Size AI

AI-generated sizeart. #sizetwitter #sizeart

Disclaimer: I refuse to take credit for anything. Intended for size artists to reference. Support size artists.

[Size AI @ai_size]

@MisterM105 Use https://t.co/Cleu1X8iHK

[Size AI @ai_size]

NovelAI seems pretty advanced. A huge downside is that it requires currency, unlike Stable Diffusion which is basically open source.

[Size AI @ai_size]

@giantesshug @ZealZyznm The default sampler, but I don’t have the name on top of my head. Cfg scale is 10 and steps is 40.

[Size AI @ai_size]

AI Generation of Giantess at the Beach
Main prompt: Crowd at beach, giant butt in the distance, giant butt towering over beach, giantess

#Giantess #sizetwitter #巨大娘 #stablediffusion

[fizznibble @fizznibble]

@biing222 https://t.co/Eyjy9HFLCz

[Size AI @ai_size]

@ZealZyznm You just have to go through a lot of batches. This was probably the only decent one out of 100. I’m using the python version of the script so this is made in the 1.4 version of the ckpt model. The one on the browser is the 1.5 version, so it could have better results.

[Size AI @ai_size]

@Jefff_Jeffery_J @CHelleaven https://t.co/TFSv2k9GZ8

[Size AI @ai_size]

Main prompts:
"giant anime girl towering over clouds, airplane in foreground, giant anime bikini girl towering over clouds"

#Giantess #sizetwitter #巨大娘 #stablediffusion

[Size AI @ai_size]

@CHelleaven A ckpt file drastically changes the weights of image generation. Stable diffusion and waifu diffusion are different ckpt files, and the program can only run one at a time. You'd need to swap the ckpt file in the program folder to switch weights.

[Size AI @ai_size]

@CHelleaven You can use the waifu diffusion model to get better anime faces as it references the booru depositories. But its so close to some actual drawings that you might be able to tell what artist or drawing they sampled it from.

[GTS★HUG ジャイアンテスハグ @giantesshug]

Stable Diffusion AI Experimenting - Giantess Artwork [5]

Prompt + Settings: https://t.co/K34T2jn4qP

#Giantess #SizeTwitter #StableDiffusion @ai_size

[Size AI @ai_size]

@Doge93874653 Pose generated in Stable Diffusion 1.4, processed face with Waifu Diffusion model 1.2. Both versions can be downloaded from https://t.co/TFSv2jSDX8 for free.

[Size AI @ai_size]

Last AI generation for today. Max Sampler and CFG scale with same prompt as posts below.

#Giantess #sizetwitter #巨大娘 #stablediffusion

[Size AI @ai_size]

(don't give up on drawing just because an AI model can combine preexisting art)

[Size AI @ai_size]

To achieve decent AI results, reference and weigh artists in the prompt. AI models attempt to replicate that artist's style based on the portfolio they have already. This is why so many people ask for artists in prompts, and why artists are still and always integral to art.

[Size AI @ai_size]

@doronjom8 Not at all! It comes with most stablediffusion models. You can try out a version of it on: https://t.co/ZASZOYX4do

[Size AI @ai_size]

AI Generated Giantess
Main Prompt: "giant anime woman girl towering over city"
#Giantess #sizetwitter #巨大娘 #stablediffusion

[Size AI @ai_size]

@CHelleaven Hope I was helpful!

[Size AI @ai_size]

@MDZZ_Main I've been using a lot of tutorials, but anything that has been up to a week recent will do. I'd also recommend Aitrepreneur for details.

[Size AI @ai_size]

@CHelleaven stablediffusion v1.5
I try to find a good pose on text2image first, then process it through img2img