I mean, once you’ve either paid for or logged into a remote service or set up the hardware and software locally (OP is using Stable Diffusion, so he’s probably running it locally on his PC, or possibly some system he controls in a datacenter, so no censorship or usage limits), there’s not that much work on a per-image basis. All he’s doing with these is plugging in your username.
The amount of setup required has also been on a decline as people package more stuff. I expect that the barrier to use will steadily decline.
Now, there are more-advanced things that require more familiarity and practice, like inpainting or regional prompts, or setting up directed acyclic graphs that generate images, as with the ComfyUI frontend to Stable Diffusion. Those give you more control over the output or make it easy to run complex processes with small amounts of tweaking to various stages in the process. There will, I am sure, be more-powerful and more-complex tools that will come out targeting professional artists. OP is also using some nonstandard settings, and a nonstandard model, so he has done some additional one-off configuration on his system. But doing what OP is doing here, once the software is set up on compatible hardware, is not complicated on a per-image basis.
This is all really cool. Wished I had the time to learn to do this myself.
Clueless lemmy ng Steps: 20, Sampler: DPM++ 3M SDE Karras, CFG scale: 7, Seed: 2213099523, Size: 768x1280, Model hash: 74dda471cc, Model: realvisxlV20_v20Bakedvae, Clip skip: 2, RNG: CPU, Version: v1.6.0
Ha! Thank you! Dude definitely looks loopy
I mean, once you’ve either paid for or logged into a remote service or set up the hardware and software locally (OP is using Stable Diffusion, so he’s probably running it locally on his PC, or possibly some system he controls in a datacenter, so no censorship or usage limits), there’s not that much work on a per-image basis. All he’s doing with these is plugging in your username.
The amount of setup required has also been on a decline as people package more stuff. I expect that the barrier to use will steadily decline.
Now, there are more-advanced things that require more familiarity and practice, like inpainting or regional prompts, or setting up directed acyclic graphs that generate images, as with the ComfyUI frontend to Stable Diffusion. Those give you more control over the output or make it easy to run complex processes with small amounts of tweaking to various stages in the process. There will, I am sure, be more-powerful and more-complex tools that will come out targeting professional artists. OP is also using some nonstandard settings, and a nonstandard model, so he has done some additional one-off configuration on his system. But doing what OP is doing here, once the software is set up on compatible hardware, is not complicated on a per-image basis.