Open tzwenn opened 9 years ago
Profiling on rPi B shows most delay comes from ffmpeg h264 decoding, but also SDL rendering is at sub-interactive rates.
ncalls tottime percall cumtime percall filename:lineno(function)
56 8.515 0.152 8.515 0.152 {method 'decodeFrame' of 'h264decode.Decoder' objects}
43 1.335 0.031 1.335 0.031 {method 'display' of 'pygame.overlay' objects}
56 0.003 0.000 0.096 0.002 blockalgo.py:246(decrypt)
Comparison to ffmpeg on i7 (hardware accelerated AES+H.264):
ncalls tottime percall cumtime percall filename:lineno(function)
267 2.874 0.011 2.874 0.011 {method 'display' of 'pygame.overlay' objects}
267 0.824 0.003 0.824 0.003 {method 'decodeFrame' of 'h264decode.Decoder' objects}
267 0.001 0.000 0.033 0.000 blockalgo.py:246(decrypt)
How to do it ? can u tell me the step ?
Haven't done it yet, since for a weekend project the first glimpse of the OpenMAX API was a bit discouraging.
Yes,I know how to decode h264 packet, but i don't know how to run PyOpenAirMirror on Raspberry, I never success run it on ubuntu ,so please tell me how to run the PyOpenAirMirror on rpi,if i can,i will decode using OPENMAX API to decode it and re-coded it using c++,please help me
Well, I haven't had any issues after installing avahi. And I used a macOS and fplyServer for an external fp-setup handshake. Alternatively you can use https://github.com/espes/Slave-in-the-Magic-Mirror, if you don't have access to macOS.
Thank,I will try it right now
Pretty slow with current setup. First investigate if delay is in decryption, decoding or rendering.