Yeah it’s pretty good for “close enough” stuff but every stupid image I could make wastes so much energy that I prefer to just do an image search and steal some real human-made images. From an art theft perspective it may as well be the same thing anyway, and if it’s just for a silly game between me and my pals who gives a shit.
That’s more or less how I feel nowadays. When I really need an NPC image, I just find a deviantArt image or something now. But I honestly eschew images entirely now and just go for proper descriptions so people can create them in their mind’s eye.
Not even close enough if you have a mind for detail. I’m running a weird campaign rn (cyberpunk red ported to 5e) and I really want to nail the visual vibe of things, somewhere close to Jean Giraud. AI can’t do perspective for tokens to save its skin, and cannot keep a style consistent no matter how you try.
Ended up having my players do their characters on heroforge, sending me a screenshot, and then me tracing over it and putting my spin on it. More work for me, but I just can’t get what I want from AI.
Yeah it’s pretty good for “close enough” stuff but every stupid image I could make wastes so much energy that I prefer to just do an image search and steal some real human-made images. From an art theft perspective it may as well be the same thing anyway, and if it’s just for a silly game between me and my pals who gives a shit.
That’s more or less how I feel nowadays. When I really need an NPC image, I just find a deviantArt image or something now. But I honestly eschew images entirely now and just go for proper descriptions so people can create them in their mind’s eye.
Not even close enough if you have a mind for detail. I’m running a weird campaign rn (cyberpunk red ported to 5e) and I really want to nail the visual vibe of things, somewhere close to Jean Giraud. AI can’t do perspective for tokens to save its skin, and cannot keep a style consistent no matter how you try.
Ended up having my players do their characters on heroforge, sending me a screenshot, and then me tracing over it and putting my spin on it. More work for me, but I just can’t get what I want from AI.