Diana Grodik
Generative AI model fine tuning and custom LoRA training
New to Contra
Diana is building their profile!
Stylized Character LoRA — Pose and Role Consistency
Stylized character trained for high pose variability, action scenes, and role-based outfits while preserving a recognizable character identity.
Well-suited for game development, character design, visual novels, and animated...
Consistent AI Character Creation – “Rock Star Masha” (LoRA-based)
This project demonstrates a fully consistent AI character built with custom LoRA training.
Masha is a virtual rock-inspired character designed to maintain facial identity across different:
– outfits
– hairstyles
–...
Consistent Virtual Model - Multi-Scene Character Generation (LoRA)
This project demonstrates a custom-trained LoRA for a single realistic character, maintaining strong facial and identity consistency across multiple scenes, outfits, poses, and lighting conditions.
The model was...
High-Consistency Character LoRA for SFW & NSFW Projects
A custom character LoRA designed for projects requiring strict identity consistency across stylistic extremes, including dark themes, sci-fi aesthetics, and NSFW-safe production.
The pipeline supports controlled variation...
Hi!
I’m a Generative AI engineer specializing in custom LoRA training and model fine-tuning for Stable Diffusion (SD 1.5 / SDXL).
I help teams scale visual production when resources are limited.
• If you have a small art team, I can turn a single character into a consistent image...