Builders & open source
Nataniel Ruiz
AI researcher working on pre-training of generative models, Google DeepMind
- AI researcher working on pre-training of generative models, Google DeepMind (GenMedia team)1
Nataniel Ruiz is an AI researcher working on pre-training of generative models in the GenMedia team at Google DeepMind. Ruiz worked on large-scale pretraining and new capabilities for Gemini Omni, an any-to-any multimodal model with video generation. Ruiz is the lead author of DreamBooth, a method for personalizing text-to-image diffusion models from a few images of a subject, presented by Google Research with an official dataset repository. Ruiz's other work on image and video generation includes HyperDreamBooth, StyleDrop, ZipLoRA, RealFill, Magic Insert, and ReCapture.1234
At a glance
- Current role
- AI researcher working on pre-training of generative models
- Company
- Google DeepMind
- Sector
- Company
- Work listed
- 2
- Honors
- 1
- Sources cited
- 14
- Profile reviewed
Research
Ruiz's PhD research in computer science at Boston University concerned simulating synthetic data for training and testing neural networks, beginning with "Learning To Simulate" (2018), written with Samuel Schulter and Manmohan Chandraker. With Sarah Adel Bargal and Stan Sclaroff, Ruiz also studied adversarial attacks that disrupt deepfake generation by image translation and face manipulation systems.156
At Google, Ruiz was lead author of DreamBooth (2022), which fine-tunes text-to-image diffusion models to generate a specific subject from a few example images, and of HyperDreamBooth (2023), which speeds up this personalization with hypernetworks. Ruiz co-authored StyleDrop for generating images in a given style, ZipLoRA for merging subject and style LoRAs, and RealFill for completing images from reference photos, and led Magic Insert (2024), a style-aware drag-and-drop method for placing subjects into images.237891011
Video, games and multimodal models
Ruiz's work has extended to video and interactive generation. ReCapture (2024) re-renders a user-provided video from new camera viewpoints, and Unbounded (2024) is a generative game of character life simulation with open-ended mechanics produced in real time. Ruiz also worked on Camera Coach, a feature for Google Pixel phones announced at Made by Google 2025.12131
In Google DeepMind's GenMedia team, Ruiz works on pre-training of generative models and contributed to large-scale pretraining and new capabilities for Gemini Omni, an any-to-any multimodal model whose Flash version was released at Google I/O 2026.114
Work and contributions
DreamBooth Published research
First author, with Yuanzhen Li, Varun Jampani, Yael Pritch, Michael Rubinstein, and Kfir Aberman (listed with Google Research), of the 2022 DreamBooth paper; Google publishes the paper's evaluation dataset in the google/dreambooth repository.2341
Gemini Omni Proprietary product
Worked on large-scale pretraining and new capabilities for Gemini Omni, an any-to-any modality multimodal model with video generation capabilities, per Ruiz's homepage.1
Timeline
- Posted "Learning To Simulate" as first author.5
- Posted "Disrupting Deepfakes", on adversarial attacks against image translation networks, as first author.6
- Posted DreamBooth, a method for subject-driven fine-tuning of text-to-image models, as first author.2
- DreamBooth received a best paper award at CVPR 2023.1
- Co-authored StyleDrop, a method for text-to-image generation in any style.8
- Posted HyperDreamBooth as first author.7
- Posted Magic Insert, a style-aware drag-and-drop method, as first author.11
- Co-authored Unbounded, a generative infinite game of character life simulation.13
- Gemini Omni Flash, to which Ruiz contributed pretraining work, was released at Google I/O 2026.1
Honors and recognition
- CVPR 2023 Best Paper Award (DreamBooth) · IEEE/CVF Conference on Computer Vision and Pattern Recognition, 20231
Where they work
- Google DeepMind 64 people in the wiki
Related people
- Saining XieAlso at Google DeepMind
- Ian GoodfellowAlso at Google DeepMind
- Kaiming HeAlso at Google DeepMind
- Heiga ZenAlso at Google DeepMind
- Jascha Sohl-DicksteinAlso at Google DeepMind
- Jiaming SongImage, video, and 3D generation · Multimodal models
- Xu TanImage, video, and 3D generation · Multimodal models
- Amit JainImage, video, and 3D generation · Multimodal models
- Alexandre DéfossezImage, video, and 3D generation · Multimodal models
- Wei-Ning HsuMultimodal models · Image, video, and 3D generation
Suggest a correction or removal
Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.
References
14 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026
- Nataniel Ruiz — homepage
- DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation (arXiv 2208.12242)
- DreamBooth — project page (Google Research)
- google/dreambooth — DreamBooth dataset repository
- Learning To Simulate (arXiv 1810.02513)
- Disrupting Deepfakes: Adversarial Attacks Against Conditional Image Translation Networks and Facial Manipulation Systems (arXiv 2003.01279)
- HyperDreamBooth: HyperNetworks for Fast Personalization of Text-to-Image Models (arXiv 2307.06949)
- StyleDrop: Text-to-Image Generation in Any Style (arXiv 2306.00983)
- ZipLoRA: Any Subject in Any Style by Effectively Merging LoRAs (arXiv 2311.13600)
- RealFill: Reference-Driven Generation for Authentic Image Completion (arXiv 2309.16668)
- Magic Insert: Style-Aware Drag-and-Drop (arXiv 2407.02489)
- ReCapture: Generative Video Camera Controls for User-Provided Videos using Masked Video Fine-Tuning (arXiv 2411.05003)
- Unbounded: A Generative Infinite Game of Character Life Simulation (arXiv 2410.18975)
- Gemini Omni — Google DeepMind