Sign up
Log in
Sign up
Log in
Live Webinar: June 25th
Join our Builder's Roundtable to learn all about fine-tuning LLMs
Register now
Home
Blog

Introducing Stable Video Diffusion (SVD) on OctoAI

Blog Author - Janisha Anand
Blog Author - Michal Piszczek

Mar 7, 2024

4 minutes
Man with arms folded wearing sunglasses on in a city intersection with the camera panning around him

Production grade animation API for GenAI applications

We’re excited today to announce the launch of Stable Video Diffusion (SVD) 1.1 on OctoAI, empowering developers to easily add engaging animations and motion to GenAI-powered images. Stable Video Diffusion on OctoAI marks the start of a new commercial partnership between OctoAI and Stability AI, which grants access to Stability AI’s core models, including SDXL Turbo, Stable Zero123C for 3D object generation, and the highly anticipated Stable Diffusion 3. With animations in under 30 seconds and proven reliability delivering millions of daily customer inferences, Stable Video Diffusion on OctoAI is the first step in this journey.

You can get started with image animation using Stable Video Diffusion on the OctoAI Media Gen Solution, powered by Stability AI.

GenAI image animation to power next-generation visual experiences

OctoAI customers like NightCafe Studios generate and deliver millions of images to end users with Stable Diffusion and SDXL on OctoAI. And these customers are eager to do more with GenAI. And key areas of interest among several of these customers have been graphics and animation — the ability to make images come to life.

Stable Video Diffusion unlocks this demand, and we’re already starting to see hints of a new wave of creative visual applications emerging. Early application areas we have heard from customers include motion graphics for rich visual advertisements, entertainment, and creative enrichment for traditional photography.

Product walkthrough: OctoAI Media Gen Solution

To get started, head over to the OctoAI Media Gen Solution homepage, select ‘Image Animation’ card. Sign up for an account if you don't already have one, and get $10 of free credit.

SVD 1.1 serves as the designated engine for video generation. Explore this page to access an interactive demo, allowing you to adjust settings to your liking, as well as an API page for interacting with OctoAI endpoints.

Using the Advanced Settings on the left side, you can customize your video animation by choosing the number of steps, cfg scale, frames per seconds (fps), motion scale, and noise strength.

The API screen describes how to use our endpoints in your application. You can easily get started by copying and pasting the API requests in Python, TypeScript, or cURL, available in the UI. When you need to scale your application, all you have to do is continue sending inferences to the same API endpoint. OctoAI handles the infrastructure scaling needed to support your on-demand changes in usage, with no upfront pre-provisioning or service degradation.

You can also kickstart your video generation journey directly from the image generation card on the Octo AI Media Gen homepage. This feature utilizes text prompts to create AI-powered images. Simply click on the 'Animate' button located on any of the resulting images to generate a corresponding animation or video.

OctoAI is built to deliver predictable latencies with scale. Early testing shows image animation inference calls with Stable Video Diffusion on OctoAI consistently completing in 32 seconds, with p50 to p95 variance of approximately 0.5% (100 runs with default configuration of 1024*576 and 25 steps). This predictability ensures a consistent experience to end users as usage of your applications grows.

OctoAI and Stability AI partnership agreement, production-grade GenAI platform serving industry leading media generation models

Stable Video Diffusion on OctoAI builds on OctoAI’s strong track record serving Stable Diffusion and SDXL models. OctoAI’s ability to optimize and run diffusion models effectively delivers reliable, low latency and low costs to end customers, and OctoAI serves multiple customers generating over a million images on the platform every day. Thanks to a newly minted partnership between the two companies, OctoAI will be among the select partners allowed to deliver core models, including Stable Video Diffusion, commercially to customers.

The agreement includes access to Stability AI’s core models, including Stable Diffusion Turbo and 3D model Stable Zero 123C. Uniting OctoAI’s ML systems expertise with Stability AI’s established track record in model development will bring media gen customers a sound foundation on which to build and scale AI powered experiences for a variety of use cases.

In the coming months, the team will be working to integrate the full breadth of OctoAI Media Gen Solution capabilities with Stable Video Diffusion and broader Stability models. This includes the ability to fine tune the models, easily apply your choice of fine tunes at inference time using the OctoAI Asset Orchestrator, and pipelines to integrate generation workflows across multiple models in the solution.

I recently had the opportunity to test the new Image Animation feature, and I’m pleased with the results. The reduced generation time of 30 seconds significantly improves the user experience, making it a useful feature for our AI companion robots. I look forward to exploring its potential for marketing and sales use cases.

David Packman, CTO @ Packabby Robotics

Sign up for the OctoAI Media Gen Solution today

Sign up and start building today at no cost, with the free tier of the OctoAI Media Gen Solution.You can go on to build and scale commercial applications with our Pro tier. For specific SLA, performance, or deployment needs, contact us for details about our Enterprise tier. You’re also welcome to join us Discord and engage with our team.