HEY Lots more at https://sliders.ntcai.xyz ($5 a month that gets dumped back into GPU costs). Leave a comment if you want me to publish a specific one here.
Give your images flat even illumination (negative) or dramatic chiaroscuro lighting (positive). No trigger required.
The notrigger code has been completely redone and this is a lora 16 resulting in better quality than prior releases.
Trained on https://civarchive.com/models/215796/top-rated-turboxllcm-super-mario-merge
Support my work on patreon and get access to SO MANY loras at https://sliders.ntcai.xyz! Plus, easily train your own - no hardware needed!
Any proceeds go to more gpus- support what you enjoy and enable my addiction!
Description
FAQ
Comments (18)
WOW, what a great Lora, just what I've been looking for all this time! Light shadows are edited, otherwise the Lora does not change the picture any further, very good work, thank you very much!
thanks
Really cool, does exactly what you'd hope for and doesn't seem to mess with the image much. Works especially well with Cyberreal checkpoint.
Definitely wish I found this sooner. Especially useful on Civitai since its generator tends to give a lot of bright images or light up the source, (person). I'm still learning stuff, but this is a definite must have for dark images.
Can you adjust the settings to use values higher than 1 on civitai? Up to 2 would be enough.
stopped working with pony for me all of a sudden
Updating: Working as before now! Seemed to be a site wide problem.
What's going on? Not working for Pony anymore?
No longer working with pony suddenly?
Hey
What types of photos did you use for your training? Did you use high-contrast characters, buildings, locations, pets, cars, etc .?
Seems he dont wanna tell us... But seems not high quality.
@MrSmith2025 I think so. I was trying to make a LoRa with dark, high-contrast images, 1024x1024 resolution, but I couldn't get the desired results.
@Noah150996 These days resolution dont care. You can put in every kind of image ratio. The right captions (best hand written) are the way to go for a decent result. Did u made good captions?
@MrSmith2025 I trained a new parrot using images of a character and modified them with low light quality and a vignette effect, and it worked perfectly.
I used 10 images, all with a resolution of 1536x1536.
I left all the images dark so the character's face is barely visible, but they look excellent when generated.
I haven't used that labeling method. I use the default one.
Thanks, this is awesome.
Can you add more range on civitai to go darker? Up to 3 would be good. It doesnt go dark enough.
Works very well!
Thanks ❤️
This sort of LoRA is still one of the best ways that I show new users how to use a technique LoRA to tell the AI what NOT to do... using this with careful negative weighting is a beautiful way to lighten a picture with the same smooth result as a proper chiaroscuro normally darkens it. Excellent.
Details
Files
Available On (3 platforms)
Same model published on other platforms. May have additional downloads or version variants.