Please reply to this post in order to make a request.

You’re strongly advised to fill in the following fields every time you’re asking for a new request:
>specific artist / art style (greyscale, monochrome, anime screenshot, flat colours etc. etc.). When requesting specific artists, verify that there’s a LORA for that artist for Stable Diffusion or that the artist is tagged in at least 30-40 posts on Danbooru for NovelAI. When requesting anime screenshot, always name the anime series.
>specific character / peculiar body traits (eye colour, skin colour, hair colour etc.etc.; for canon characters, just name them)
>camera angles (from below, from above, close-up etc.etc.)
>framings (full body, lower body etc. etc.)
>setting (location,day time)
>pose/actions

Be as descriptive as you can! Vague or effortless requests might NOT be considered.

> How can I view the prompt of an image?
1. For Stable Diffusion: install https://gist.github.com/catboxanon/ca46eb79ce55e3216aecab49d5c7a3fb then ask the user for the catbox of the image.
2. For NoveAI: ask the user for the .png file of the image, then download the image and upload it on https://novelai.net/inspect

File: 2ee621a4-e23e-4bd7-92f7-1(...).png (2.46 MB, 1024x1024)

2.46 MB PNG

>>11357043
yeah so the general image was not that time consuming to create (took me about 45 mins), but then I was stuck for 2 or 3 hours trying to create the damn footprints, I might get back at it one day but ill skip it for now, and If I find a way ill be sure to share it. Till then, enjoy some makima.

File: ComfyUI_34334_.jpg (1.18 MB, 1664x1160)

1.18 MB JPG

>>11357107
since theres no specific tag for that and probably not a lot of images of it used in the training set, if a combination of danbooru tags doesnt do it then you won't be able to do it reliably unless you get a lora for it or summon your paint editing skills to draw in the prints and img2img...
but i can only speak about what works from a local perspective, doesnt nai have a vibe transfer feature that can reproduce concepts if you have a reference image?

File: 1736946985946918.png (1.44 MB, 955x955)

1.44 MB PNG

the summer heat can be a mixed blessing

File: 1725253745682935.png (1.47 MB, 1216x832)

1.47 MB PNG

Finally managed to get NAI to behave with multiple characters.

File: 1736465945278939.png (1.72 MB, 1216x832)

1.72 MB PNG

And for those that would prefer the backdoor.

File: big officer.png (975 KB, 1024x1024)

975 KB PNG

>>11357107
Not normally into armpits, but this is really hot.

File: a3bb4741-191f-4af4-ac08-0(...).png (1.89 MB, 1024x1024)

1.89 MB PNG

>>11357145
what's up with specifically needing to use danbooru tags or danbooru in general, does the AI only get their training from there or smth?

And if it does, would be possible to manipulate danboorus database to include tags related to the generations I or anyone else would want to create?

As for the vibe transfers, yeah I use them and even tried to use them for the foot prints but I don't think the AI recognized them well enough for what I was trying to generate.

>>11357243
have something else instead then, this ones got some variety.

>>11357045
Requesting shrinking with Asuna bunny from Blue Archive, size gambling concept, no specific artstyle, colored, any angle, teasing, fullbody, taking place in a casino

>>11357223
>>11357225
>>11357227
>>11357229
These are really great anon. Keep it coming.
Prefer futa.
Maybe praxis + vale from xenoblade2 with mini morag?

File: 1751561758751936.png (1.49 MB, 1216x832)

1.49 MB PNG

>>11357482
Fresh out of credits but I'll make more soon-ish.
>Prefer futa.
Next ones probably will be, in the meantime I got this one out of it right before credits ran out (maybe could have used a few more touch-ups, oh well).

File: file.png (2.5 MB, 832x1216)

2.5 MB PNG

>>11357477
Is this something close to what you have in mind?
>>11357498
>>11357229
These gens are pretty solid. Thank you for your experiments with NAI.

>>11357498
Nice forced cock vore on horsecock with lamia tongue on top.

File: file.png (2.56 MB, 1024x1024)

2.56 MB PNG

We need more tiny girls sniffing testicles

>>11357498
Nice gen. can you make more like that^?

>>11357549
agreed have a cock sitting on girls face be it from a man, futa, shota or loli futa
>>11357581
MOAR, need giant mommy tsunade

>>11357045
giga giantess phosphora wearing a bikini showing off her sexy body as multiple floating islands crash into her, preferably with a small inset showing a panicking tiny palutena

File: ComfyUI_25663_.png (2.55 MB, 1872x1160)

2.55 MB PNG

>>11357459
it's just a really convenient site to scrape for training data because it has a shitload of anime images that are all autistically labeled with a wide range of tags, which saves a ton of work for the model maker and saves us from the badly automated captions with minimal to no artist info that leads to boomer prompt models that only know 5 artstyles
danbooru being the biggest booru out there means their tag system is pretty much the standard, and from an art appreciator perspective it makes it easy to come up with prompts, because youre already used to seeing those tags
i think theres probably a way to request new tags via the forum, but even if you get the tag in
there you would still need to wait for someone to train a completely new finetune on it, so lets just say it could take a while

File: nagi.mp4 (3.23 MB, 640x640)

3.23 MB MP4

any tips for wan img to vid prompt so giantess hand reaches towards viewer/screen making it look like shes grabbing you?
also been trying to make vids with Daz renders but guess that should go in /aco/ thread right

File: Bath time.png (2.58 MB, 2096x1536)

2.58 MB PNG

>>11357630

File: 94af07b4-3b91-4417-bd94-2(...).png (1.46 MB, 1024x1024)

1.46 MB PNG

>>11357852
Gotcha, well kinda gotcha, I'm not that into the model maker space, by that I mean I'm not really a tech guy, so just to make it clear, Danbooru is only used for finding tags you want to use? NovelAI doesn't actually use Danbooru as a training source for their AI? I'd really like to know more details about how the NovelAI's AI works if you have any insight and if you're willing to share, cause I'm mostly interested in finding out more about NovelAi as I've tried setting up comfyUI only to find out that my PC is dogshit for it.

