Reposting here, as the last thread quickly filled with images:
Those of you taking notes or making structured tagging systems, artists, etc. What are y'all using? I've been making notes per qualifier and artist that I've been using but I'm looking to move said notes to a system where I can recall partial prompts according to their tag (e.g. "dark colors", "scifi", or "concept artist")
Any advice for programs or software? I'd rather make my own and not be limited with the "SD Artist collection" notion page.
Ryan Adams
adding film grain to give the image a bit more texture
>((Cyberpunk)) back alley with brutalist architecture, vibrant colors, Marc Simonetti, Wenjun Lin
Minimal difference between the artists. Wenjun Lin makes things a bit floaty, not sure I appreciate it here. Tom Bagshaw pops nicely across all colour mods. Aldred Kubin adds sharpness to corners. Wadim Kashin somehow feels more lived-in, though I can't put my finger on why.
The various colour mods all seem pretty impactful, and all in exactly the ways I expected. I've had trouble getting warm and cool to work in the past, but they seem sufficiently impactful here, if a bit subtle. Pastel remains a personal favourite.
Can someone recommend good prompts for game concept art? I want to see a POV shot in the back of a troop transport, like you're a soldier being transported, but try as I might I can't get it from inside the truck looking out. Nor is the style particularly good. >inside troop transport personnel carrier, scenery showing road and traffic and trees and sky, soldier sitting in the truck in military gear wearing a plate carrier and holding a rifle, charcoal sketch concept art storyboard, by Greg Rutkowski and Marc Simonetti, trending on Artstation, 4k f/11 HDR
>generate a picture while a game is running >tab out >generation stops >tab back into game >generation continues >tab out again >generation stops What the fuck
Oliver Williams
could someone enlighten me as to why my medvram isn't working. I'm wanting to generate bigger images with 8gb vram
I'm tracking shit in Obsidian, but it's kinda garbage for images.
Angel Price
If the AI is unwilling to generate something you want with txt2img you need to sketch up a loose concept in drawing software, blocking things in with color (high contrast).
I'm modifying the Krita plug-in right now to support a remote server, since my client is Linux and server w/ my 3090 Windows. But I'll try that once it's working. Already have txt2img work, just need img2img and upscaling
Christian Peterson
how do i install the waifu diffusion so that I can launch it and the other model without having to change names and shit