By Dax Grove -
Hello all,
I am currently working on a large IP camera project and would like some advice. I have decided on the Axis P1357, an IP camera with a native resolution of 2592x1944 (5MP) because it supports both audio recording and the ability to change the lens.
The project I am working on requires me to record 30 speakers in our conference rooms simultaneously (cameras would be placed about 15-20 feet from the speaker on the ceiling). The speakers have white boards to use at their disposal, most of which will write information down on them, or use them for power point presentations. I need to capture their presentations simultaneously, compress the video, and upload it to our CDN within a few hours so our subscribers can watch the data.
If you were do to an hour of continuous record at native resolution (5MP) using H.264 @30fps it would be 8.28MB/s or 29,808MB/hr also known as 29.11GBhr. Luckily, these IP cameras have a wide array of on-board compression. But even dropping the resolution in the camera to 1080P it's still 12.34GB/hr and 720P is still 5.55GB/hr
Now here's where I need help. I need to get the video to sub 1GB/hr before I upload it to the CDN. Even with changing the on-board compression to 90% at 720p @30fps it's still about 2.5GBhr and looks terrible. So now I know I need to do compression using at least a small render farm, but here are a few questions.
1) What resolution should I send the video to the render farm that would allow me compress the video to the size I want, without losing too much quality. Do I do the full H.264 5MP??H.264 1080P? H.264 720P??
2) What programs would you utilize to do the data compression? I was thinking a mixture of the render management software Deadline (which I already have licenses and machines for) in combination with Thinkbox's Draft. Or do I go with an open source piece of software?
The project has a large budget, and the infrastructure in place to at-least store the data and compress it (10 node render farm in place).
Here is some source material below.
Thank you!