Work in progress, there will be more versions coming out to address issues
VERSION 1.1
This model is intended for a photorealistic style (still works with anime) and uses categories like periledascent676‘s Minigiantess SD to create giant women from 10ft tall to kilometers tall. The trigger words are mesogts for 10ft to about 30ft and macrogts for 30ft to mountain sized.
You can use the showcase pictures prompt as a template and change it to your liking. The images from the version 1 gallery are good prompts too for this. Bear with me I will be adding many more with cool new poses and contexts in the gallery! 😎
I recommend using Easynegative with this unless you’re really good with negative prompts, also Albedo base is giving me good results. But sometimes if a scene doesn’t render well try changing the sampler or checkpoint. These are only recommendations, experimenting is key. I you have trouble doing something specific, ask me in the comments, chances are I can help you.
Credit to katsumend for the idea of using the categories ✌️
Credit to stapfschuh for prompt tips that really make a difference and help you fix your image, go read his articles!
Description
I have added a whole lot of training on top of what was already there and I attempted fixing issues like a lack of interactions, boyfriend and man words being confused by the ai. Hopefully it will be better after this update. Have fun and any feedback is greatly appreciated as long as it’s not aggressive. ✌️😎
FAQ
Comments (15)
just a suggestion from my side: prompt "giantess-pov" to get the same result.
it can be useful for training your lora.
have already published dozens more prompts. maybe you can find something useful. just check my articles if you want.
Thank you but what do you mean by get the same results?
I see what it does thanks for the trick. Works a 100 times better if you prompt that with the lora though. Scaling is alot less stubborn with it. Depends on what u wanna do though
@Gts_Fox you mean you use that prompt on your lora? that's what my actual goal was: testing it together with this prompt.
@Gts_Fox i am a prompt guy and my brain is overloaded with ideas. ask me if you need help! i will try to find a solution. on anything
Thanks for the shoutout! Been following this pretty closely. Can't wait to try out the new version.
I noticed it’s mostly better for real giantess action like crushing buildings or cars, almost everything else will need a little prompt engineering to achieve good results
Seems to show them wearing sneakers a lot even when trying to give them other footwear 😕
Thanks for your feedback, sorry about that. Did you try weighing your prompts like: (high heels:1.2)? I usually don’t have a problem overcoming that with some weighing.
found out while prompt testing that "low-angle" generates very good size differences. maybe you can use this for your model.
and, just for fun, i added "on-a-city-rampage" and got exactly what i was hoping for :D
I’ll try it more, I actually put a couple "low angle macrogts" as captions in the training pictures
Hi thanks for creating this! Curious, I can't seem to download the training data, even though I see it
Thanks for the feedback, actually the training data is private, maybe in a future version it will be public. it’s 363 giant women images 95% from dall-e 3, size varying from about 9ft to larger than a town. Captioned with mostly 1 sentence per image but v1.2 will come out soon with a different captioning method as a test.
a few minutes ago i tested a pov-variant together with gigantic animals and had directly your model in mind during prompt testing. and here they are:
"first-person-pov" + "gigantic-woman"
during tests i used "gigantic-woman-on-city-rampage" for size comparism. will upload one image to show you the result. maybe these prompts can help you.
Thanks I’ll try it out
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.