Type | |
Stats | 6,390 6,744 |
Reviews | (485) |
Published | Aug 18, 2024 |
Base Model | |
Trigger Words | sideboob |
Hash | AutoV2 C4A49E869C |
.jpeg)
--- New versions for Flux are out, will improve when I have enough buzz again, Flux E3 is new (2024-08-25) ---
This LoRA should improve the appearance of sideboob in loose tops and shirts.
SDXL -> obviously for SDXL
Flux -> based on AI generated images with extreme sideboob, easy to use
Flux E3 -> all is not lost, save your failed image with Flux E3, more info below (new!)
Flux Photo -> based on photos, very hard to get decent sideboob, photorealistic/fluxlike
Flux Photo E3 -> Epoch #3 of Flux Photo, kind of a compromise between Flux and Flux Photo, bit easier to use
Prompt examples (FLUX)
a beautiful scandinavian girl posing in the mountains, the sun is shining, there is snow on the background, she is wearing a black and white top that shows sideboob
beautiful girl in an abandoned warehouse wearing hotpants and a loose top showing sideboob, cinematic light
Flux E3
Do you like your image, but it failed in some way, like flaws in the hands or weird nipples, all is not lost, you might save the initial generation with Flux E3. Best use is only when the original fails! =)
Suggested Resources
.jpeg)
Discussion
Somehow I used a zip file instead of the original safetensor, that's why it wasn't recognized in the generation data, should be fixed now! 👌
So much downloads and likes but no reviews, if you use the node Save Image With Metadata, your images will be posted in all used LoRA's, here is an example
https://civitai.com/images/25390580
Used almost the same training data as the original SDXL model and the results are a lot less clear. I also see this with other creators where Flux LoRA's barely do anything.
In some way this LoRA also gives better results on the most random prompts that have nothing to do with sideboob, or women at all ¯\_(ツ)_/¯
