Roka3D
An image-to-3D and text-to-3D webapp focused on making miniature figurines for TTRPGs.


I specialize in building software that bridges creativity and technology. From node-based AI IDEs to 3D tools and ML models, I focus on clean architecture, practical tooling, and shipping real applications with real users.
Selected software and tools I've built.
An image-to-3D and text-to-3D webapp focused on making miniature figurines for TTRPGs.

A node-based integrated development environment for building AI-enabled applications and webapps. ComfyUI integration, OpenAI/Anthropic/Ollama support, custom nodes, community center, native GUI apps, and an intuitive web server.

An experimental OS-like workspace for executing GUI apps made in Neu.

Fine-tuned models and LoRAs for image editing and generation. Click original/edited cards to compare before and after.
A LoRA that greatly increases the clothes transfer capabilities of Qwen Image Edit with a focus on maintaining the user's body and accurate depiction of clothing.
View on Civitai


A LoRA that takes an image of a person and outputs the outfit extracted to a white background.
View on Civitai





Fixes the age-old problem of people taking a photo of their screen instead of just using a screenshot.
View on Civitai





Helper for getting more accurate pose transfer with Qwen Image Edit.
View on Civitai





A LoRA that corrects Z-Image Turbo’s tendency to produce an overly polished “actor face,” yielding more natural and realistic depictions of women.
View on Civitai


A pixel art LoRA for generating Game Boy Advance and DS-era Pokémon trainers.
View on Civitai


Email or LinkedIn — say hi.