Z-Image seems to be the first successor to Stable Diffusion 1.5 that delivers better quality, capability, and extensibility across the board in an open model that can feasibly run locally. Excitement is high and an ecosystem is forming fast.
Z-Image seems to be the first successor to Stable Diffusion 1.5 that delivers better quality, capability, and extensibility across the board in an open model that can feasibly run locally. Excitement is high and an ecosystem is forming fast.
I’ve paired my Z-Image Turbo with SeedVR2 upscale, running on a RTX3060 12gb, 32gb sysMEM, generates in 40sec. I’m holding out for Z-Image Edit that is a larger model, once that is out… going to be interesting. Oh and to train your own ZIT LoRA, takes 5hrs for 3000 steps. So fast.
Z-Image Base and Z-Image Edit have been announced as being the same size (or, at least, the whole set has been announced as being in the 6B size class) as Turbo, but slower (50 steps with CFG, apparently, from the announced 100 NFEs compared to Turbo's 9 NFEs, where turbo doesn't, in the use they reference, use CFG.)
Did you forget about SDXL?
Clearly you have, but while on the topic, it is amazing to me that only came out 2.5 years ago.