FPS is not accurate #3138
Comments
|
Transcoding only or some heavy editing? |
|
No editing whatsoever. Here are the steps I took: 1.Download a sample video: I used this: https://www.sample-videos.com/video123/mp4/720/big_buck_bunny_720p_1mb.mp4 |
|
Cannot reproduce this with the one of the latest Daily Builds ( https://www.openshot.org/download/ ). Can you give it a try? |
|
I checked it again using the latest build and I still see the problem. See
video: https://a.cl.ly/mXum1Rdr Notice that the final file is 30.19 f/s
when it should be 30. Audio is 48K as it should be.
…On Mon, Dec 23, 2019 at 9:06 AM SuslikV ***@***.***> wrote:
Cannot reproduce this with the one of the latest *Daily Builds* (
https://www.openshot.org/download/ ). Can you give it a try?
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#3138?email_source=notifications&email_token=AC4NL5H4WYAQLGEZ37E4IX3Q2DHWVA5CNFSM4J6UHKHKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHRJV2A#issuecomment-568498920>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC4NL5AWT23OZT6RKJKFPS3Q2DHWVANCNFSM4J6UHKHA>
.
|
|
Can you share the exported file as well? |
|
You can find both files here: https://a.cl.ly/jkune79q |
|
The file has constant FPS. Deltas is 159 samples of 512 deltas + last one at 0 delta for video, and 249 samples of 1024 deltas for audio - thus only two constant values but the last sample. No other variations is present. I not quite sure how Windows properties calculates the fps field for now. But the samples are OK in terms of constant fps and the last ISO I have seen for MP4 format. Am I wrong? Some info. audio It looks like current version of the Windows simply takes the length of the video track (5.3 sec) and total number of the video samples (160) to calculate the FPS.
|
|
I am not an expert about this but I do know that when I submitted to one of
amazon services, it was rejected because the fps was not exactly 30. After
I converted it using VLC it was accepted. Here is the converted file after
VLC conversion and this is compare of both files FPS.
https://a.cl.ly/qGuDXz9e converted file: https://a.cl.ly/o0uQEQbq The rate
is different but you can clearly see that the frame rate is exactly 30 in
the converted file vs the file produced by OpenShot is 30.19 and not 30
…On Mon, Dec 23, 2019 at 4:30 PM SuslikV ***@***.***> wrote:
The file has constant FPS. Deltas is 159 samples of 512 deltas + last one
at 0 delta for video, and 249 samples of 1024 deltas for audio - thus only
two constant values but the last sample. No other variations is present. I
not quite sure how Windows properties calculates the fps field for now. But
the samples are OK in terms of constant fps and the last ISO I have seen
for MP4 format. Am I wrong?
*Some info.*
output:
video
track timescale = 15360
track duration = 81408
thus track's duration is 5.3 sec
audio
track timescale = 48000
track duration = 254976
thus track's duration is 5.312 sec (the longest track = the length of the
movie)
It looks like current version of the Windows simply takes the length of
the video track (5.3 sec) and total number of the video samples (160) to
calculate the FPS.
@ashmetry <https://github.com/ashmetry>
1. Is this causes any real difficulties when you importing the file to
some service?
2. Just for the experiment, can you say what corresponding Windows
field says about the next file sample? - it was created using other
software (OBS Studio),
1280x720_444_709_tv_10sec_10sec.zip
<https://github.com/OpenShot/openshot-qt/files/3996410/1280x720_444_709_tv_10sec_10sec.zip>
3. What build of the OpenShot you actually used for the posted
exported file (in the .zip)? It has no *faststart* swap of the indexes
tables for mp4. Did you made some changes to the source of the OpenShot or
just posted the output file from the old setup? (it has nothing to do with
the issue itself but I'm just wondering why it is not working for you)
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3138?email_source=notifications&email_token=AC4NL5GFURHBD2AXGCHYHHTQ2E32DA5CNFSM4J6UHKHKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHSBPNA#issuecomment-568596404>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC4NL5GZFYQAC66BFMDMFXLQ2E32DANCNFSM4J6UHKHA>
.
