| CARVIEW |
Johanna Karras
Welcome to my website!
I am a fifth-year PhD student focusing in generative AI at the Graphics and Imaging Lab (GRAIL) at UW.
My LinkedIn Profile
Curriculum Vitae
Download Resume
Research Interests
My current research interests lie in generative AI, specifically video and image synthesis. I am pursuing my PhD in the Graphics and Imaging Lab (GRAIL) at the University of Washington, advised by Ira Kemelmacher-Shlizerman, Steve Seitz, and Brian Curless.
Highlights
- September, 2025: HoloGarment: 360° Novel View Synthesis of In-the-Wild Garments is published on arxiv.
- June, 2025: I return to Google as PhD Student Researcher.
- April, 2025: Perturb-and-Revise: Flexible 3D Editing with Generative Trajectories is accepted to CVPR 2025 in Nashville, TN.
- December, 2024: I attended SIGGRAPH Asia in Tokyo, Japan
- July, 2024: Fashion-VDM: Video Diffusion Model for Virtual Try-On is accepted to SIGGRAPH Asia 2024 in Tokyo, Japan.
- June, 2024: I return to Google as PhD Student Researcher.
- September, 2023: DreamPose: Fashion Image-to-Video Synthesis via Stable Diffusion is published at ICCV 2023 in Paris.
- June, 2023: I start working as a PhD Student Researcher at Google.
- September, 2021: I start my PhD at the University of Washington in the Graphics and Imaging Lab (GRAIL).
- June, 2021: I graduate Caltech with a B.S. in Computer Science.
Research Projects
HoloGarment: 360° Novel View Synthesis of In-the-Wild Garments, Arxiv 2025
Johanna Karras, Yingwei Li, Yasamin Jafarian, Ira Kemelmacher-Shlizerman
HoloGarment synthesizes photorealistic 360° novel views of real-world garments in images and videos, even in the presence of occlusions, wrinkling, and significant pose variations. We introduce a novel implicit training paradigm with 2D videos and synthetic 3D assets in order to learn in-the-wild garment novel view synthesis without real-world paired data.Project Page | Arxiv
Fashion-VDM: Video Diffusion Model for Virtual Try-On, SIGGRAPH Asia 2024
Johanna Karras, Yingwei Li, Nan Liu, Luyang Zhu, Innfarn Yoo, Andreas Lugmayr, Chris Lee, Ira Kemelmacher-Shlizerman
Given a garment image and person video, Fashion-VDM synthesizes a photorealistic try-on video. We introduce a state-of-the-art video virtual try-on model based on diffusion, split classifier-free guidance, joint image-video training for try-on, and a progressive temporal training scheme.Project Page | Arxiv
Perturb-and-Revise: Flexible 3D Editing with Generative Trajectories, CVPR 2025
Susung Hong, Johanna Karras, Ricardo Martin-Brualla, Ira Kemelmacher-Shlizerman
We propose Perturb-and-Revise, which makes possible a variety of NeRF editing. First, we perturb the NeRF parameters with random initializations to create a versatile initialization. Then, we revise the edited NeRF via generative trajectories.
Project Page | Arxiv
DreamPose: Fashion Image-to-Video Synthesis via Stable Diffusion, ICCV 2023
Johanna Karras, Aleksander Holynski, Ting-Chun Wang, Ira Kemelmacher-Shlizerman
Given a person image and pose sequence, DreamPose generates an animation of the input person following the pose sequence. DreamPose equips Stable Diffusion with pose-and-image guidance, using a novel encoder architecture and finetuning strategy.
Project Page | Arxiv
Deep Neural Networks for Black Hole Imaging
- As an undergraduate at Caltech, I worked in with Professor Katie Bouman and Dr. He Sun on deep neural network techniques for black hole imaging. The project was presented at CVPR 2021. See the project and extended abstract in the Github Repo.
Internships
-
Google, PhD Student Researcher (Summer 2025)
-
Google, PhD Student Researcher (Summer 2024)
-
Google, PhD Student Researcher (Summer 2023)
-
Streetscope Inc., Computer Vision/AI Intern (Summer 2021)
-
J.P. Morgan, Software Engineering Intern (Summer 2019)
-
Microsoft, Explore Intern (Summer 2018)
Awards & Recognition
-
UW Reality Lab – Amazon Fellowship (2022)
-
Google Computer Science Mentorship Program (2021)
-
NCWIT Collegiate Award Finalist (2020)
-
Caltech Summer Undergraduate Research Fellowship (SURF) (2020)