When was the last time you saw a talk that had so many questions for the speakers that the speakers were answering questions for almost an hour after their talk? For us, it was last night at the OSS4AI event. Ari Korin and Christian Safka revealed the research around video #AI that we've been doing at Tavus. They talked about: - 2D vs 3D models - Neural Radiance Fields vs Gaussian Splatting - Scaling this kind of ML tech Want to catch the whole talk? Check out the YouTube link below.
Tavus
Software Development
San Francisco, CA 7,256 followers
Build with generative video APIs
About us
Tavus is a generative AI video research company that enables developers to build digital twin video experiences through easy-to-use APIs.
- Website
-
http://www.tavus.io/developer
External link for Tavus
- Industry
- Software Development
- Company size
- 11-50 employees
- Headquarters
- San Francisco, CA
- Type
- Privately Held
- Founded
- 2020
Locations
Employees at Tavus
Updates
-
Tavus reposted this
This week on America's next top model … 14 pre-seed/seed stage startups pitched at OSS4AI's Awesome AI Dev Tools event for July. Not an easy task to follow up two amazing presentations from Ari Korin and Christian Safka about Video #AI at Tavus and Yi Ding about the complexities behind existing frameworks that he learned while at LlamaIndex. Big congrats to: Wordware (YC S24) Ana Robakidze Fai Sukontanit WeCP | We Create Problems Jake Anderson Simon Farshid Alexander Ng Jovy Yu Aiswarya Sankar Aibek Yegemberdin Dhiren Mathur Oksana Riabichko Sachin Dharashivkar Sahil Manocha Catch the YouTube video in the comments!
-
We're really excited for our dev team to talk about their favorite parts of building in #video #ai See you tonight!
17 people will be presenting at the OSS4AI July meetup tonight. Two full on presentations: - Video AI by Ari Korin and Christian Safka from Tavus - LLM Frameworks by Yi Ding formerly at LlamaIndex 15 2-minute demos by: Wordware (YC S24) Theneo Beluga WeCP | We Create Problems BRXai Assistant UI Valyu GitRoll mutable.ai Entelligence.AI Toolhouse Adapt Botprise, Inc momentum sh Element AI TrojanVectors See you there #SF #AI community - RSVP below.
-
Talking to our users and customers is one of our favorite things. OH in an interview in the past week: “The quality of Tavus models is market leading. We get access to a world-class AI team and don’t have to worry about building models. This is huge! We have more resources to actually focus on the application side of things: the workflows, the user experience, and building features quicker.”
-
The field of avatar creation within video #AI has grown significantly over the last few years, with recent innovations vastly eclipsing earlier methodologies. Current pioneering efforts are predominantly driven by the growing sophistication of 2D Generative Adversarial Networks (GANs) like: - Wav2Lip - Write-a-speaker - DINet and the innovative 3D reconstruction and rendering techniques offered by Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3D-GS). We've not only kept up with these fields, but released our own research and our own model. Our groundbreaking model, Phoenix, demonstrates the capability of generating high quality full-body replicas that capture a broad spectrum of human appearance, expression, and emotion. Try it out for yourself by signing up at tavus.io today!
-
Having a digital twin that can do things on your behalf (with approval) is an exciting prospect. What's your dream use-case for deploying your digital twin via video? An interesting use case we've seen is doctor-patient communication - but maybe not in the way you're thinking. Our CEO and Co-Founder Hassaan Raza recently shared this example in a PYMNTS interview. We're seeing doctors using a digital twin to send out patient instructions before an appointment in multiple languages, building trust with a digital interaction first and reducing anxiety by communicating in the patient’s native tongue. What would you use a digital twin for? 👬
-
New: we have 10+ new stock replicas for your end-users 🤖 🤳 Selfie style replicas, great for social media 🖼 New backgrounds, including office & home See the full list of our updated replica library in our docs: https://lnkd.in/gnZe7KVY
-
"For businesses looking to leverage hyper-personalized, AI-generated videos, using a solution like Tavus via its developer APIs is relatively straightforward." Our Co-Founder and CEO Hassaan Raza talks to Austin Prey at PYMNTS on all things Tavus replicas and APIs. “It’s a night-and-day difference of what we can accomplish now with the realism, and what we can actually generate versus we could a couple years ago,” Raza said. Check out the video + write up here 👇
-
So how does our text-to-video replica model a.k.a. Phoenix work? The model can be broken down into four stages: 💬 Text-to-speech (TTS) 👨💼 3D reconstruction of the head and shoulders 😛 Script-driven facial animation 🎨 High-fidelity rendering Get into more nitty gritty details in our research paper https://lnkd.in/gQXdApP3
-
-
Livestream happening now 🔊 Tune into a conversation on all things Video + AI with our CEO Hassaan Raza and Wistia CEO Chris Savage https://lnkd.in/g_3DrXve Hosted by Sandhill Markets & Sacra.