Light Control - Up Down for SD15
Suggested Weight: -5 ~ +5 (or higher)
Effect is more obvious in darker themes without external light sources (window, sun, moon, lamp, etc).
It is difficult to prompt light direction in SD, so I made this tool. If you like this tool and want to see more of them, please follow me and leave a positive review. Thanks😘
Positive: Add lightsource to the top
Negative: Add lightsource to the bottom
Extra tip: Would be better to use only OUTALL blocks.
Like what I'm up to? Feel free to treat me a Ko-fi. Cheers!
As my user name suggests, I am trying to find my NextMeal.
so I put a revised version(weight adjusted, stabler composition, smaller filesize, later epoch) on Ko-fi for supporters.
It has been months with negative balance sheet for me, because I've been struggling for too long to find a full-time job as a foreigner and I am living abroad with my partner. Your donation would really greatly help me out with my life. I am also open to commissions if you have some custom requests that I can help with, you can just leave your comments on this page or on Kofi.
Description
FAQ
Comments (7)
Nice work.
If you think just using the OUTALL blocks works better, it might be a good idea to post the block merged lora.
I suspect most people don't have the extension for block merging at inference time or even if they did wouldn't think to use it most of the time.
Thanks💪
As mentioned in the description, there’s indeed a modified version. Tbh I am trying to save my own situation in real life but I don’t want to kill the open source spirit here. basically people still have the same model, the only difference is the file uploaded here need a little bit extra effort. It is just the way how I compromise for the moment. If anyone gets to test it and like it, they can decide for themselves whether to make a donation or not.
@NextMeal Sorry I missed that part of the discription. I'm totally fine with people charging for content. I was just trying to make a helpful suggestion because I see a lot of people release loras and put things in the description like "works best at a weight of 0.4" or "use the following block weights" instead of just normalising the lora to work well at a weight of 1.
Also the question you asked me about training on 1.5 the other day disappeared before I could respond, but my "slider" models are trained using the difference in image pairs, rather than with LECO. Because it's image difference being trained, it doesn't matter if the text encoder doesn't recognise the concept.
@m99 oh I remember that now,. it has been a while, I almost forgot about that. 😂😂I see you are making some drugs there.
I've used this so many times on SD, especially on models that always seem to cast shadows on the bits I want well-lit, but I'm missing it badly now I've moved to SDXL.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.