Adithya Iyer

To a large extent - yes, we wanted the model to use as much context from the original video for consistency of subjects/movement. Logically, it's easy to train one which can fix issues in the original video too, but for the purposes of this work: subject consistency with source > aesthetics , in terms of importance.

@adithy2 Assuming quality of the Reshot is wholly dependent on input video quality?

Introducing Reshoot-Anything 🎦, our method to change perspective on video, while being trained largely on *in-the-wild* videos 🧵1/8

@hckinz Hi! Vista4D was concurrent work to ours, and takes some different arch decisions, which are quite interesting in itself. Our method builds on Wan2.2, and more so tackles this pressing issue : can you scale a vid2vid by using raw footage, using very scarce synthetic data

@adithy2 Is this same as Netflix vista 4D which they released recently, it does virtual reshooting without physical camera rigs and based on Wan 2.1 👀

Weights, code and links: Paper : arxiv.org/abs/2604.21776 Website : adithyaiyer1999.github.io/reshoot-anythi… Github : github.com/morphicfilms/v… 8/8

Work done with my awesome collaborators: @avinash_plwl ,Shivin Yadav, @not_ali_afridi, @HarikumarMidhun. Special shoutout to @jdkanani and @morphic for the support, compute and inspiration! 7/8

vision🍌 is here vision-banana.github.io if you got into computer vision the way I did, starting with pixel-level labeling tasks like segmentation, edges, depth, or surface normals, you’ll probably feel the same seeing these results -- something big has quietly shifted, and it’s going to change how we approach these problems for good 🧵

@TongPetersb @sainingxie @ylecun @mengyer @YiMaTweets @LukeZettlemoyer @liuzhuang1234 Congrats peter!

I defended my thesis today! Sincere thanks to my advisors @sainingxie @ylecun and committee members: @mengyer @YiMaTweets @LukeZettlemoyer @liuzhuang1234. I could not have wished for a better PhD life, and I want to thank everyone who was part of this journey. Slides Link: tsb0601.github.io/data/defense_s…

@sanghyunwoo1219 @amilabs congrats @sanghyunwoo1219 !

Building something new. I’m joining @amilabs from day one to work on world models with an amazing founding team. A bigger challenge, a bigger bet, and an exciting road ahead.

i’m joining forces with @ylecun and an incredible group of people to start AMI Labs @amilabs. AMI isn’t a conventional lab. we don’t intend to become one. a lot to say about why this moment matters, but for now we’re heads down building. join us: amilabs.xyz

🚀 Unlocking Standard Diffusion Transformers on Representation Encoders Why do standard DiTs fail to converge on high-dimensional features like DINOv2? 📉 We found the answer isn't just "more parameters"—it's Geometry. Introducing Riemannian Flow Matching with Jacobi Regularization (RJF) 📄 Paper: arxiv.org/abs/2602.10099

Neat! Been waiting for something like this

Hiring a Content & Narrative Operator to work with me at @morphic. - Someone young, obsessed with AI and startups - Uses ChatGPT / Claude for writing tasks DAILY You could be working remotely, but it’s very crucial that you live on Twitter. 😺 Just DM me. Or send this tweet to someone you know.

Inside @morphic you can extract frames directly from any grid images generations Saves the mundane task of manual cropping this is how it should be!

What if Odyssey was reimagined as an anime? Inspired by the timeless epic, this is our anime interpretation of Odysseus’ journey. Same myth. New visual language. Made entirely with @morphic.

Turns out Nano-banana pro can do ambigrams, and I'm truly blown away