/bbwai/ - bbwai

I ran into an old acquaintance a while back, and much to my surprise, she's let herself go a bit since high school. She wasn't the athletic type, but she was as thin as a rail. As we were talking, I caught a glimpse of her little potbelly slightly spilling over her tight jeans through her tucked-in t-shirt, and it sent my brain into overdrive. Guessing by her newly rounded belly and fuller face, the past six years of the married life have been generous to her figure—and she hasn't even had kids yet, though she said she's been trying for one. Anyway, since then, I've discovered a new facet of my fat fetish: weight gain with age. The idea of a thin, youthful, energetic woman growing older, heavier, and lethargic over time. The seedance algorithm is particularly handy for video transitions, and I've made these two clips so far, to try and capture what I find so sexy about this kind of transformation.

Anybody into cholas and sassy Latinas?

It's incredible that Grok allows you to make this for free.

some jav experiments.

Made this using Grok's Imagine.. If you pick last frame of the video and use it to generate another, you can splice a sequence together without too many issues. The only problem is the limit and the stupid default auto-generation which wastes time. I fear had Grok not upped moderation, I'd have signed up bc this seems like crack. Thought I was jaded but being able to animate any picture is almost like seeing nice boobs for the first time.

There aren't enough middle-aged women into gaining, so I thought I'd generate my own.

just started messing around with grok using a trish stratus photo and screen shotting last frame of each video generated to make a longer sequence I clipped together sadly by the 3rd or 4th clip her face face becomes more generic looking https://gofile.io/d/x9hjie

>>37759 Watching this I eventually began to appreciate the AI cannibalizing itself with every screenshot being more degraded than the last. It was like going on a journey from FLUX to SD 1.0

z-image is crazy

>>37922 This is your LoRa, not pure Z-Image, isn't it? I tried Z-image, but the model does not allow to achieve such overweigh. Yes, following the prompts, anatomy, and details of faces and limbs are at a high level, but such shapes cannot be achieved without LoRa's, none of which were posted publicly on Civitai, TensorArt, SeaArt, etc. I'm trying to figure this out and also train a slider-LoRa for Z-Image, but while I'm collecting a dataset with my perfectionism, someone else will probably post something similar.

>>37927 yes I trained two loras im using right now, you can drag and drop my images into comfyui for workflow, prompt and everything, here are the loras I trained: https://gofile.io/d/WSLkmg Im curious to other results

Imagine a world where people ultimately grew tired of the effort required to stay thin in the age of abundance. At first, mankind turned to appetite suppressants, but when their prices grew exorbitant, they simply gave up. Thus, humanity ushered in The Age Of Corpulence. With the world's dietary inhibitions now thrown to the wind, obesity rates skyrocketed. Those who were naturally fit and had never known the pleasure of being in a bigger body were left scrambling to catch up to their neighbors' expanding waistlines. However, it wasn't long until advancements in nutritional science led to the availability of calorie-dense liquid compounds for the purpose of rapid weight gain for aesthetic purposes. These compounds were administered via feeding tube in the mouth by board-certified aestheticians and cosmetic surgeons. Fattening procedures, or "swelljobs", quickly overtook the likes of facelifts and breast augmentation in popularity, as many believed excess fat to be a natural alternative with comparable results. I invite you to step into... The Twilight Zone.

>>37930 I tried, but I can't even get the workflow to work. You've got nodes in there that it just doesn't want to download (I updated comfyUI to the newest version because the one I had before was too old) and taking it from the github sources doesn't make them show up in comfyUI. What I'm missing is FastUnsharpSharpen and FastFilmGrain which I tried to get from vrgamedevgirl.

>>38181 okay that took way longer that I'm willing to admit, but getting the stuff from here also helped: https://comfyanonymous.github.io/ComfyUI_examples/z_image/ also I didn't realize the fastunsharpsharpen and fastfilmgrain nodes weren't even enabled, so they didn't matter.

