A versatile supermix of loads of models.
V4 is nearly done... Using up to date source models, reimagined recipe, and higher precision merges!
I find that the model behaves best at higher resolutions, so if your 512x512 render come out strange, pump up the res.
Clip skip 1 is best for photorealism, clip skip 2 is best for a simple, drawn style.
The only portion of the recipe that I can discern from days of scribbling terrible notes is this:
[[EimisSemiRealistic_1-0v + trinart2_step115000] + [[Float's Mix + Fruity Mix] + CandyMissionBerryF222-hassan]] + [samdoesartsultmerge + [samdoesarts v2 + [CopeSeetheMald-berry200_20400 + jhSSamdoesarts_v5]]] + ???
With the amount of sda in its inheritance, "samdoesarts style" does influence outputs.
For easy one-shot portraits:
Prompt: Cinematic portrait, professional fashion shot. Feminine, attractive, stunning, pretty girl.
Negative Prompt: ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, extra limbs, extra legs, extra arms, disfigured, deformed, body out of frame, bad art, bad anatomy, text, watermark
20 steps at cfg 5 with DPM++ 2M
Description
3rd iteration of my personal daily driver.
FAQ
Comments (4)
works pretty well with good prompting
hey may I ask if I can transform this from a safetensors into a ckpt file? I'm using diffusion bee and it doesn't look like I can run the model :S
If you hit the drop-down next to the download button, you'll find the .ckpt version
oh woah thank you you're awesome! :)
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.








