If you enjoy my work please leave a like or a review. Pics of your work are also appreciated :)
Posting the images directly to the lora page also helps a lot.
Pony instructions:
Use the main tag along with the style you want.
'latexHood' is the main trigger and should always be used, you can add a color before it to change the hood color.
'noHair' will give you the classic latex hood look.
'twinTailsHair' will give you the twin tails hairstyle, can use a color before it to change the color of the hair.
'ponytailHair' will give you the ponytail hairstyle, can use a color before it to change the color of the hair.
Adding 'nostrils' to the negative can help improve the results too.
There are two other tags that don't work most of the time but if you are prompting to that they might help 'noEyes' and 'noMouth'.
Illustrious instructions:
Use the main tag along with the style you want.
'latexHood' is the main trigger and should always be used, you can add a color before it to change the hood color.
'noHair' will give you the classic latex hood look.
'twinTailsHair' will give you the twin tails hairstyle, can use a color before it to change the color of the hair.
'ponytailHair' will give you the ponytail hairstyle, can use a color before it to change the color of the hair.
'noEyes, noMouth' will give you a faceless hood.
It currently has a rather unfortunate (or fortunate) side effect of turning some outfits to latex if they aren't prompted, mostly happens if you only put character name. The 'ponytailHair' tag is a bit hit or miss, changing it to 'single ponytailHair' can help.
Flux instructions:
Flux is awesome just go wild with it. Even without using the hood the lora is great for improving the latex look in flux and giving the subject bigger busts. If you get hair in the images you can add "no hair" to the prompt to easily fix that.
SD 1.5 instructions:
Currently only supports one style.
This Lora is a bit finicky and I will need to revisit it later but I found it worked better than most of what is currently available.
Some tags can clash with it by trying to "fix" the look, so if you get bad results try playing a bit with removing things such as masterwork, disfigured, ugly, etc...
The recommended strength is about 0.6.
Description
FAQ
Comments (9)
Because you installed the main version of FLUX, your LORA is no longer searched for under Pony in the "add resource" tab
Huh, well that seriously sucks
This LORA works very well in Swarm UI, but unfortunately because it is some unusual format it does not want to work in Invoke UI. Unfortunately, it is a great pity because Invoke UI makes working with Flux much easier thanks to its friendly graphical interface.
If you are looking for a good friendly ui for flux give Forge a try.
I tried, but Invoke is on a completely different level and I've been working with it for a year now. When I work with graphics I like to have a graphics program and not something that looks like Excel. That's the problem with most UIs, that they are created by programmers for programmers. INVOKE is a completely different philosophy of operation. More with creators and artists in mind.
Is there any chance for updated version?
@GordanGontaro Do you have any problems with other loras because I suspect it might be an issue on your end. This Lora was made in a very standard way. I wouldn't even know where to begin to know what needs to be changed in the training for invoke tbh
@JaneB No. This is not a problem on my side. I have talked to the creators of Invoke AI. They are aware of this problem, but since it affects a small number of lora models that are trained in some rare format, they decided that the problem is not worth solving for now. If I at least knew how your lora was made (because it probably wasn't in Civitai), it might be easier to convince them by giving them on a plate what format it is or in what program it was trained or maybe this is some training settings issue.
By the way. Another your model latex catsuit work perfectly fine.
@GordanGontaro Pretty sure it was trained on civitai but it was a while ago so its hard to tell. Might have been an issue that was fixed since it was in the early days of flux but to be honest I am not seeing myself retraining the lora unless I make some significant changes to it which I currently dont have any planned.