A new character for the Cast. Dark hair, hazel-green eyes, that soft-edgy young streetwear vibe — space buns and colored clips are part of the signature but not locked in, she holds across other styling too.
Second Cast character trained on Krea2.
The story behind this one
This drop has a bit of technical background I want to share, both because it might be useful to someone and because it explains the texture June carries.
I built the training dataset using a specific model I love: Krea2-SAT-DirtyRealism by @Sateluco . That model has a grain, a light quality, and a way of framing poses that I keep coming back to — especially the grain. So I intentionally trained June on a dataset generated with it, hoping some of that character would transfer into the LoRA itself.
It worked. June carries that texture.
The ironic part: June doesn't pair perfectly with Krea2-SAT-DirtyRealism itself. Both the model and the LoRA push in the same direction, and stacked together the effect goes over the top — grain on grain, look on look. Dropping the LoRA weight to 0.7 – 0.8 compensates a fair amount, but not fully. Something I honestly didn't think about while building the dataset — I was too focused on getting the effect into the LoRA to consider what would happen when I put them back together.
The good surprise: June pairs really well with the two other Krea2 models I use regularly — GonzaLomo Krea2 V4 by @GBRX and CielBleu Krea2 by @cielbleu9991 . Both fantastic models, each with their own strengths, and together with June they take the best of the base model plus the grain texture I was chasing. Really happy with the result on both — not so much for the character itself, but for the technical outcome: I got the effect I wanted into the LoRA and it transfers cleanly onto the two models I love most.
Specs
Trigger word:
K_ScJun3_KSCBase model: Krea2
Recommended weight: 1.0 — sweet spot
Working range: 0.8 → 1.2
With Krea2-SAT-DirtyRealism specifically: drop to 0.7 – 0.8 to keep the grain effect in check
Don't go below 0.7 — character starts to fade
Personally I never push above 1.0
Notes
Stress-tested across a wide range of prompts and contexts — character holds, flexibility is there.
All preview images here were generated with June alone, no style LoRA stacked. What you see is the LoRA on its own.
Pairs naturally with the ShinobiSat Krea2 Workflow — same pipeline, tested together (link below).
The grain texture is a specific taste — I know it won't be for everyone. If you have thoughts on the effect, or on how June pairs with your favorite Krea2 model, I'd love to read them in the comments.
If you cook something nice with her, drop it in the gallery.
June is the seventh drop of the ShinobiSat Cast — a growing series of photoreal character LoRAs, now across multiple base models. More coming.
— ShinobiSat