Sergey Tulyakov (@sergeytulyakov) 's Twitter Profile
Sergey Tulyakov

@sergeytulyakov

Director of Research at Snap Inc. Tweets are my own.

ID: 888203246266753024

linkhttp://stulyakov.com calendar_today21-07-2017 01:05:30

201 Tweet

1,1K Followers

104 Following

Kevin Chih-Yao Ma (@chihyaoma) 's Twitter Profile Photo

GenAI Media Generation Challenge Workshop #CVPR2025 is today (6/17): 📍 Summit 423-425 ⌛️ 1:15 - 5:40 pm ⌛️ We have exciting keynote speeches from Jun-Yan Zhu (Jun-Yan Zhu), Sergey Tulyakov, Richard Zhang (Richard Zhang), Yuanzhen Li, and Tim Salimans to share the latest progress on

Sergey Tulyakov (@sergeytulyakov) 's Twitter Profile Photo

Stop by our #CVPR2025 posters to meet the team! We present 7 poster today, 2 papers are highlights. Video generation, 3D scene generation, 4D generation, improving quality of synthesized images and more!

Stop by our <a href="/CVPR/">#CVPR2025</a> posters to meet the team!

We present 7 poster today, 2 papers are highlights. Video generation, 3D scene generation, 4D generation, improving quality of synthesized images and more!
AK (@_akhaliq) 's Twitter Profile Photo

Snap presents VIMI Grounding Video Generation through Multi-modal Instruction Existing text-to-video diffusion models rely solely on text-only encoders for their pretraining. This limitation stems from the absence of large-scale multimodal prompt video datasets, resulting in a

Sergey Tulyakov (@sergeytulyakov) 's Twitter Profile Photo

At Creative Vision, Snap Research, we’re excited to announce our 2025 call for interns. Our past interns have built some of the world’s best image, video, and 3D generators. We’ve introduced new personalization methods and developed the world’s fastest mobile foundational image

Sergey Tulyakov (@sergeytulyakov) 's Twitter Profile Photo

Future of video generation and editing! Videos differ from images in that they have an additional dimension - time. We know about curse of dimensionality, but in this case it's a blessing! It unlocks so many new ways to be creative!

Moayed Haji Ali (@moayedhajiali) 's Twitter Profile Photo

Where are good old progressive diffusion models? 🤔 Breaking generation to multiple resolution scales is a great idea, but complexity (multiple models, custom diffusion process, etc) stalled scaling. Our Decomposable Flow Matching packs multi-scale perks into one scalable model.

Willi Menapace (@willimenapace) 's Twitter Profile Photo

Why is progressive generation so complex? 🤔 It doesn't have to be. Our Decomposable Flow Matching (DFM) simplifies the process into a single, straightforward flow model, 🚀 beating prior work in image and video synthesis. #AI #Research #MachineLearning