Resized to 73% of original (view original)
| Prompt | masterpiece, best quality, very aesthetic, very awa BREAK jz235, gritty, faux traditional media, painting \(medium\) BREAK 1girl, red hair, red eyes, ponytail, completely nude, nipples, flat chest, parted lips, half-closed eyes, on back, from above, legs up, looking at viewer, uncensored, heavy breathing, steam <lora:vpred1-EXTREMELYshitty1dimlora-tsuntsuke-step00002112:1><lora:vpred1-EXTREMELYshittylowsteptest25:.7> |
|---|---|
| Negative prompt | sound effects, lowres, (bad), text, error, fewer, extra, missing, worst quality, jpeg artifacts, low quality, watermark, unfinished, displeasing, early, signature, extra digits, artistic error, username, scan, bruise on face, bruise, tanlines |
| Sampler | Euler |
| Seed | 415786760 |
| Steps | 20 |
| Cfg Scale | 5 |
| Model Hash | ea349eeae8 |
| Advanced Sampling Enabled | True |
| Advanced Sampling Mode | Discrete |
| Controlnet 0 | Module: tile_resample, Model: xinsir controlnet tile [4d6257d3], Weight: 0.35, Resize Mode: ResizeMode.INNER_FIT, Processor Res: 0.5, Threshold A: 0.5, Threshold B: 0.5, Guidance Start: 0, Guidance End: 0.45, Pixel Perfect: False, Control Mode: ControlMode.BALANCED, Hr Option: HiResFixOption.BOTH, Lora hashes: vpred1-EXTREMELYshitty1dimlora-tsuntsuke-step00002112: 99671485df44, vpred1-EXTREMELYshittylowsteptest25: 1d6acda1d135 |
| Denoising Strength | 0.8 |
| Discrete Sampling | v_prediction |
| Discrete Zsnr | True |
| Model | NoobAI-XL-Vpred-v1.0 |
| Schedule Type | Normal |
| Size | 1152x1632 |
| Version | f1.2.0dev-v1.10.1RC-previous-955-g456f2771 |
- ‹ prev Search: spread_legs next ›
Anonymous 10/06/25(Mon)13:55:30 No.8738008
>>8737953
>>8737972
slight further testing
https://files.catbox.moe/6wu0e9.png
styling requires a fair amount of cheating and ends up losing some minor details(like the direction the lines of the hair being pulled back and some random bangs getting added), but those are kind of fixable in post. The cheating comes in with the base t2i being generated with only with the low stepcount lora(acting as a "character" lora) and the manual i2i upscaling got a second 1:1 res pass to force the style to a significantly higher capacity. original upscale output for reference https://files.catbox.moe/1o7ijf.png
I'll probably try it again and change some things with a 3 image dataset later.