Type | |
Stats | 2,125 |
Reviews | (77) |
Published | Jan 10, 2024 |
Base Model | |
Training | Epochs: 10 |
Usage Tips | Clip Skip: 1 |
Hash | AutoV2 45754998AA |
If you do like this work, consider buying me a coffee :)
Use v4 of this model for free on Stable Horde
The long awaited followup to ICBINP, this model is merged to try and get the most realistic images possible. It's the ICBINP you all know and love, but now with all of the benefits of SDXL (recommended 832x1216)
Update: v5a
Lots of comments that this was starting to drift towards cgi, so have added back v2 with a grad A merge, and then added the fantastic FP100C lora into the model to make it more like a photo.
Update: v5
No new training on this one, just a merge with the fantastic Voight Kampff model, and some detail and aesthetic loras merged in (specifics in the about this version). Should give you the same realism, but with a more midjourney'esque cinematic feel.
The first six sample images are made in Forge with the DPM++ 2M Karras, 35 steps, 832x1216, cfg 3, clip skip 2, and no additional tooling so should be reasonably easy to replicate.
The remaining samples (all the ones with line breaks in the prompt) were made using cfg at 2, the Forge Coupling extension, the Peturbed Attn Guidance extension (scale 3, adaptive 0 or scale 6 adaptive 0.3), the FreeU extension (1.1, 1.1, 0.9, 1.1), and the LatentModifier (5 sharpness, 20-40 extra noise multiplier) all enabled.
If you want to push this to make anime, change the sampler to DPM++ SDE 3M Exponential and crank the CFG up to 6-7.
I'm sure that some more experimentation and better workflows you will be able to achieve amazing results!