Stuck your favorite character into wall, with this LoRA.
Training is based on AbyssOrangeMix3 model.
In the belowing image, first column is with no LoRA, second is with 0.8 weight, third is with 1 weight. Through rows, there's comparison amoung concepts and character sittings. | 下图中第一列是不用LoAR、第二列是0.8 LoRA权重，第三列1权重。每行每行是不同的概念，比如卡墙里上半身视角、下半身视角、以及带有照片的。
You can customize the character with prompts together with "stuck in wall, multiple views, lower body with profile". 卡墙里带有照片的可以用prompt控制照片上的人物。
Notice that "lower body with profile" often show no face photo if you doesn't define the character's facial feature with prompt. Should add "XXX hair, XXX eyes" to prompt. 注意，如果不描述一下角色的面部，通常不会带有头像照片。
Description
Trigger Words for concept「lower body with a profile photo」:
stuck in wall, lower body with profile, multiple views, kneepits
Resolution may influence image composition, I recommend use 512x1024 for this concept, otherwise image maybe not long enough to contain both the lower body and profile photo.
Trigger Words for concept 「upper body only」:
stuck in wall, upper body only, hanging breasts
FAQ
Comments (7)
Can't seem to get this to work, think you could add an example with a full prompt?
This should honestly be a requirement on the site at this point.
it is a bit wonky but I kind of got some of a result using:
<lora:StuckInWallWithA_epochs25:1> , stuck in wall, lower body with profile, multiple views, kneepits, ass, adorable face, masterpiece, best quality, incredibly absurdres, unity cg, ultra-detailed illustration, hatsune miku, aqua eyes, green hair, striped panties
Negative prompt: (low quality, worst quality:1.4),(bad_prompt:0.8), (monochrome:1.1), (greyscale), deformed face, deformed, blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, extra_limb, poorly drawn hands, two heads, gross, mutilated, horrible, scary, evil, old, conjoined, morphed, text, error, glitch, lowres, extra digits, watermark, signature, jpeg artifacts, low quality, unfinished, cropped, siamese twins, robot eyes, bad_prompts, by bad_artist, extra fingers, missing limbs,
Steps: 70, Sampler: Euler a, CFG scale: 7, Seed: 2926738770, Size: 544x480, Model hash: d1facd9a2b, Model: Liven's Color Max
Will continue to test and attempt to achieve similar results.
Resolution influences image composition, I recommend use 512x1024 otherwise image may not be long enough to contain both the lower body and profile photo.
A side view concept would be great too.
您好，非常感谢您的分享，我使用了您的lora来生成图片，但这似乎有一定的问题（奇怪的画风或怪异的脸），我希望能得知您使用的大模型（稳定的扩散），如果您愿意，我将不胜感激
Hello, thank you very much for sharing, I used your lora to generate pictures, but this seems to have a certain problem (strange painting style or weird face), I would like to know the large model you use (stable diffusion), I would appreciate it if you wish
看上去对画风影响有点大，还会有更新吗
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.