VTDecompressionSessionDecodeFrame ERROR -12916 (kVTFormatDescriptionChangeNotSupportedErr) - ios

Question

I got a problem I can't get my head around.
First I create a VTCompressionSessionCreate (h264) then in my compression callback when I start feeding images I get a CMSampleBufferRef sampleBuffer as expected.
Just for debugging the code stream I then create a VTDecompressionSessionCreate and feed the 'sampleBuffer' containing the H264 stream to a VTDecompressionSessionDecodeFrame and I would expect a CVImageBufferRef imageBuffer in my decompression callback.
Now to the problem:
If I create VTDecompressionSessionCreate using the 'sampleBuffer' from the compression callback like this:
CMFormatDescriptionRef format = CMSampleBufferGetFormatDescription(sampleBuffer);
Everything works as expected and I get CVImageBufferRef's in my decompression callback.
However my intention is to send the data over a network so I need to get my format discription from the in stream SPS and PPS information.
So then I must 'fake' getting the SPS and PPS by first extracting them and then using them like this:
CMFormatDescriptionRef format = CMSampleBufferGetFormatDescription(sampleBuffer);
size_t spsSize, ppsSize;
size_t parmCount;
const uint8_t* sps, *pps;
CMVideoFormatDescriptionGetH264ParameterSetAtIndex(format, 0, &sps, &spsSize, &parmCount, NULL );
CMVideoFormatDescriptionGetH264ParameterSetAtIndex(format, 1, &pps, &ppsSize, &parmCount, NULL );
const uint8_t* const parameterSetPointers[2] = {sps, pps};
const size_t parameterSetSizes[2] = {spsSize, ppsSize};
CMFormatDescriptionRef format2;
status = CMVideoFormatDescriptionCreateFromH264ParameterSets(kCFAllocatorDefault, 2, parameterSetPointers, parameterSetSizes, 4, &format2);
I would expect format and format2 to contain the same information but:
format = <CMVideoFormatDescription 0x17004fd50 [0x19483ac80]> {
mediaType:'vide'
mediaSubType:'avc1'
mediaSpecific: {
codecType: 'avc1' dimensions: 1280 x 720
}
extensions: {<CFBasicHash 0x170270cc0 [0x19483ac80]>{type = immutable dict, count = 2,
entries =>
0 : <CFString 0x194935fa0 [0x19483ac80]>{contents = "SampleDescriptionExtensionAtoms"} = <CFBasicHash 0x170270c40 [0x19483ac80]>{type = immutable dict, count = 1,
entries =>
2 : <CFString 0x194939fa0 [0x19483ac80]>{contents = "avcC"} = <CFData 0x1700c9920 [0x19483ac80]>{length = 35, capacity = 35, bytes = 0x0164001fffe100106764001fac56c050 ... 28ee3cb0fdf8f800}
}
2 : <CFString 0x194936000 [0x19483ac80]>{contents = "FormatName"} = <CFString 0x17003a160 [0x19483ac80]>{contents = "H.264"}
}
}
}
format2:
format2 = <CMVideoFormatDescription 0x174051c70 [0x19483ac80]> {
mediaType:'vide'
mediaSubType:'avc1'
mediaSpecific: {
codecType: 'avc1' dimensions: 1280 x 720
}
extensions: {<CFBasicHash 0x17426f9c0 [0x19483ac80]>{type = immutable dict, count = 5,
entries =>
0 : <CFString 0x19499a608 [0x19483ac80]>{contents = "CVImageBufferChromaLocationBottomField"} = <CFString 0x19499a648 [0x19483ac80]>{contents = "Center"}
1 : <CFString 0x19499a328 [0x19483ac80]>{contents = "CVFieldCount"} = <CFNumber 0xb000000000000012 [0x19483ac80]>{value = +1, type = kCFNumberSInt32Type}
3 : <CFString 0x194935fa0 [0x19483ac80]>{contents = "SampleDescriptionExtensionAtoms"} = <CFBasicHash 0x17426b100 [0x19483ac80]>{type = immutable dict, count = 1,
entries =>
2 : <CFString 0x174031560 [0x19483ac80]>{contents = "avcC"} = <CFData 0x1740c4910 [0x19483ac80]>{length = 35, capacity = 35, bytes = 0x0164001fffe100106764001fac56c050 ... 28ee3cb0fdf8f800}
}
5 : <CFString 0x19499a5e8 [0x19483ac80]>{contents = "CVImageBufferChromaLocationTopField"} = <CFString 0x19499a648 [0x19483ac80]>{contents = "Center"}
6 : <CFString 0x1949360e0 [0x19483ac80]>{contents = "FullRangeVideo"} = <CFBoolean 0x19483b030 [0x19483ac80]>{value = false}
}
}
}
format works
forma2 don't and VTDecompressionSessionDecodeFrame throws error -12916.
Thank you for helping.
.

Answer 1

Solved it. It was the way I created CMFormatDescriptionRef containing the code stream that was causing the error.
The SPS and PPS was taken from a CFSampleBuffer. Then I create a CMVideoFormatDescriptionCreateFromH264ParameterSets so far so good. But in the same application I turned around the stream and decoded the picture using the same CFSampleBuffer. That's not working and was causing the error. I had to convert the payload to NSData first then create a new CFSampleBuffer from the NSData. Then it works