File: file.png (2.46 MB, 1024x1024)

2.46 MB PNG

>>11357633
What about this?

>>11357504
NTA but I fucking love this concept

File: ComfyUI_30873_.png (3.6 MB, 1456x1872)

3.6 MB PNG

>>11358068
>Danbooru is only used for finding tags you want to use? NovelAI doesn't actually use Danbooru as a training source for their AI?
No, I'm sure they do use it and any other booru site they can scrape. To train an image model you need labeled data, which is to say the labelling must represent the prompt that should lead to said image. This means Danbooru suddenly became one of the best sources of training data for this specific (tag based, anime style) type of models, because if you just finetune a pretrained model (like SDXL) on a whole dataset of booru tagged images, the model learns to generate the right images based on the tags. Because there are so many distinct artist tags and because models can generalize these styles and combine them, you end up with a model that can do a wide variety of styles and all you need to do to get those styles is know the tags. And you can easily find new tags by simply browsing the site. That's a completely different approach from the usual base models which require you to form whole sentences describing the scene and the vibe, and they usually dont know many artists or styles so you end up needing loras for everything.
afaik NAI started out as an SDXL finetune just like pony, illustrious and noobai, but more recently i think they also train models from scratch (though not entirely sure on this)
>I've tried setting up comfyUI only to find out that my PC is dogshit for it
How dogshit exactly? 8gb vram is enough for SDXL based models

File: bc8e0563-9dff-4dc8-9ab6-4(...).png (1.79 MB, 1024x1024)

1.79 MB PNG

>>11358201
I got 7gb of it (don't ask how, don't know), I can do basic models and like medium models but my pc starts lagging like crazy and overheats in minutes, so not worth it.

One thing I did note about NovelAi tho is that it seems to be training itself with the user, by that I mean and this just might be an illusion but I've had easier time generating stuff that I want then before, and the results I get do sometimes be very similar to what I did before. So either the AI is actually remember what I generated before or I'm just a dumbass who improved over time at prompt picking and vibe transfer usage.

Also thanks a bunch for the info, appreciate it, if I'm not boring you to death, I just got 1 more question for you, in you humble opinion, do you Think NovelAi is heading towards a good direction or are they going to fall off in the future?

File: 0256eaca-4dfb-4c4c-8478-8(...).png (1.67 MB, 1024x1024)

1.67 MB PNG

>>11358525
I just noticed my horrible english here and would like to take a moment to apologize to anyone who is going to read it.

I need advice on how to make the feet look good bros. I read the guides, I've teied all sorts of tags and they're still always awful.

File: 4b7c9500-ff66-439b-bfa2-5(...).png (2.94 MB, 1856x1280)

2.94 MB PNG

>>11358096
Glad you like it! Casino shrinking is actually underrated.
>>11358201
Most of what you wrote about the data training is correct.
> i think they also train models from scratch (though not entirely sure on this)
Yeah, both V4 and V4.5 are their first SDXL-independent image generation models.
>>11358525
>improved over time at prompt picking and vibe transfer usage.
This is it! For what I know, NAI doesn't keep track of prompts, moreover I also can notice a definitive improvement in your creations over time, anon. >>11358527 is especially great!
>>11358615
Add "bad feet, extra toes, missing toes" in the negative prompt then write in the main prompt box the name of those artists who are good at drawing feet: the first ones coming to my mind right now are kairunoburogu, k-rha's, casino (casinoep), nayt layt. Hope this helps!

File: ComfyUI_00094_.png (1.49 MB, 960x1408)

1.49 MB PNG

>>11358844

>>11358071
pretty good! although, would you be able to have the islands actually crashing into her and falling apart?

File: 83e2f081-1463-4b6f-948f-d(...).png (1.22 MB, 1024x1024)

1.22 MB PNG

>>11358844
Thank you very much anon.

>>11357045
How do you guys do clean inpaint to add a tiny in a scene in SDXL ? I just can't make it work

Throw me some basic ideas to prompt. Character, outfit, pose, artist/style. I'm too retarded to do multiple characters.

>>11359397
drunk goldship passed out in a parking lot

File: 1752172423514154.png (1.27 MB, 1152x768)

1.27 MB PNG

You've disappointed her...

>>11359169
kek

>>11358615
In my experience positive tags help here more than negs. Detailed feet or foot focus will get you higher quality shots assuming the rest of the tag structure reinforces that perspective. In some of my best sets, I've not used neg tags at all.

File: 223428_%KSampler.seed%_00001_.png (3.65 MB, 2352x1744)

3.65 MB PNG

>>11359408
Here my low effort gens

File: file.png (3.54 MB, 2352x1744)

3.54 MB PNG

>>11359531

File: 224142_%KSampler.seed%_00001_.png (3.57 MB, 2352x1744)

3.57 MB PNG

>>11359532

>>11359397
giantess futa Tsunade with a bit of muscle tone and a fat cock (no hyper) looming over a shrunken girl in Norza style if possible
standing in bedroom or sitting on bed

>>11359169
damn that's awesome

File: 035038_%KSampler.seed%_00001_.png (3.93 MB, 1944x2352)

3.93 MB PNG

>>11359547

File: 035751_%KSampler.seed%_00001_.png (3.91 MB, 1944x2352)

3.91 MB PNG

>>11359641

File: 71.png (990 KB, 800x1131)

990 KB PNG

>>11357045
Inpaint Sabrina, Red, and Leaf onto attached image. Replace Lusamine with Sabrina, Elio with Red, and Lillie with Leaf.

>>11359642
thanks anon nice seeing Sakura going thru her usual training
great body proportions for Tsunade

File: 234027_%KSampler.seed%_00001_.png (2.98 MB, 1584x1912)

2.98 MB PNG

>>11360040
Np, apprerciate the ideas otherwise I get too into generating the same slop for myself. I could prompt the same line of gacha waifs for days.