>>38188 hi yes you are right they are not even required, share your results if you want im curious, for the ass pictures I used a different lora I trained: https://gofile.io/d/e0ONt9

>>37974 Swelljob Subject #B31: This woman (39 F) consulted with us regarding a weight adjustment for her son's high school graduation. She insisted on needing to look "good and healthy" for his big day. The size 18 dress she purchased months ago should have no trouble slipping over her new curves.

I tried fiddling around a bit and ended up building my own workflow, even though I barely understand what all the settings do. It feels like I'm walking a fine line between "looks like the lora", "broken" and "that's just not big enough"

>>38252 playing around with rain produced this little gem though,,,

that got me into attempting showering scenes

and finally some giantess pictures

>>38255 I think 404 is my favorite.

>>38258 Where? I can't find it.

>>37930 Thanks for sharing, good LoRA :) It's not my favorite body type, but I like the realism of the skin textures, even though it's a bit too much. I've had some time to train my own for Z-Image for the first time, and I have some early results, but it needs some refinement. I trained it on the De-Turbo model, and it seems that was a mistake, as there is a pattern of vertical stripes everywhere. I want to add more of larger sizes and sharper skin textures. As I understand, your entire dataset consisted of images with severe stretch marks and cellulite? And are there any nuances you have discovered that would be useful for training LoRA like this? There's a ton of stuff online about training characters, but I couldn't find anything about concepts. I will continue posting in the Z-Image thread. I posted here because you would see the reply here.

>>38234 Swelljob Subject #C147: Caroline (42F) opted for a swelljob because she had grown tired of her gaunt appearance and low self-esteem. Her self-admitted hyperthyroidism proved a challenge for our weight gain compound. We administered dose after dose but failed to see the expected weight gain. Because of her condition, Caroline actually has the distinct honor of being our first patient to need cutting-edge AdiSynth Bio-Insulation, a synthetic lipid that is injected under the skin but sits atop the muscle for the fat look and feel without any of the adverse health affects. Even if Caroline won't see the benefits of a stronger heart from narrow arteries like most of our patients, she is nonetheless ecstatic about filling in the seat of her new mobility scooter.

>>37930 thanks a lot, looks really impressive! any chance to re-upload the loras? they got archived and required premium account

I think I got shadowbanned from grok, has this happened to anyone else? it just comes up with a message saying "Grok was unable to finish replying." whenever I try to do a video or image

Swelljob Subject #D256: Lydia (36F) has been a regular at our clinic for years, having come in for various Botox treatments in the past. This time however, the petite lingerie model insisted on a weight adjustment, citing diminished demand for petite clothing and a lack of opportunities for petite models as a result. Her years in the modeling industry have left her food-averse, so she opted for a synthetic swelljob, courtesy of AdiSynth Bio-Insulation. Over the past 12 weeks, she's blossomed into a new woman. With every injection comes another job offer, she says.

>>38593 Are we getting AI generated Bo voiceovers now? That almost sounds like her (unless I'm getting model's voices confused)

Wondering if the original person who started the thread made more, or if they decided to bail after the AI uploaded giantess stuff. While I'm not into giantess stuff, they DID manage to blob up a cute person into SSBBW size and beyond, hoping we get more of her! Also! On the Bo thread, someone (who's post has sadly been deleted) mentioned the ability to create ACCURATE nudes and apparently did so for Bo...which means we could get AI Bo porn (since the real deal won't). Of course, as soon as I put out the idea, others shot me down despite the point I made....but I still think it could be worth a shot! Someone else did make an accurate visage of the angel of adipose, but apparently the examples I was able to snag may have been lost to the void of time (unless I just missed them, but I posted what I was able to get here). Anyone know if there even is a way to get examples similar to the 2 I was able to get? (Preferably for free and without the site asking for 50 billion things just to make an account?)

>>38194 Could you please reshare your lora's? theyre the best ones ive ever seen but cant find them anywhere :(

>>37930 >>38194 Can these be re-uploaded?

Or could someone else who downloaded the lora's reshare them?

nothing guys? no lora reshares available? - didnt expect the thread to die/ OP to leave.

nude sex

can someone make her really fat with fat rolls plz