Zeitstein - Timestone
Zeitstein is an artistic, primarily 2.5D, custom merge with a rich, wide variety of subject applications. It brings flair and detail to the animated side of the 2.5D space, and is able to venture into 3D or flat color art, though it's not great at realism, unfortunately; future versions will implement more versatility.
Samples include prompts and settings that I recommend. My images are made with ForgeUI using the DDPM sampler and Align Your Steps 32/GITS scheduler, and in as is usually the case with my samples, no necessary embeddings or LORAs, as well as a variety of prompt structures (tending toward simpler prompts to showcase what the model itself can do for you.) I use, and recommend, ADetailer, as it significantly improves eyes in particular.
Much like SteinCustom before it, this model has finally diverged far enough from its predecessors to indicate future development will take it even further. At its core, this is a merge of SteinCustom-V11 and SteinIllustriousV13, with a bit of my own NemoMix (one of my older 2.5D Pony models) thrown in, and the result is something distinct from both its forebears.
Description
More refinements, more advancements, still keeping within the same style. This one's enhanced with a little injection from one of my other models (the new Augenstein version, posting alongside this), giving it a bit more color and creativity.
FAQ
Comments (5)
Which scheduler do you recommend for using with the DDPM sampler on this checkpoint?
1. Thank you!
2. Can you please mention "What was changed" between versions. Please.
The change is listed in the "About this Model" section of the page.
At the moment, I’m using this model for img2img work. I’ve tried more than 50 models, and so far this one is my favorite — but only with the Detailer.
I see a big difference between version 1 and version 2. I don’t see any difference between 2 and 3. I haven’t tested version 4 much yet.
@waxattack 3 was a back-merge at about 20% with SteinIllustrious, one of my other models, primarily I found that it added or gave different aesthetics to creative details (like clothing, poses, hair, and such) and a bit more color, without altering the composition of the image very much. I tend to prefer versioning like this with my stuff, because it's jarring to me if there's a massive style jump from one version to the next. V4 is a bit more of a change, which can be tested if you generate images off the same seed from each model; you'll see gradual changes, some more than others.
All that said, I'm not what I would consider a professional at this, and simply enjoy sharing what I've created, customized, or learned in my own journey with AI, so I apologize if I lack a more expansive explanation.