Resized to 78% of original (view original)
| Prompt | masterpiece, best quality, very aesthetic, @arako, @deal360acv, @dino \(dinoartforame\), @ciloranko, 1girl, megumin, open mouth, tongue out, upskirt, holding person, full nelson, vaginal, uncensored, faceless male, rolling eyes, (colorful:.3), chromatic aberration, gritty, faux traditional media, painting \(medium\), horror \(theme\) <lora:anima-arako-april29-step00002112:.8><lora:anima-dim1-dinov3-step00002464:.6> |
|---|---|
| Negative prompt | lowres, (bad), text, error, fewer, extra, missing, worst quality, jpeg artifacts, low quality, watermark, unfinished, displeasing, early, signature, extra digits, artistic error, username, scan, deviantart, ye-pop, score_1, score_2, score_3 |
| Sampler | Euler |
| Seed | 1277148121 |
| Steps | 20 |
| Clip Skip | 2 |
| Lora Hashes | anima-arako-april29-step00002112: 0x1e0893c607, anima-dim1-dinov3-step00002464: 5ced7a714e8b |
| Model | anima-preview3-base |
| Module 1 | qwen_3_06b_base |
| Module 2 | qwen_image_vae |
| Schedule Type | Beta |
| Size | 1088x1536 |
| Version | neo |
Anonymous 04/29/26(Wed)11:25:48 No.8869091
https://files.catbox.moe/stoal2.png
>>8869082
>By the way, the way this dino lora looked on noob can't be resurrected for anima, can it?
Maybe? I'm 90% sure the secondary media tags are doing a lot of heavy lifting on the general tone/aesthetic that you're looking for and they don't work quite as well on anima as they did on illustrious based models, probably due to the whole not using CLIP anymore thing.
you might be able to snafu the TE into cooperating by including weird subtext but who knows.
for what it's worth I'm pretty sure I reused the same dataset. it's just difference in models at play.