivial use cases— the ability to decide a) how best to visual

And that's why laymen with prompts will never replace trained artists/designers for anything but trivial use cases— the ability to decide a) how best to visually communicate the right thing to the right audience, b) whether generative AI is the best choice to achieve that and the capability to use more precise tools if not, and c) that what you have is cruddy enough, or the message is superfluous enough, that not using an image is more effective. This image fails and this is a trivial use case. While it's depressing to see so many lower-end commercial artists lose their livelihood and their depressing wages on the market upstream, I can't help but have a little schadenfreude seeing things like this after so many tech people with dunning-krueger-informed confidence about visual communication have gleefully called me a buggy whip manufacturer.

stale2002 2y ago
> why laymen with prompts Thats the baby mode stuff dude. You have missed the forest for the trees if the only thing that you can engage with, is the simplest, most low effort usecase for AI. There are so many other things that can be done with AI. The immediate, obvious stuff would be anything to do with Video To Video AI generation. IE, imagine someone shoots a video the normal way. With all the creative input that this involves. And then you take that video, and you change things in it, using AI. I don't know, you see the video and you realize that you need to add an additional lightsource in, and you want all the shadows to autocorrect. You could use AI to do that. Thats just a random/intermediate usecase off the top of my head, that involves a lot more creative input than just "prompt in, video out". I am sure that there could be a lot more crazy usecases, but you aren't going to be able to see then, because you are instinctively losing your mind by only talking about the dumbest and most easy usecase of prompt engineering.
- chefandy 2y ago
  I'm a technical artist so I'm pretty familiar with professional film and game production pipelines, and how generative AI is used within them. The genuinely useful tools for NN in production workflows will look like Nuke's copycat tool. Being very familiar with the current state of these tools, I'd be shocked if something like lighting could be controlled anywhere close to precisely enough for real professional work in the foreseeable future. Even with a full 3D scene, forward-thinking tools like Octane can only muster up modest optimizations with NN lighting processing. Starting from a 2D image? Maybe for a real estate listing, online product photo, or personal media, but that's about it. Prompts are just fundamentally one of the least useful interfaces for precise work, and most people involved just don't realize how unforgiving the requirements are for high-end media production.

stale2002 2y ago

> why laymen with prompts Thats the baby mode stuff dude. You have missed the forest for the trees if the only thing that you can engage with, is the simplest, most low effort usecase for AI. There are so many other things that can be done with AI. The immediate, obvious stuff would be anything to do with Video To Video AI generation. IE, imagine someone shoots a video the normal way. With all the creative input that this involves. And then you take that video, and you change things in it, using AI. I don't know, you see the video and you realize that you need to add an additional lightsource in, and you want all the shadows to autocorrect. You could use AI to do that. Thats just a random/intermediate usecase off the top of my head, that involves a lot more creative input than just "prompt in, video out". I am sure that there could be a lot more crazy usecases, but you aren't going to be able to see then, because you are instinctively losing your mind by only talking about the dumbest and most easy usecase of prompt engineering.

chefandy 2y ago
I'm a technical artist so I'm pretty familiar with professional film and game production pipelines, and how generative AI is used within them. The genuinely useful tools for NN in production workflows will look like Nuke's copycat tool. Being very familiar with the current state of these tools, I'd be shocked if something like lighting could be controlled anywhere close to precisely enough for real professional work in the foreseeable future. Even with a full 3D scene, forward-thinking tools like Octane can only muster up modest optimizations with NN lighting processing. Starting from a 2D image? Maybe for a real estate listing, online product photo, or personal media, but that's about it. Prompts are just fundamentally one of the least useful interfaces for precise work, and most people involved just don't realize how unforgiving the requirements are for high-end media production.