TDA4VE-Q1: H264 Encode Spec clarification & performance discussion

Hi Gibbs, 

Gibbs Shih said:

Q1:Ref TRM p12 (spruj28.pdf). Does it mean TDA4VE "only" support max "8" video H.264 Encoding "simultaneous" from camera input? (If we assume camera input bandwidth is OK)

Yes, that means you can have up to 8 channels of encoding as long as you stay w/in the pixel throughput constraints for your specific device.

Gibbs Shih said:

Q2: Base on Q1, base on software point of view, Could we use "one H.264 Encoding unit" for video file recording and network video streaming? Do we have example for gstreamer  or vision app?

Yes, the use-case you're specifying here is supported through the GStreamer framework in our Linux SDK. Could you clarify if you are using Linux or QNX? Also, what SDK version?

Gibbs Shih said:

Q3: As far as I know, TDA4VE only has "one VPAC instance" and "one H264 encode instance", Base on software point of view, could we implement 2M@30FPS * 5 camera input for video date file recording and network video streaming?

Could you specify what the exact resolution is for 2MP? Also, how many streams are being written to file sink vs. network sink?

Thank you,
Sarabesh S.

Hi, Sarabesh

Thanks you replies.

Few replies.

Q2/A2: We should use linux or FreeRTOS w/ vision APP.

Q3/A3: 

2M@30FPS * 5  =  (1920x1080 @30FPS) * 5 cameras.

For this example, there are 5 video stream are being written to file sink, and 5 network sink simultaneous.

Because We only support 8 channels encoding, Does it have any limitation?

Thank You.

Gibbs