Ancient theme WG, happy with how it turned out, making these are mad addicting.

>>12834 Part 2, Not sure why the last post was all spoilered

Fall Of The Queen Brat 1/2

>>13328 Fall Of The Queen Brat 2/2 I also have captions to go with the sequence at reddit.com/r/WeightGainAI/comments/16w2r8z/fall_of_the_queen_brat/

Lifeguard pt 1/6

>>15536 Pt 2/4 (it's actually 4 parts, sorry for typo)

>>15537 Pt 3/4

>>15538 Fin

Messed around with some of the models on the Model Madness thread and got some pretty good results: https://huggingface.co/mariaWitch/biggerModelMerges/tree/main?not-for-all-audiences=true I mainly used the bbw150kg model it was very responsive to sequence prompts and had a generally great style. It was extremely hit or miss which isn't surprising since you basically just have to get lucky with the poses, continuity, and quality but when it works I think it works quite well. I'll be posting the rejects from this run on >>10270 since they came out extremely well as individual art but failed as sequence images.

> parts 8,9,10 from ongoing WG comic that's up to pt. 16 so far > gordobabosoff on deviantart / patreon

>>15768 Im sorry, im on mobile and meant to post these in the dall e 3 thread

Made this on a whim.

🍰🐉

Grumpy hardbody turns into happy lardass.

>>15539 This sequence is phenomenal! I tried to reverse engineer your results in the second to last image, but I can't track down one of the loras you used, and wanted to ask if you could point me in the right direction: I couldn't track this one down "<lora:m-fat:1>" and wasn't sure if I'm using the same model as you. Is the AuroraONE_F16 a bbw model? When I run your same parameters but with the AuroraOne_F16-bbw model I have, and without that lora, the result looks like this. Did you have to inpaint her belly a bunch to get the desired effect? Any help would be appreciated! The look you created here is exactly what I've been trying to nail down. Great job on the sequence, it's fantastic.

Read up on a lot of what's been posted here and on the rest of the board/guides. Check out other stuff I'm working on at: https://www.deviantart.com/syntheticsoftness

Tried to generate a before/after effect using a prompt only. Maybe there is a Lora for split-sequence? Found only dressed/undressed and blocks like comics. It's convenient to generate two at once

Pardon, forgot the second one

Little messy

Schoolgirl

Schoolgirl

Weight gain sequences

>>21900 The husband gets drunk beats wife. Husband cant stop drinking liqour. Punching holes in so many walls. FUCK!

>>21903 I figured he did. Getting hot outside. The same. Wesson. You better put that beer belly in check. Before you bet your ass in front of the Feds I will.

>>21899 >>21900 >>21901 keep it up champ you've almost generated an image

>>18701 What model/recourses did you use?

>>22698 Sorry that I'm this late, but: Model: https://civitai.com/models/162207/super-ani-invicible Loras: https://civitai.com/models/48197/obese-girls-or-concept?modelVersionId=52816 If you drag the image into ComfyUI you should get the workflow, but to quickly summarize: 1. load model -> FreeU_v2 node -> load loras 2. For the prompt I concatenate the default good image prompt with what I want (generally makes it follow the prompt better). Same thing with the negative (I use a few embeddings here). 3. Afterwards it's just a simple generation with dpmpp_3m_sde at like 40 steps then completely basic latent upscale (no fancy nodes, I find that this works almost as well but speeds up generation times on my GTX 1660 super). 4. Finally I just use face detailer at the end.

From thin tp very fat immobile

>>23992 Not how this works, retard.

>>17365 I should probably check this board more often AuroraONE is not a BBW model - just the standard Aurora model you'll find on huggingface (at least back in June of 2023). m-fat is the 'obese girls' lora by LeonTalon on civitai. I use a toroboro LoRa that you can't see in the final image. I get the entire image sorted out, then in the last step I drop the Toro lora, move "photorealistic" from negative to positive, and run img2img at a low weight.

I've made some sequence for Tracer

And here's a little bonus

>>25987 >>25988 >>25989 >>25990