3 ms·

All very good considerations. The "live" VR 360 capture market is really just getting started. I like the idea of "post-processing" as a service, especially.

All very good considerations. The "live" VR 360 capture market is really just getting started. I like the idea of "post-processing" as a service, especially. It might be down the road a few months though as the market for it develops.

i336_ 10y ago
:) To me, it sounds like you have the perfect head start to this whole scene. I can see why you want people to know this exists... Your quoting "live" there made me hesitate on the word for just long enough to start wondering about realtime processing, which I get the impression your system is capable of (if you were able to do SD card tests and instantly preview the results). Now I'm wondering what would happen if you ran your software on a box in the cloud with a 1Gbit link, where people feed in their dual- or tri-camera raw footage, and perfect 3D (and maybe with a hefty premium, calculated depth info) comes out the other end - all in near-realtime. Can you manage less than 1s of latency? 500ms even? (Okay maybe I'm being too optimistic here) Hmm, and maybe you could provide direct ingress support for broadcast packs.... And thinking about it, depending on how CPU-intensive your system is, you might be able to run this on relatively inexpensive hardware per 1G/2G link. (I know EPB offers commercial 10G installation in Chattanooga for $300/mo, but I'm not sure what their datacenter pricing is.) I really like the idea of post-processing as a service - considering what options are on the table, that direction seems the least complex and the most straightforward. (PS. Please excuse the rather non-provisional and overly concrete/direct/instructional wording style of my previous post - my brain's a little under the weather at the moment, and grammaring how I'd prefer to is proving occasionally challenging.)
- opticalflow 10y ago
  Well, I think you're on the right track. Right behind me are 30 supermicro pizzboxes (some have GPUs) that are racked but powered down and looking very lonely (but with 6 dark fibers also feeling lonely at the moment). Perhaps I'd consider an "encoding.com for VR". Honestly, the main part I'm missing is real-time stitching, but maybe Facebook's open source release for their Surround 360 software later this summer will solve that!

i336_ 10y ago

:) To me, it sounds like you have the perfect head start to this whole scene. I can see why you want people to know this exists... Your quoting "live" there made me hesitate on the word for just long enough to start wondering about realtime processing, which I get the impression your system is capable of (if you were able to do SD card tests and instantly preview the results). Now I'm wondering what would happen if you ran your software on a box in the cloud with a 1Gbit link, where people feed in their dual- or tri-camera raw footage, and perfect 3D (and maybe with a hefty premium, calculated depth info) comes out the other end - all in near-realtime. Can you manage less than 1s of latency? 500ms even? (Okay maybe I'm being too optimistic here) Hmm, and maybe you could provide direct ingress support for broadcast packs.... And thinking about it, depending on how CPU-intensive your system is, you might be able to run this on relatively inexpensive hardware per 1G/2G link. (I know EPB offers commercial 10G installation in Chattanooga for $300/mo, but I'm not sure what their datacenter pricing is.) I really like the idea of post-processing as a service - considering what options are on the table, that direction seems the least complex and the most straightforward. (PS. Please excuse the rather non-provisional and overly concrete/direct/instructional wording style of my previous post - my brain's a little under the weather at the moment, and grammaring how I'd prefer to is proving occasionally challenging.)

opticalflow 10y ago
Well, I think you're on the right track. Right behind me are 30 supermicro pizzboxes (some have GPUs) that are racked but powered down and looking very lonely (but with 6 dark fibers also feeling lonely at the moment). Perhaps I'd consider an "encoding.com for VR". Honestly, the main part I'm missing is real-time stitching, but maybe Facebook's open source release for their Surround 360 software later this summer will solve that!