Futa was a pain since the having a dick and tiny girl tended to merge them into an unholy clusterfuck.

Gotta learn to in paint with comfy.

File: 00094-942200709.png (3.24 MB, 1600x2000)

3.24 MB PNG

>>11360220
>I could prompt the same line of gacha waifs for days
been doing this trying to get something decent, i hate it
cant get Wendy to fuck smaller Erza since she always jobs even in my outputs
been trying to turn some of my daz stuff into clips but that is also failing

File: Repeated Fan FoooolyStyle.png (3.77 MB, 2688x1536)

3.77 MB PNG

it has been a bit since I posted a Turbo

>>11360203
hhhnngg mommy
great work
>>11360204
how did you get it so she gets larger when getting closer to the camera , what program?
I havent had any luck with this when using pic or renders of giant girls walking or getting closer to camera they keep the same scale instead of looking bigger the closer she gets
>>11360461
mind if request a 10-15' tall Jet crouching and smugly petting the head of her smol trainer guy or girl

File: ShrinkLabs2.mp4 (1.05 MB, 768x1152)

1.05 MB MP4

>>11360204

>>11360721

I'm not sure if it's because it was trained well enough, i described it well enough, or if I just got lucky.

File: ShrinkLabs3.mp4 (1.6 MB, 768x1152)

1.6 MB MP4

>>11360765

>>11360765
>>11360768
ah still nice work especially the second one
>>11360797
loving this futa dark elf muscle mommy
want to be the tiny and jump on her bulge hugging it as it grows

File: 00284-1618549580.png (2.04 MB, 1536x1536)

2.04 MB PNG

>>11357045
Mordred funny idea I got

>>11357498
Been away goonin to ai rps. But holy fuck this is fantastic. More of this idea

File: Paired Accelerator Headpat.png (3.98 MB, 2048x2048)

3.98 MB PNG

>>11360721
>mind if request a 10-15' tall Jet crouching and smugly petting the head of her smol trainer guy or girl
I don't know how best to convey 'trainer', so it's more of a smol manager in a suit.
This was horrible gacha, even with regional prompting.
picrel and the next one are the best(and just about only great ones) from almost 200 different gens where I lightly changed the prompt, regions and loras used.

File: Paired Accelerator Headpat Seo.jpg (412 KB, 2048x2048)

412 KB JPG

>>11361388
in the end I think the Seo style lora helped the model grasp the scenario the best, and the lightly sketch-like style had me overlook small mistakes by the model easier.
>model refused to do high heels on grass so they became loafers

>>11361036
>rps
of what exactly, if you dont mind me asking

File: 1751687742252039.jpg (3.98 MB, 4800x4800)

3.98 MB JPG

Presenting: Anon & Goth Elf GF

>Visi, are you sure this is a good idea?
>Of course it is! A graveyard at night under a full moon is the perfect place for magic!
>Please remind me why you need to cast a spell to grow taller than your mother.
>Oh stop it, you'll be very happy with the results.
>Now then...
>Klatu, Vecna, Nico....!
>!
>!
>Wh-what's happening to me?
>It's working! I can't believe it, it's working!
>Just wait 'till mom sees me now!
>Hey, wait! Why did I shrink? Why am I a girl? This isn't what I agreed to.
>Dear, dear Anon, I said you would be happy with the result of this spell, and rest assured I will make you very happy...
>(You look like the one that's happy...)
>Oh, so much to do, so much to do!
>Let's go back to my place and, ahh~, in a minute. I just need to, ahh~, settle down first, and...
Several hours later:
>That's it! I don't care how long it takes me to get back to your place like this, I'm leaving!
>No, Anon, wait. I'm almost done. Just a little longer and...
Several more hours later:
>Ahhh...ahhh..okay, I'm done. We can go.

>>11361388
>>11361393

Cute as fuck.

>>11361547
Ahahaha.
No.
I do very depraved and fucked things.
Things that's top 90% of the users here i'd bet.
Not a chance sorry.
Anon status isn't enough for me to air that.

>>11361430
Fucking geez, I can only get so hard... Please gen more of this pair. I'll take anything.

File: ComfyUI_34891_.png (2.83 MB, 1160x2080)

2.83 MB PNG

>>11358525
>training itself with the user
That sounds unlikely because training on its own outputs would be very bad for the model. It's normal to get better over time, even if you're not actively trying to. For example when you copy the prompt from an old gen you liked and slightly change it for a new gen, you are performing natural selection on your own gens.
>do you Think NovelAi is heading towards a good direction
No idea, I don't use their service but many anons here do and the models seem to work well, so good on them for training good models and providing a good service. If they do fall off, another company will simply take their place. With a small bit of effort you will be able to use all your experience and old prompts on those other services so dont worry about being locked in.

>>11358615
>>11358844
>Add "bad feet, extra toes, missing toes" in the negative
I don't think this has ever worked for me. Picking the right model / lora and artist / perspective tags is the best way. To add to your list of artists, footer is definitely a good one. It does impact the pose of characters a bit if used at a high weight. Foot focus is very hit or miss for me, sometimes it increases the chance of getting 4 toes, probably due to some clash with another style or artist tag.
And clashing / contradicting tags are usually the cause of consistent fuckups. When I notice the model starts doing certain things wrong a lot more than usual, the first thing I do is remove tags that are related to the thing being fucked up. That also includes perspective tags sometimes.
The final thing that often gets overlooked is the actual setup: try more or fewer steps, different cfg levels, a different scheduler, switch from euler ancestral to euler or the other way around, change dimensions of the image, add another hiresfix pass with medium denoising, etc.

File: 00003-685721505.jpg (301 KB, 1600x2400)

301 KB JPG

>>11361388
>>11361393
thanks anon thats what i wanted especially the second one
>>11361548
haha that would be so weird

>>11361956
I really want to get a footjob from her.

File: eating out a titaness.png (925 KB, 1024x1024)

925 KB PNG

>>11361388
>>11361393
I want to do unspeakable things to turbo.

>>11361791
What's the worst that could happen beyond giving someone proompt ideas?

>>11361971
>>11361972
Those are amazing. I'm assuming it's a paid service to animate them?

>ai text rping with gemini as little man
>girl kidnaps me after i offend her and throws me in her mouth
>swishes me in her mouth, drinks boba tea while avoiding swallowing me
>spits me in her boba tea, takes a picture of her posing with it for social media, sucks me back up into the straw along with the drink
>after she finishes drinking boba, spits me into the empty cup, leaving me stuck inside because of residual sugary drink
>throws the cup away with me in it. later i get trash compactored to death
ai rp spits out some real gems sometimes.....

File: w87puo.png (1.24 MB, 1920x1080)

1.24 MB PNG

>>11357045
Requesting giant Aphrodite from Record of Ragnarok with tiny people climbing her boobs or stuck in her cleavage

>>11362337
I know that feel.
>Roleplaying as cat boy in sci-fi setting
>Make a catgirl space cop card
>Testing shrink ray tech greeting
>Pretend {{user}} is 3 inches tall
>Write that her breath felt really good, and that my character got hard
>Smug, she leans inward exhaling deeply
>Moan loudly, reveling in the treatment. "I wish I could kiss you. Haha."
>She picks up my guy, kissing and licking, before putting him in her mouth
>Cooms
>Spits him into her cleavage and relaxes for a bit

>>11362383
BASED
i know its just an advanced auto complete but its amazing how it needles the thread sometimes

File: file.png (1.75 MB, 1216x832)

1.75 MB PNG

>>11362381
Current AI softwares struggle at making mutliple tinies look coherent and well-rendered. This is probably my best result so far.

>>11357045
can i get giga giantess futa takanashi kiara in lingerie sleeping as her huge cock floods numerous tiny cities with cum and a tiny mori happily looks on with what appears to be a growth ray?

File: ComfyUI_35160_.png (2.57 MB, 1160x2080)

2.57 MB PNG

>>11362024
Well anon, what are you waiting for? Go ahead and ask her nicely!
>hey lady can I get a footj-ACK

File: 0417f44e-04bc-411d-a6cb-f(...).png (2.68 MB, 1024x1536)

2.68 MB PNG

>>11357044
>take shitty collage pic from gts website
>plug into chatgpt and say make anime version
this very simple method has gotten me some insanely good results

>>11362664
Could you describe the scenario a bit better, please?

>>11363539
basically have takanashi kiara (the vtuber) as a giga giantess futa, sleeping in her lingerie unaware of her size, on top of/around multiple cities that are tiny compared to her, while her huge erect cock leaks out cum taking out several of the cities. latter part was just wanting to have an implication that mori calliope used a growth ray on kiara to make her grow massive while she was sleeping, but i don't mind it being excluded since now that i write this all out it does sound like quite a lot

File: file.png (2.1 MB, 1216x832)

2.1 MB PNG

>>11363553
Ok this is what I've just come up with. Due to the scale of the giantess, I decided to place Calliope inside an inset.

>>11363576
ohhh that looks just perfect~
thank you so much!

File: ranko.jpg (274 KB, 1600x2000)

274 KB JPG

just want BIG Akane with smaller Ranko is that too much
keeps making Ranma the bigger one in most gens

File: ComfyUI_35628_.png (2.1 MB, 1160x1872)

2.1 MB PNG

i guess this is my new oc
slightly different from my old oc because her hair is purple and she wears a sarong (which i think is neat)
long live originality!

File: nerdy glasses girl become(...).png (1.48 MB, 1024x1024)

1.48 MB PNG

>>11362394
Yeah. It really makes me curious to see how stuff is by the end of the decade.
>>11363342
I cri errytime.

File: (Nether)worlds are stored(...).png (3.9 MB, 1664x2432)

3.9 MB PNG

Somehow I ended up making this.
it's size enough

>>11357045
Requesting a giga Chun Li with a massive squishy ass obliterating a city with a colossal nuclear fart. Sweaty, naked, hairy pussy, relieved look on her face, mass destruction, fire.

File: ezgif-3971278b1857ec.mp4 (2.39 MB, 768x1152)

2.39 MB MP4

>Anon-kun, I'm a fairy, let me in...

File: Izuku_Sits.mp4 (1.38 MB, 720x480)

1.38 MB MP4

>>11364465

File: Izuku Sits2.mp4 (1.11 MB, 640x640)

1.11 MB MP4

>>11364841

File: dg.webm (1.77 MB, 480x832)

1.77 MB WEBM

experimenting with wan firstframe/lastframe

File: hy.webm (2.29 MB, 480x832)

2.29 MB WEBM

>>11365231

Someone please kill the cuckfaggit ruining the thread. Tyvm.

>>11365231
Really solid. Adore it.

>>11365231
Can you send the catbox for this please

>>11364197
>>11364465
>>11364841
>>11364843
Please stop, no one comes to these threads to see dudes.

>>11365338

Parts:
https://files.catbox.moe/95lsmz.mp4
https://files.catbox.moe/mu4aw6.mp4

Combine and Interpolate:
https://files.catbox.moe/nsyrrs.mp4

Keyframe images:
https://files.catbox.moe/vpmiyb.png
https://files.catbox.moe/xrk15b.png
https://files.catbox.moe/bbbcyv.png

File: a.jpg (172 KB, 1152x1024)

172 KB JPG

>>11363763
Do you need something like a 4090 to get these to work?

Pictured: Two ai images merged in Krita... my autism is failing me.

>>11365366
>>11365276

> /gdg/ - Giant Diffusion General - size/giantess/ """giant"""" AI slop thread #25
> 1) All genders (futanari/newhalf included), body types and races are allowed.
> giant
> All genders

>>11365454
Doesn't change the fact that absolutely nobody comes here for that shit.

>>11365500
wrong
that said i wish they were cuter, i'm not a fan of bara

>>11365454
>>11365510
>every picture in the thread is of a woman
>literally every single picture
>"yeah, I bet there are tons of people who are here for dudes, lemme post some"
you can post dudes sure but don't kid yourself into thinking it's gonna be popular, you gotta admit that only 1% or 2% of the anons in this thread are gay. not that we need another slop thread or another size thread on this board, of course

>>11365500
Honestly agree with the other 1 or 2 anons, as long as it showcases somewhat decent AI usage, I'm fine with it.

>>11360461
allow me to steal the prompt, pretty plz

File: Tinyschool1.mp4 (1.02 MB, 768x1152)

1.02 MB MP4

Trying some different scenarios, seeing what I can get.

>>11365636
If the AI wants to be bad at scale in this direction I won't mind nearly as much.

File: el.webm (2.16 MB, 480x832)

2.16 MB WEBM

Didn't realize the comfyui metadata was missing from the 2 parts in >>11365398

Reuploaded them here:
https://files.catbox.moe/1h0bup.mp4
https://files.catbox.moe/m8am0x.mp4

Also another FFLF test.

>>11365454
Idea.
Kill yourself.
At least post a half decent male if yiu gon a fag up the thread.
Not that sniveling beta bitch loser cuck.
Have some self respect and dignity.

>>11365398
thank you!

Requesting more growth and sequences pleassee

File: ComfyUI_35766_.png (2.46 MB, 1160x1872)

2.46 MB PNG

>>11365454
I for one think its hilarious
keep at it

>>11365997
Growth is shockingly hard to get the AI to go for, since it doesn't really have a concept of space, and size doesn't translate as obviously in video as like, other physics things do.The easiest way to do 'growth' at the moment is like you see in this thread where the giantess leans in or walks towards the camera, because that's a lot more common in the training data than something actually growing in size. Unfortunately that usually means if you actually -try- for growing, the camera will often just zoom in or the character will walk forward instead of actually growing larger.

More city butt crushing sequences, now that would be good.

File: nopan Repeated Fan Fooool(...).png (3.92 MB, 2688x1536)

3.92 MB PNG

>>11365553
here's the image before any multidifusion upscale: https://files.catbox.moe/syq4w5.png
The fooooly lora does a fair amount of the work.

File: platinum giga.jpg (1.12 MB, 1536x2304)

1.12 MB JPG

Big Brat

>>11366683
Still very new so don't even know what a lora is, but I'll look into that!

File: file.png (3.02 MB, 1856x1280)

3.02 MB PNG

>>11364276
Here, I've just cooked a napalm fart for you, anon. I've decided to take on your request since nobody wanted to do it.
>>11365500
Giant men are totally allowed: no need to bitch about this.
>>11360768
>>11365638
>>11365633
These are pretty solid videos. I bet you're using img-to-img Wan 2.2 for these, aren't you? Can you do a NSFW version?

File: Hotaru Shidare's Stomach.jpg (401 KB, 2432x1664)

401 KB JPG

How much candy could a Giantess eat?
>>11366751
>lora
Those are a thing for Stable diffusion models; aka via Local or CivitAI mostly.
they are plug-ins of sorts, specifically to bring something into the 'knowledgebase' of the model, Like artist styles or a better grasp of vore.
civitai is also a good starting point to hunt for lora's other people have made.

File: IMG_20250830_012512_388.jpg (160 KB, 876x1280)

160 KB JPG

>>11357045
Can I request an Ochiko style image of Tomoe Mami from Madoka Magica or Hatsune Miku of Vocaloid fame?

I'd love to have POV shot like this.

>>11366818
Oh man it managed to get Hotaru's design down perfectly. She's such a perfect giant pred and if you could gen more of her that'd be amazing.

File: Hotaru Shidare's Stomach (...).jpg (450 KB, 2432x1664)

450 KB JPG

>>11366907
It's using a lora I spotted during my occasional check of Civit. I knew I had to gen her with a stomach full of candy.
I also made this onkinky styled one before moving on to other subjects.
I'll keep her in the back of my mind for when I next do something vore related, though no promises.

>>11366915
Nice. The background looks nicer on this one but >>11366818 came out way better overall imo. I just checked Civit and looks like it's blocked in my country so I'll just have to keep checking the thread for any more of her.

File: sakura mad.jpg (1.44 MB, 1536x2304)

1.44 MB JPG

Sakura got a Black Iced Tea when she wanted a Black Iced Tea Lemonade.

>>11366830
This pic has to be one of the hottest things I've ever seen

>>11367268
fucking diamonds over here
keep it up anon
can you make one where she reaches towards the viewer with her hand getting larger as it gets closer to viewer and just fades to black

>>11367206
yeah i wish who ever uploaded it included the meta data for it.

>>11367139
God, I love these bratty scenarios so much. I need a sequel where we see her from the ground as she actually pours her drink onto the Starbucks, crushing the entire street and flooding the rest of the city.

File: Tnbt.mp4 (3.74 MB, 1070x1080)

3.74 MB MP4

>>11367511

File: die idiots.jpg (2.01 MB, 1536x2304)

2.01 MB JPG

>>11367697
No need to worry about the refund, they can just have the drink back

File: gwensqc06.png (736 KB, 1000x696)

736 KB PNG

Eyh I know this is not a slop critique thread but I'd love if some more experienced people could point out some of my mistakes, I don't have much of an eye for it yet but I'm learning.

File: Yua-a02.mp4 (1.4 MB, 984x1440)

1.4 MB MP4

I play Pot of Greed! It allows me to draw two tits onto the battlefield!

>>11368386
>himecut
>huge breats
>glasses
>chocker
>small top
hhhhnnggg

>>11368007
>>11367697
What loras do you use to make these? The style reminds me of Yilx.

I know this isn't the place to actually post it, but does anyone know where to find decent photorealistic stuff? Specifically tall-to-minigiantess stuff, women in that range interacting with normal sized men, maybe growing too if it can manage it. I remember when video gens first started appearing I saw some photorealistic ones and a couple of them were really good, but I've seen very little of them since and wouldn't know where to look. I remember there was a general on /gif/ but it didn't have any of the right sort of thing, mostly just horrendous middle-aged fat women as I recall.

File: file.png (2.01 MB, 1216x832)

2.01 MB PNG

>>11368911
They may have used Yilx, indeed, although it's well-concealed.
>>11368987
There a couple of TikTok AI-focused accounts who create videos and mages about giantesses and size difference.
>>11368196
I think this is pretty solid. Local SD users might give you more tips and hints to guide you to improve further.

>>11368553
>short hair Korra
bueno.jpg
still wish she turned giant against the giant robot
>>11369076
MOAR

File: magic.jpg (1.17 MB, 1536x2304)

1.17 MB JPG

>>11368911
NSFW-illustrious-SDXL is the checkpoint. Artist tags are yilx:1 and wamudraws:0.5, and the only lora I'm using is 'Tiril ( KeyHat / キー帽 )' firstly because I'm a freak but I ended up liking what it does for shortstack/oppai loli proportions.

>>11369076
Sex with Aqua on Sylvia's dick

>>11369076
What sort of tags did you use for this? I want to shrink some giant waifus.

File: 1741240226855227.png (1.5 MB, 1152x768)

1.5 MB PNG

The best ride at the park.

File: 5e2b15a0-3f4a-4320-8bfe-7(...).png (3.11 MB, 1856x1280)

3.11 MB PNG

>>11369147
>>11369228
Here, anons.
>>11369229
Since I made all these images with NovelAI, you can check the prompt by yourself by uploading the pics on novelai/inspect.

File: Hotaru Shidare interrupts(...).jpg (390 KB, 2688x1536)

390 KB JPG

Did this on a whim,
mostly to check if I can still make these kind of shots.

File: Is this the human hive.jpg (353 KB, 2048x2048)

353 KB JPG

>>11369652
Also a random whim
Big Bee

>>11369652
Looks great. The only nitpicks I can see are the vampire teeth and the black nails when Hotaru's are usually red.

>>11369226
top tier shortstack/oppai loli proportions
>>11369530
anon pls already diamonds
mind if i request futa Aqua fucking a smaller female Kazuma while still on top of Sylvia
mfw Sylvia would be really gentle with a tiny person when they tell her they love her

File: paul.jpg (1.5 MB, 1536x2304)

1.5 MB JPG

Butte

File: paul2.jpg (1.11 MB, 2456x1384)

1.11 MB JPG

>>11369840
+1

>>11369530
Now it just needs a giant Wiz about to give Sylvia a blowjob/paizuri

Two quick tips for people wanting to do growth in a generated video:
1) If you have them move towards the camera at the same time, you tend to get better growth (Though this can also trick it into just tilting the camera, so it's not a surefire thing)
2) First Frame/Last Frame is going to get you better results, but you need to be overly descriptive in the prompt to get it not to do a fade or dissolve or camera movement instead of growth. Using a tracking shot with a chase scene where the one in back gets bigger tends to work most consistently, and the AI doesn't care as much about what the background looks like.

>>11369883
I'm really like the kissing retro anime waifu series. You do good shit.

Maybe one day AMD GPUs can make videos too but for now I'm stuck with images and online gen vidyaslop.

File: big lilith.png (3.2 MB, 1600x2000)

3.2 MB PNG

>>11370085
AMD really needs to step up their game
was able to get Rocm (9070XT) on linux working i think but renders take so long especially compared to my 5070ti
well after first render the others go faster
got their sample video workflow to work but that also took a really long time

>>11369076
>>11369530
Ok ok. Good,very good.
Now have aqua feed him to slyvia's cock

File: 1729495225013637.png (1.04 MB, 1024x1024)

1.04 MB PNG

Helping big women nut has been on my mind.

>>11369652
>Hotaru maw
Didn't know I needed this. Hot.

Trying to prompt novelai to create an image where the character is standing like directly on the viewer so that her foot isn't exactly visible, just her toes framing the sides. Can't seem to get it to work, anyone got tips?

File: Satisfied Gwen.png (3.62 MB, 1088x1920)

3.62 MB PNG

I didn't hate this NAI gen, so here it is.

The furry model's better for this, and most size things, actually, because it has tags for size stuff instead of just a single one. Just make sure you have 'human' and the skin color you want in the character tags and it usually generates humans fine. Also, find a pic that matches the general idea you want, then use vibe transfer. It cares about what colors are in what spot of the vibe transfered image, so a picture that's similar can really help with generating what you want.

>>11370613
Hmm okay sounds good. Although not sure how to find the furry model.

>>11357045
>Any artist style is fine, you can choose
>Blond adult male about to sit on a tiny 13 year old on the couch
>The blonde adult is average, the teenager is skinny and wearing glasses
>Preferably a couch POV but any POV works
>The teenager should be waving their arms or shouting as if trying to get their attention

>>11370654
Also
>Teenager is black-haired and brown eyed

>>11370650
Use "fur dataset" as the first thing in the general prompt, I also think there's a button near the general prompt that lets you swap

>>11370717
>>11370650
Useful tags: "Worm's-eye view" for extreme POV, "Giga" for giga sizes, "Landscape Dwarfing" for giga sizes that include a lot of tiny land, like ochiko's stuff. "Planet Dwarfing" for any tera-sized stuff, "imminent stomp" for when someone's about to get stepped on. Instead of using dabooru's tags it uses e621's so you can check there to see if there's any specific tags that'd be helpful for you.

File: 1739611455120170.png (1.85 MB, 1216x832)

1.85 MB PNG

>>11370131
Not that anon but since you mentioned that specifically I may as well post something I made a pair of weeks ago. Far from perfect, I dumped it in the "needs fixing" folder and forgot about it

>>11370917
While I'm not into vore of any kind, this kind of reminds me of one of the most /d/ things I ever saw. It was a futa getting a blowjob with a tiny person in the giver's mouth.

>>11370917
loving this. so if you can make a follow up, or just more like it please do

File: 1616832103794.jpg (403 KB, 1450x2048)

403 KB JPG

>>11357045
Requesting some vore stuff/mawshots of Klara from Pokemon in Karbo's style

File: he giant woman bends.mp4 (1.2 MB, 768x1122)

1.2 MB MP4

>When you walk into the villa and Cynthia surprises you, but you're not terrified this time

>>11371373
nice, Randy my beloved

File: 224b989e-6721-4ab4-8657-e(...).png (1.75 MB, 1216x832)

1.75 MB PNG

>>11370401
Try "3::worm's-eye view, foreshortening::, perspective, depth of field" in the main box and in the chaaracter box try "standing, standing over the viewer, knees". Avoid "full body" as it tends to include feet as well.
>>11370550
Pretty good! I'm looking forward to seeing more from you.
>>11370613
True, but since the release V4.5 Full e621 tags have been working also on the Anime model.
>>11370917
Solid gen! What I can only recommend is to try artist tags or artist mixes, as the output results in being usually more coherent and overall better-looking. If you feel like it, maybe you could also do the requests of >>11369800 and >>11369857.
>>11371190
Here, anon. Hope you'll like it!
>>11371374
Nice growth sequence.

>>11371513
Oh man aside from the teeth being a little messed up that turned out pretty much perfect. Thanks anon.

File: Determined Bono in Sleepy(...).jpg (467 KB, 1664x2432)

467 KB JPG

Nothing against Akebono, just an uma that came to mind.

File: 1756416465202813.png (747 KB, 1152x768)

747 KB PNG

Lewd interactions with a big Ruby Rose.

>>11371913
This is so hot. Great job.

File: part2.jpg (1.26 MB, 1664x2432)

1.26 MB JPG

>>11372105

File: part3.jpg (1.29 MB, 1664x2432)

1.29 MB JPG

>>11372107

File: part4.jpg (1.17 MB, 1664x2432)

1.17 MB JPG

>>11372108

File: part5.jpg (876 KB, 1664x2432)

876 KB JPG

>>11372109

>>11371375
>but you're not terrified this time
Excuse me but my dick is absolutely terrified, positively trembling even!

>>11371806

Moar big Ruby Rose!

File: jeep.mp4 (1.84 MB, 984x1440)

1.84 MB MP4

Been messing around with start frame / end frame animations. Couldn't convince vidu q1 to make any fast motions, and the other vidu versions just have no stability whatsoever.

Tried making other variations like tiny guy climbing the cat girl to get into the jeep but it only ever resulted in weird shit like him noclipping to it.

File: Sagi09.png (1.95 MB, 1536x1536)

1.95 MB PNG

Particularly liked how this degenerate slop turned out after much inpainting.

Also have a version where there is no pink swimsuit cup, just the netting but had a hard time decididing if I wanted the more dumbass design one or not.

>>11357045
requesting some giantess Gold Ship content with this body build >>11372360
close-up of female trainer (just a girl in gym teacher clothes or office wear) climbing a 50' tall GoldShip's boot or thigh with smug Goldie unaware
if possible pov shot from tiny trainer and camera looking upwards to see Goldies face (face blocked by breasts if frontal view)

I have an impossible time inpainting tinies, Any recommendations?

>>11373149
Paint an actual blob of color the vague shape of a person before inpainting, and reduce the strength of the inpainting to .8 or .9. This generally works with the majority or things you want to inpaint -into- a scene.

File: Inpaint Shit.png (1.25 MB, 1664x1617)

1.25 MB PNG

>>11373149
Like the other anon said, inpainting tiny people works best with Inpaint Sketch. You need to add the tiny's traits the best you can to the prompt, remove the giant girls traits that the inpainting area can't see, or both. This is because as you probably know, traits will try to bleed over onto other characters, but assiduous negative tags can also help alleviate that. So giving the tiny different traits from the giantess helps, as vague traits will just try to use the giantess's ones to create a tiny clone of the bigger person.

Inpainting WILL fuck up some small details and make it look inconsistent, but frankly the only way you're fixing that is either more tedious lower denoise inpaint around it, photoshop, or you get lucky. After you get a general character shape, you can use normal Inpaint to give it an extra detail pass and change things up better, and maybe fix some of the shit Inpaint Sketch fucked. The important part here is the AI will recognize something it made MUCH easier than nothing or something you drew in photosohp or whatever, and it becomes vastly easier to work with. So even if the AI outputs a vague human blob because the Sketch denoise was too low, it can still probably work with it just fine since the AI made it to start with.

These are the settings I used for the entire inpaint process; I removed most the actual image details from the prompt. More mask means it can see more of the image but the tiny will have less quality, less mask means the tiny will be higher quality but more likely to fuck up consistent details that tie it to the whole image.

>>11373354
Oh, and forgot to mentioning but giving your tiny a light outline (some brush opacity and softness) helps a lot.

File: he video should be in the sty.mp4 (433 KB, 768x1152)

433 KB MP4

>Which will it be? Your size back, or your company? I don't really mind either way...

>>11373460
Amazingly degenerate slop. I love it.

File: wan22_00247.mp4 (1.94 MB, 640x896)

1.94 MB MP4

>>11372854
gave it a go
>>11372864
really like this one

>>11373897
Sick, did you use any loras to make that? Every time I try to use Wan without a lora it does some hilarious but fucked animations.

File: 00090-317993150.png (3.15 MB, 1496x1800)

3.15 MB PNG

>>11373906
just used
GGUF loader: wan2.2_iv2high/low_noise14bQ6
lora loader: wan2.2lightning_high/low noise
took about 3 tries to get this one

File: ComfyUI_35556_.png (2.74 MB, 1160x2080)

2.74 MB PNG

>>11372854
absolutely love this framing and interaction, and the design of the girl in the background is great
how did you make the base image?

File: 1742821956658235.png (1.39 MB, 1152x768)

1.39 MB PNG

It's dangerous to guess your mistress' age, so undershoot to be safe.