|
|
@ashmetry the technical info is not for you personally. I just using the site to inform other developers who may find something new. @ashmetry Anyway, can you give the answer to other two questions as well? From my point of view if there is problem then it is only in the last encoded video sample created by libopenshot lib. The posted video audio Thus the 5.13332(2) sec / 154 (total video samples) possibly truncated to 0.03333 ms of about 30 fps. The deviations in video fps is small in both cases. The standard I mentioned earlier is ISO/IEC 14496-12 edition 2015. There was info:
That means that the Is the mentioned Amazon service is free? (to test with) |
|
Thanks for your input. Unfortunately it’s part of a bigger service on
amazon that is not free. (Amazon video advertising network). I agree that
the deviation is minimal but the final result according to Windows is still
not exactly 30fps. I appreciate your input and hard work on this.
…On Tue, Dec 24, 2019 at 3:16 AM SuslikV ***@***.***> wrote:
@ashmetry <https://github.com/ashmetry> the technical info is not for you
personally. I just using the site to inform other developers who may find
something new.
@ashmetry <https://github.com/ashmetry> Anyway, can you give the answer
to other two questions as well?
From my point of view if there is problem then it is only in the last
encoded video sample created by *libopenshot* lib. The posted Untitled
Project-converted.mp4 has 1 sample of 2999 delta + 153 samples of 3000
delta for video, and 236 samples of 1023, 1024, 1025 deltas for audio (all
values cycled in row and each last for 1 sample):
video
track timescale = 90000
track duration = 461999
thus track's duration is 5.13332(2) sec (the longest track = the length of
the movie)
audio
track timescale = 48000
track duration = 241663
thus track's duration is 5.03464583(3) sec
Thus the 5.13332(2) sec / 154 (total video samples) possibly truncated to
0.03333 ms of about 30 fps.
The deviations in video fps is small in both cases. The standard I
mentioned earlier is ISO/IEC 14496-12 edition 2015. There was info:
The time to sample boxes must give non‐zero durations for all samples with
the possible exception of the last one. Durations in the stts box are
strictly positive (non‐zero), except for the very last entry, which may be
zero. This rule derives from the rule that no two time‐stamps in a stream
may be the same. Great care must be taken when adding samples to a stream,
that the sample that was previously last may need to have a non‐zero
duration established, in order to observe this rule. If the duration of the
last sample is indeterminate, use an arbitrary small value and a dwell
edit.
That means that the Untitled Project-converted.mp4 (that has the *first*
sample aside of the others by the 1/90000 sec) is actually not constant fps
and the file mentioned in the issue is actually comply to the mp4 rules. Is
it has variable frame rate? In my point of view - No.
Is the mentioned Amazon service is free? (to test with)
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3138?email_source=notifications&email_token=AC4NL5H5ONSNCFCXV76GXXLQ2HHNBA5CNFSM4J6UHKHKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHS3OJQ#issuecomment-568702758>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC4NL5HY7F36DQUBKRHB2U3Q2HHNBANCNFSM4J6UHKHA>
.
|
|
Hope, Windows will change it back to the Win7 result - where it shows no variable fps for this file. Still waiting for the answer on the two other questions. |
The OP says OpenShot 2.4.4, I assume that means 2.4.4 release. Didn't faststart go in after the release? |
|
@ferdnyc Release version? It always in few steps off of the daily build. Maybe you are right. I totally forgot about the release version. |
|
This is fixed in the daily build I believe. Closing issue for now. |
|
@jonoomph I still experience the same issue on version 2.5.1. As a result I am unable to upload the video to Amazon. |
|
@cohlar were you using a daily build? |
FPS is not accurate. Selecting for example 30 fps will generate 30.xx in the final video. This causes problems with services only accepting videos with specific FPS quality.
Steps to reproduce the behavior:
Steps I took:
Expected behavior:
A clear and concise description of what you expected to happen.
System Details:
The text was updated successfully, but these errors were encountered: