SchmusRealism: Shininess
Remove the typical plastic look of ai generated images.
Take the oily shine off skin, or add a natural glow, without changing the person. One number sets how shiny the skin is, from matte to oily. Face, pose, clothes and background stay how they are.
No trigger word. Strength −2 to +2. Needs the two sampler.
What it does
−1: forehead, nose and cheek highlights calm down. Pores and wrinkles stay. Best balance for retouching.
−2: fully matte. Strongest effect, with a slight skin colour shift.
+1: a natural, healthy sheen.
+2: oily, sweaty or post-workout skin.
Use cases
Clean portraits and headshots: remove shine without a retouching pass.
Flash and hard light: tame hot spots on the forehead and nose.
The + side: athletes, summer heat, dewy beauty shots.
Examples
Why not just prompt "matte skin"?
Krea 2 Turbo has no negative prompt, and in my tests a skin-finish phrase changed the skin tone and contrast rather than the shine. This slider was trained on the highlights alone, so it only changes the shine and works as a dial.
How to use
Krea 2 Turbo, 8 steps, cfg 1, euler / simple, with two samplers:
Sampler 1: steps 0 to 3 on the plain model. This lays out the picture.
Sampler 2: steps 3 to 8 with the slider. This changes the skin finish.
Set the amount with strength_model on the Load LoRA (model only) node.
Keep the change point (
start_at_step/end_at_step) at 3.With other LoRAs: Put the slider last in the chain on the sampler-2 model.
Strengths and weaknesses
Strengths
Moves real highlights in both directions while pores and wrinkles stay.
identity, pose and framing hold across the range.
Weaknesses (v1, please read)
Small details and backgrounds can change a little (a buckle, a distant skyline), most visibly at +1 / +2 on busy scenes and warm light. At −2 a street scene or body can shift too.
Non-skin can respond. Glossy objects may change slightly on high strengths, though much less than skin.
Thanks for checking out my LoRA and I hope you find it helpful in your artworks.
Feel free to leave some feedback or let me know what could be improved. 🫶