This checkpoint has a Turbo-Lora backed in!!! Use CFG: 1-1.5 and 6-12 steps, the style will come up at step 6!!!!
CFG 1 will ignore negative prompts (actually you do not need negative prompts) This will bang out images insanely fast!
CFG1.5 will consider the negative prompts! But will be slower!
All sample images are generated WITHOUT ADetailer!!! The Checkpoint itself is super crisp!
Okay buckle up! Let me introduce you guys to my newest kryptonite!!!!
The creation if this checkpoint made me find my love for illustrious again :DDD
Hours of trying out, adding styles, using sliders for colors, contrast, sharpness and now I can’t stop using it! There is a Turbo-Lora backed in!!! So use 1-1.5 CFG and 5-12 steps (I use 8)
Many people wrote me about the style I used for my last three-strap Swimsuit. Here it is! I also created a Style-Lora but this her is more capable.
It is ultra artistic and has such a cool flair that the pics become insanely cool even with just a bit prompting!
Description
FAQ
Comments (7)
Bro, this checkpoint is amazing! It generates images so much faster than other models. Could you tell me which Hires. fix settings are best to use for a 2x upscale to preserve the anatomy? The model often adds two navels or places one too high near the chest. Also, what's the trick behind Hires. fix making the style so much juicier and more detailed by adding a lot of extra strokes?
Super happy that you also see the beauty in this style. I am also freaking loving it :D
honestly hard to tell cause I do not use hires that much.... Also I use draw things on mac, so the settings are probably totally different compared to a windows pc. I will test it and maybe I can give you some insides tomorrow.
@quadratrund Thanks for replying, bro! This checkpoint stole my heart, honestly. I’ve always been looking for a model that I could use non-stop without getting bored, and I finally found it. Regarding the OS, I’m actually on a Windows PC running Stable Diffusion Forge. I’ve been thinking about switching to ComfyUI, but I haven't taken the plunge yet.
"Sorry about my English, it’s not my first language."
@idana3120400 hey man all good, it is not my first language either. So I experimented a bit, I do not get this issue with "double navels" but my settings are that the hi-res-fix orientates itself at 70% of the first generated image. Can you use this information? when I go to 90% I get your navel-issue to. but this seems to be a general hi res fix problem.
A fast and stylistically consistent illustrative checkpoint with a clear artistic identity. Shadow‑Anime excels at delivering a painterly, stylized look without relying on negative prompts. Linework, shading, and color harmony are stable across seeds, and the integrated Turbo‑LoRA makes it extremely responsive. As a refiner, it performs even better: it enhances contours, adds clarity, and reinforces the stylized aesthetic without overwriting the original composition.
Pros
No negative prompts required — the model stays clean and coherent on its own.
Strong stylized identity: consistent linework, soft painterly shading, and cohesive color palettes.
Excellent as a refiner at 40–60%, adding definition without destroying the base image.
Lightweight prompting: responds well even to minimal or simple descriptions.
Fast generation thanks to the integrated Turbo‑LoRA.
Cons
Without explicit age control, the model tends to skew toward adolescent proportions.
Extremely sensitive to CFG scale: small changes can drastically alter style, anatomy, and lighting.
Strong sampler dependency: results vary significantly between samplers, with some producing drift or flattening the style.
Facial variance is low: the model tends to generate the same face unless forced via prompt or by injecting a character embedding.
Body proportions can compress toward “younger” silhouettes unless explicitly corrected.
Verdict A refreshing alternative to realism‑driven checkpoints. Shadow‑Anime embraces stylization with intention and personality, and deserves more visibility — especially among artists who value illustration over photorealism.