HOME | DD

Description
Similarly to the previous image, SD 1.5 struggled to handle the gigantic belly, constantly trying to either shrink it or turn part of it into legs. I do wonder if an angle with less chances of being interpreted ambiguously would've helped the AI along.

The base image is cropped from this image: www.deviantart.com/zel1302/art… (can't share, because it's not something I have permission to share!)

Disclaimer: img2img is an iterative process, usually involving feeding output images back into the model to come closer to the ultimate image. While the starting image and generation data are provided, you shouldn't expect the base image to directly yield the final images seen here.

Number 1
This is legitimately as far as I could take it from the original image before the belly started being lost. It really isn't far at all...

Prompt: Morbidly obese woman, blonde hair, green bra, massive belly, sagging belly, beautiful face, art by artgerm
Negative prompt: deformed
Seed: 163814572
Other: Steps: 20, Sampler: Euler a, CFG scale: 7, Denoising strength: 0.2, Mask blur: 4, Size: 512x512
Model: Stable Diffusion 1.5

Number 2
The best image that still remained slightly similar to the original image, albeit not as close as desired.

Prompt: Morbidly obese woman, blonde hair, green bra, massive belly, sagging belly, beautiful face, art by Greg Rutkowski
Negative prompt: deformed
Seed: 1100011456
Other: Steps: 20, Sampler: Euler a, CFG scale: 7, Denoising strength: 0.6, Mask blur: 4, Size: 512x512
Model: Stable Diffusion 1.5

Description
Similarly to the previous image, SD 1.5 struggled to handle the gigantic belly, constantly trying to either shrink it or turn part of it into legs. I do wonder if an angle with less chances of being interpreted ambiguously would've helped the AI along.

The base image is cropped from this image: www.deviantart.com/zel1302/art… (can't share, because it's not something I have permission to share!)

Disclaimer: img2img is an iterative process, usually involving feeding output images back into the model to come closer to the ultimate image. While the starting image and generation data are provided, you shouldn't expect the base image to directly yield the final images seen here.

Number 1
This is legitimately as far as I could take it from the original image before the belly started being lost. It really isn't far at all...

Prompt: Morbidly obese woman, blonde hair, green bra, massive belly, sagging belly, beautiful face, art by artgerm
Negative prompt: deformed
Seed: 163814572
Other: Steps: 20, Sampler: Euler a, CFG scale: 7, Denoising strength: 0.2, Mask blur: 4, Size: 512x512
Model: Stable Diffusion 1.5

Number 2
The best image that still remained slightly similar to the original image, albeit not as close as desired.

Prompt: Morbidly obese woman, blonde hair, green bra, massive belly, sagging belly, beautiful face, art by Greg Rutkowski
Negative prompt: deformed
Seed: 1100011456
Other: Steps: 20, Sampler: Euler a, CFG scale: 7, Denoising strength: 0.6, Mask blur: 4, Size: 512x512
Model: Stable Diffusion 1.5