Commander Valen (The Betrayer) | DOOM

I updated this model in order to achieve flexibility of this model. As I previously state before, this model is almost impossible to make flexible due to lack of dataset (heck even R34 isn't well rendered because The Betrayer R34 didn't existed, yet). So, I experimented with incorporating data science and data mining approach in order to reach some of support of clothing and R34. Well... this model supported for this. Here's some update changes:

• Decreased image usage for dataset from 60+ to 50+

  • Originally, I bumped up to 70+ images for this update. Upon inspection, this create dataset imbalance that could result undoing from what I want

  • Many of images on this model using kinda same-y style and blurred as hell. I use Valen/The Betrayer in-game screenshot too much that I need to remove it

  • Added some high quality and much better clarity for better char accuracy

• Fixing model mislabel during labeling

  • I found out that his scarf didn't labelled at all along with his armor

• Decreased resolution base trainning from 768 to 640

• Added support of clothing and R34 version of The Betrayer.

  • I still worked on for absolute flexibility for this model. So, I still updated this model soon, but not to soon as my roadmap of update/making LoRA is mountainous

  • Proper R34, unfortunately best works on weight of 0.7 - 0.75, which decreased rate of accuracy of his facial feature. For minimize this, properly tag Valen face features like 'beard, bald, old man', Please note that this thing may not accurate as checkpoint that I use is different with you. In this case I use, OhMenToontastic, Airfuck's Brute Mix, Photon, and Airfuck's Wild Mix.

• Simplified Trigger words just to one trigger, the_betrayer