Dan Ruta

Dan Ruta, PhD

ML Research Scientist @ Disney Research | Studios

Building the future of Generative AI.
Neural Style Transfer, Representation Learning, Speech, and High-Performance ML Engineering. Also a photographer documenting wildlife, landscape, and aerial work across the world.

Research

2025 • ICCVW

Leveraging Diffusion Models for Stylization using Multiple Style Images

Scaling diffusion models for state-of-the-art stylization with multiple style references.

Generative AIDiffusion
Read Paper
2023 • ECCV VisArt

DIFF-NST

Diffusion Interleaving For deFormable Neural Style Transfer. Strong form deformation via attention manipulation.

Read Paper
2023 • ECCV VisArt

NeAT

Neural Artistic Tracing. Style transfer for high resolution images via novel optimization strategies.

Read Paper
2022 • ECCV

StyleBabel

Artistic Style Tagging and Captioning. Multi-modal vision/language representation learning on a massive dataset.

Read Paper
2022 • ICCV

ALADIN

All layer adaptive instance normalization for fine-grained style similarity. Representation learning of artworks.

Read Paper
View Full Publication List →

Engineering

xVA Synth
3M+ downloads, 100k+ Users

xVASynth

Neural speech synthesis tool. Generates high-quality voice acting lines with fine-grained pitch/emotion control.

MLText-to-speechElectron
Steam Nexus Code
jsNet
High Performance

jsNet (2017)

Deep Learning library for Web/Node.js. Achieved 15x performance boost via custom C++ WebAssembly backend.

MLWebAssemblyC++JavaScript
Code
WebSight
Assistive Tech

WebSight

Real-time VR video augmentation via WebGL shaders (edge detection) to aid the visually impaired.

WebGLThree.jsShaders
Read Paper Code
vrscribbe
Productivity, Privacy

Tools

A collection of free, privacy-preserving web tools with no accounts required, no limits, and no sending of any data to the server from the browser. Fast and to-the-point, with modern browser tech and UI.

JavaScript
Website
vrscribbe
Personal fun

Food

An ad-free, life-story free, highly searchable database of well-labelled recipes from around the world with macros, step-wise timing, and tablet-friendly UI.

JavaScript
Website
vrscribbe
Personal fun

VR Scribble (2016)

An experimental browser based 2D 360 multiplayer virtual reality painting app, using 2 phones, WebSockets and WebVR.

Three.jsPug (Jade)Node.jsJavaScript
Website
View All Projects on GitHub →
About this site

Personal portfolio of Dan Ruta, PhD, Machine Learning Research Scientist at Disney Research | Studios. Research spans generative AI, neural style transfer, diffusion models, and browser-based WebAssembly tools.

Featured work

Research areas

Generative AI, neural style transfer, diffusion models, computer vision, and high-performance web tooling with WebAssembly.