LATEST: X-RAY 2.5
if you wanna enter my small discord server to discuss about my LoRA's: https://discord.gg/jqmqnkdNBZ
I am sorry for my trouble caused, after hours of troubleshooting with the model, the model won't upload and I did a mess... I am really sorry. I clicked save for every change I made. I will finish this page later.
For my old user's: I know the 2.3 and 2.4 model are gone, but I will try to upload them later, and also, this model, outperforms the older models by 60% if you follow my steps: try to generate an image as you normally would with xray and stuff, then save that image and interroage it in deepdanbooru, then copy the generated prompt in text2image and try to generate, it will be much better. I will update this model page later, and again, I am sorry. Also, I can't upload more showcase because of an error when trying to Publish the LoRA's. Sorry.
Edit: Added them back.
It's recommended to iterrogate an anime image before with DeepBooru.
Recommeneded to add the related prompts about x-ray in general ( x-ray view, cross-section, vaginal, uterus, cervix, penis hitting uterus, internal cumshot, etc,...) and Use either Euler at Euler A at around 20 or 50 steps
Have fun!
PS: Might release a improved version in the future if it's requested enough + fertilization/impregnation model
Done!
If you would like to buy me a small coffee to keep me up at night, it will be much appreciated!!
Description
FAQ
Comments (15)
Would you mind making a version with only the x-ray, especially for inpaint, please?
you want a LoRA that has been trained only on the cropped x-ray thingy?
@grendar I think that is what he means. Would be great to be able to inpaint with that.
Oh, sorry for not answering this sooner, but yeah, it's just like you guys said.
Is X-Ray Anime 1.0 and 1.5 still Lora or is it Model (checkpoint) ?
Will there be Fallopian tubes and ovaries included in the future?
I really like this, but it's still difficult to use. Keep up the good work.
This Lora will cause serious pollution to the entire art style of the image. Is there a solution?
That's likely an issue with your workflow; turn down the lora strength, adjust your prompt, lower/raise your steps/cfg, etc.
I've used this lora with a number of different models and it's done a decent job at adjusting to the situation.
Another cause could be that you are using conflicting models; don't use 2 sex models, etc. It creates too much sampling conflict.
First, generate a picture without using this Lora, then transfer the picture you like to img2img, select a free area, this time use this Lora and set the denoising strength to greater than 0.9
Style chage happens because any checkpoints react very sharply to the presence of the "x-ray" tag, so artifacts arise
I don't know if you included it into the training set already, but it would be nice to have both male and female anal sex cross-section and urethra cross-section. Also, as someone already mentioned: oral.
my best prompt:
, ,(x-ray view:1.2), (cross-section:1.5), vaginal,cervix, penis hitting uterus, internal cumshot,,(fill cum:1.2), ,, uterus , (uterine wall:1.1),
, frontal view, looking at viewer, ,
, <lora:sex xray-2:0.7>,sex x-ray ,,,, <lora:xray2.5:0.7>, suspended on penis,,, , <lora:cowgirl_with_hands_on_knees_v1.0:1>,pov,astride,thrusting ,
xray2.5: X-Ray Hentai 2.5 - X-Ray Hentai 2.5 | Stable Diffusion LoRA | Civitai
sex xray-2: cum in pussy X-ray/体内射精X-ray - 体内射精v1.0 | Stable Diffusion LoRA | Civitai
cowgirl_with_hands_on_knees: cowgirl with hands on knees - v1.0 | Stable Diffusion LoRA | Civitai
Which other models does this work well with? So far I haven't been able to get it to actually work yet?
Although, maybe I need more steps or something?
Noice, are you going to make an XL version?
Is there a way to turn off the overlay picture? I'd just like to see the cross section over the body
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.