sean safreed cofounder product owner cloud video
play

Sean Safreed Cofounder | Product Owner Cloud Video Authoring and - PowerPoint PPT Presentation

Sean Safreed Cofounder | Product Owner Cloud Video Authoring and Delivery Platform for X-Reality Storytelling (Virtual, Augmented, Mixed) Business model Software Platform Founded late 2015 Backed by Madrona, Vulcan Capital,


  1. Sean Safreed Cofounder | Product Owner

  2. • Cloud Video Authoring and Delivery Platform for 
 X-Reality Storytelling (Virtual, Augmented, Mixed) • Business model – Software Platform • Founded late 2015 • Backed by Madrona, Vulcan Capital, Raine, and angel funding Forest Key, CEO Bill Hensler, CPO Scott Squires, CTO Product Manager for V1 end-to-end Engineering lead for Adobe Former Lucasfilm VFX Supervisor and video stacks of both Adobe Flash and Creative Suite video product and technology lead, multi-film Academy Microsoft Silverlight: server, client, servers technologies, and Award nominee for VFX, recipient of authoring, DRM, partner ecosystems. photography tools at Apple. Early Science & Technology Academy Lucasfilm CGI artist. 2x prior media- digital video editing tools and Award. Founder of Dreamquest tech founder/CEO with public hardware pioneer. Images (now Disney) and Puffin company exits. Designs (now Avid). Sean Safreed, CMO Jim Eadie, CBO Aaron Rhodes, Exec Producer Graphics processing and pipeline Business and corporate Producer of numerous corporate film tooling veteran at Apple, SGI, and development executive at Viacom, projects for clients such as Adobe, HP, founder of leading motion graphics Time Warner, AOL. Architected Oculus, and Nvidia. CGI artist and and VFX tools vendor Red Giant digital, social and mobile supervisor credits include: Ironman, The (products currently in used by 200k+ monetization strategies. Avengers, Sin City, Harry Potter, Pirates artists). of the Caribbean, etc.

  3. Pixvana SPIN Studio An end-to-end platform for spherical 
 video processing in the cloud

  4. FOCUS ON STREAMING VR Mission Solve the issues that break immersions with VR video softness: 
 Lack of resolution, stereo acuity and color depth. Solution: Increase resolution where the viewer is looking at any time and decrease resolution where you are not, this is Field of View Adaptive Streaming (FOVAS).

  5. Equirectangular Image - 10K

  6. What You See in the HMD

  7. Field of View is Narrow and Resolution is Low Most HMDs have a 96° field of view

  8. Today’s 4K Resolution is NOT Enough Table 1 ~16K 4K 5.5K 8K 12K Horizontal 4096 5400 8192 12288 15360 2048 2700 4096 6144 7680 Vertical 11 15 23 34 43 Pixel/degree FOV 96 96 96 96 96 Resolution 1092 1440 2185 3277 4096 Now Late 2017 2019 2022 ? HMD

  9. OPF Projections and Viewports Viewboxes Viewports A projection through which a viewer A individual video stream that is optimized for a perceives a video stream in Virtual Reality given field-of-view within the entire spherical video

  10. OPF Mapping - Frustum Viewport video stream -------------- The final projection map onto frustum shaped OPF viewing room OPF Mapping - Frustum The large face at the back of the frustum has the lowest effective resolution (Y) Falloff Curve (X) (Z) Lowest Highest Resolution Resolution The small face of the frustum has the highest effective resolution, and includes high-res faces on the top/bottom and sides of The FOV and pixel density value The fall-off function, which affects The FOV and pixel density value X for center-view of the current Y the rate of decay of image quality Z for rear-180-view of the current viewport as FOV of (A) approaches FOV (C) viewport

  11. Frustum Projection Highest Resolution Image Area 2 5 2 1 4 1 3 4 5 3

  12. Processing Pipeline for 16K Jobs • SuperNova image processing framework • Extended the engine to run on Linux servers • Jobs are driven with a JSON-based DAG w/OpenGL functions • A single GPU shader drives the projection process

  13. FOVAS Process on AWS HD GPU HD HD HD HD HD HD HD HD HD HD HD HD HD Hybrik HD 10K S3 Projection HD HD Encoder HD HD HD HD HD HD HD HD HD HD HD HD HD Pixvana SPIN runs a 50 node K2 GPU cluster to complete jobs in a reasonable timespan. Doing this same work on a desktop

  14. Looking Forward • Revise processing framework for greater parallelism • Extended the engine to use on GPU encoders/decoders • Switch to CUDA for easier development for compute • Test on p2 instances for faster processing + I/O

  15. Preview on Steam - HTC Vive and Oculus Rift http://store.steampowered.com/app/623970/Pixvana_SPIN_Play/

  16. Questions @pixvana, @seanpixvana

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend