Marks-Experimental Refiner V2 REDUX

Hey there!

This is my completely nude update to my first refiner, and since I've been having a lot of people ask about my generations and what I exactly do to achieve a polished look, here's something that should speed up getting a similar result. I'm going to be training another lora after this on all of my favorite/top generations I've done, to also get closer to my style for those interested!

This is trained on a batch of quality generations I've hand picked, each image carefully observed for any issues, and this is the result of it. Now, this is a mix of meh, BUT, for those who want to capture a theme similar to my style, this is the perfect mix.

Reccomendations:

Model value: (0.4-0.8) for most stable results

Can use between 0.9 and 1.3, however, it can become very unstable very rapidly, however, lower weights can reduce the overall effect on generations. 0.8 is by far the best way to use it, less if horrible, more if ineffective.

HAS A BIAS TOWARDS WOMEN, TRAINED OVERWHELMINGLY ON WOMEN

All generational parameters used for the images that were then used to make the videos, can be found in the tab at the video, and the upscaler used for them all was UltraSharp.