Den 19.12.2021 16:40, skrev Andrew Randrianasulu:

On Sunday, December 19, 2021, Terje J. Hanssen <terjejhanssen@gmail.com> wrote:

Thanks, script testrun executed with attached output here.
    Test-run_Intra-frame_key_word

As seen below from the summary table, all files except the MPEG-2 generated "SD-MP2-50.mpg" without the ffmpeg -intra flag set, contains all Intra-frames only:

-------------------------------------------------------
du -sh SD*.*            # key_frame=1    # key_frame=0
                (# I-frames)    (# non-I-frames)
284M    SD-MOV.mov            746        0
205M    SD-DV50.dv            745        0
186M    SD-MP2I50.mpg            745        0
153M    SD-MP2I-.mpg            745        0
186M    SD-MP2-50.mpg            63        682
146M    SD-MP2I50-an.m2v        745        0

good!

but a bit concerned 746 vs 745 frame count. Are we losing one frame somewhere?

To repeat:
1. SD-MOV.mov (ProRes) was the initial record source

2. SD-DV50.dv was encoded from ProRes using a simple ffmpeg target (preset)
ffmpeg -i SD-MOV.mov -target pal-dv50 SD-DV50.dv

3. The SD-MP2* were encoded from SD-DV50.dv using ffmpeg

All ffmpeg generated files has the same frame counts

My take:
The single frame difference seems to occure from step 1 - 2, or the script doesn't count 1 indentical as the ffmpeg generated 2 and 3.

Counts matched # key_frame=1 (I-frames)
./kfrm.sh SD-MOV.mov | grep -c key_frame=1

Counts other (unmatched) than key_frame=1
./kfrm.sh SD-MOV.mov | grep -v -c key_frame=1

--------------------------------------------------------