Aeontropy on Nostr: it really depends the method on local at the moment you configure your settings to ...

it really depends the method on local at the moment you configure your settings to generate with base res on sdxl models could take 10 to 15 secs. If you use upscalers around 40secs or a minute and the res max without running out of cuda easily 17xx pixels heigh x 12xx width if you have more system ram there's a method called multidiffusion it can go higher than that but obviously will take more to gen since it will start using system ram once it runs out vram.