File: EarlyVer.png (1.26 MB, 1856x1216)

1.26 MB PNG

>>11373957
Genned a bunch of face close-ups, cropped the width a bit and extended the canvas height then inpainted in the foreground stone floor (the original face close-up had her neck and I just drew over that shit), some quick img2img to blend them together better, then inpainted the smaller people in and whatever errors the previous inpainting caused. Those tiny stairs leading away unironically took as much time as the stupid jeep did.

I was originally trying to fuck around with mesoamerican architecture, like her peering onto the top of the Pyramid of the Sun or some shit, but I couldn't find a SDXL model or Lora that could do it. Just kinda left me to bullshit my way along with fancy prompting and inpainting.

Left was original base image before cropping, right was an early version after inpainting the ground and img2img with some slight inpainting to fix her hair and eye. Pretty sure I did a lot more img2img after this trying to not have a shit floor and then for some reason it added a stone circle pattern on its own.

I inpaint a lot. I've also done shit where I just inpaint a blank white canvas from the start, like I did with >>11368386

File: Sagi03.png (2.03 MB, 2112x1056)

2.03 MB PNG

>>11374067
This is another I did of the same theme but the perspective kinda got fucked adding what was supposed to be a normal Mayan stone platform that accidentally ended up too tall and a bit off in general. This stupid pile of rocks took more effort than the entire rest of the image and it still didn't come out great.

File: Screenshot_20250909_09462(...).jpg (1.09 MB, 1080x2316)

1.09 MB JPG

Would love to see a Boa Hancock with this outfit snuffing out a city under her feet.

>>11374172
Now shrink her

File: nano-banana-2025-09-10T09(...).png (1.06 MB, 1344x768)

1.06 MB PNG

Nano Banana has so much potential, can turn images into size content with ease and it's hard to even tell it's slop. Just need to find somewhere that has the model for free with no/minimal restrictions...

>>11374577
>And then she farts

>>11374016
Oh hey, a Wo-class. Finally some giant Abyssals for the thread. Got any others?

>>11371806

Need new big/giant RWBY characters like this.

Though prefer a proper depiction of the bigness instead of a white void, and actually giant instead of shrink.

Cramped indoors or outside with buildings/trees/whatever at the very least. Probably sound picky.

File: scary....png (1.39 MB, 1216x832)

1.39 MB PNG

Jessica's been acting strange lately... and eating a lot more... BUT. She's been doing much better during training, so whatever R&D's been doing to her is a total win! Very safe!

File: 00022-2650839363.png (3.79 MB, 1664x2688)

3.79 MB PNG

>>11375438
Just this other one. Just got into diffusion a few days ago and still learning. Will make more later, likely with Seaport Princess.

File: vpart8.jpg (1.16 MB, 2432x1664)

1.16 MB JPG

https://mega.nz/folder/GQpUWKqI#hEHFbilVfrHMSDiUWNPQnw

File: ezgif-706b0c1bcc9d83.mp4 (793 KB, 576x1024)

793 KB MP4

>>11375551

File: saku.jpg (1.33 MB, 1536x2304)

1.33 MB JPG

TFW everyone you know will never be mushed and turned into paste underneath the boot of a sadistic, childish Sakura stomping around for fun

>>11375544
Cock vore follow up maybe? Or condom fed to tiamat, medusa, or barqhest?

File: 5527a9d7-7be6-4dcc-be3c-2(...).png (2.01 MB, 1856x1280)

2.01 MB PNG

>>11376055
>Cock vore follow up maybe?
Here, my friend!
>>11375597
Cute!

File: Beamax.png (3.92 MB, 2688x1536)

3.92 MB PNG

Big Bea, part of some recent playing around with foreshortening.
I should see how good Illustrious can do cosplay, maybe I can do Twin Turbo cosplaying Iono

>>11376128
And then she fucks giant Yunyun

File: dkhzpr1-7ff1b8f6-2405-4ac(...).png (1.34 MB, 864x920)

1.34 MB PNG

>>11375551

File: giantess_blakes_breast_cr(...).jpg (129 KB, 1081x739)

129 KB JPG

>>11376396

File: willows_little_world_by_m(...).jpg (136 KB, 1000x800)

136 KB JPG

>>11376397

File: the_giant_threesome_of_we(...).jpg (107 KB, 824x1104)

107 KB JPG

>>11376398

>>11376092
Aww, thanks anon! I'm planning on making a comic, so stay tuned >:] (Also, really love your Astolfo expressions! Really cute and I like how it keeps the flesh fangs he usually has omg)

Is local generation worth pursuing? I want to try setting it up, but how powerful is it really compared to the online ones?

>>11376736
online is as good as you pay for
local is as good as your hardware

>>11376638
>>11376639
damn these are really nice. wish i knew how to inpaint and such.

File: Another Bea.png (3.87 MB, 2688x1536)

3.87 MB PNG

>>11376736
All my gens are local with 12gb of NVIDIA Vram.
lower Vram generally just makes it slower. Unless you can barely load the ~6Gb StableDiffusionXL model at all. Some front-ends will still try with little Vram but slow to a snails pace, I get that somethings when Windows decides to do a Windows in the background and available Vram becomes less than 4gb

A simple gen (not pic rel) I just did had these stats:
Time taken: 35.5 sec. Active Peak: 7.07 GB, Reserved: 8.29 GB, Total system usage: 10.1/11.9941 GB (84.4%)
this is including an adetailer pass, and while playing a YouTube video.

>>11376092
Nice ty anon.

>>11376128
Super lewd.

>>11376736
It's worth it, but it's also dangerously addictive once it's set up. Make sure you configure it to save things as jpgs unless you want to buy more hard drives.

>>11376736
>>11376835
A couple more things:
>put models on an SSD so they can be swapped/loaded faster
>use a RAM disk program to create a fake drive to serve as the output folder for your gens, so you don't waste SSD write cycles (manually save the ones you want to keep to actual storage of course)