fixed negative term, scale is now from -1 to 1 and no trigger words are required.
Support me and build your own sliders with just words at https://sliders.ntcai.xyz/
Example images at 1.0 strength and do not require a trigger.
Resources
https://github.com/rohitgandikota/sliders
https://civarchive.com/models/215796/top-rated-turboxllcm-super-mario-merge
https://sliders.ntcai.xyz
Support my work on patreon and get access to more of these loras with the ability to train your own. https://patreon.com/NTCAI
Description
FAQ
Comments (18)
testing it out (using socarealism merge my go to) it seems to have almost no effect even at a strength of 5
hi please include the term 'attractive'. it was also trained on 'ugly' in the negative.
If that doesn't work, import one of the example images into comfyui and use the setting / toprated model from there.
the changes are a bit subtle but definitely there, -1 made girls way more "basic" looking and 1 improved the pose, background, and made her slightly more beautiful. In short: I disagree
@Xavia91 ok I'll keep playing. I also was mostly experimenting with inpaint, maybe that's an issue
Haven't tried it yet, but man, I wish if every model had the same type of thumbnail as this. Gradually labelled, showing before-after, not too fast, not too slow, just perfect. Way better then those 2.5D "moving" sh*ts. Thanks for making it like that. This should be a standard here on civit, lol
Thanks, I'm working on making it easier to use but the source code is available at https://github.com/ntc-ai/lora_eval
@ntc thats pretty neat, would be even greater if you could turn it into a auto1111 extension :D
@Xavia91 Thanks. I don't have plans to make an A111 extension because I am busy with trying to pay for my compute addiction. It's open source so anyone can pick it up!
This is making my invoke ai crash.
Hey I haven't tried that one. Is there an error message?
Forgive my ignorance but how does one use a slider?
Hey it's just a lora, use like any other with the trigger. Slider is just the name of a training technique to differentiate it from Kohya or dreambooth loras
Hello, author Da Da, how should this be used? I am using your LORA in Additional Networks, why is it not working?
hi It's a regular lora. Use the trigger attractive and strength of 3 or 4.
It's "<lora:attractive:3>"?
@VRman yeah that should do it in a1111 if you have attractive.safetensors downloaded correctly.
@ntc what is a1111?
@ntc Hahahaha......... sorry for the confusion, but I'm not quite sure if my WebUI is the A1111 version. It was just updated this month though.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.