[0.9.25] Known Bugs

[anrayliu]

Here are a list of known bugs as of version 0.9.25.

• yt-dlp release v2024.12.23 is bugged and sometimes causes youtube videos to fail to open. Upgrade to the latest version to fix.
• Youtube videos can ocassionaly freeze or stutter due to slow frame grabbing over the network.
• Seeking in some Youtube videos seem to fail at 1080p.
• Some videos, due to how they were encoded, cannot their frame count extracted with 100% accuracy. The extracted number usually differs from the real number by a few frames. If a completely accurate count is required, use video._get_real_frame_count() or preload the frames.
• Some videos when played as bytes cannot have their frame size read properly. This will cause an exception.
• Some videos when played as bytes cannot have their frame time stamps extracted, which means VFR videos cannot be read as bytes.
• PyAV version 14.0.1 seems to cause bugs in imageIO for now. Use decord to read from bytes for the time being or downgrade to version 13.
• When using the decord reader to play a video with multiple video tracks, the first track will not be selected, causing side effects and potential exceptions. Use any of the other three readers instead.
• FFmpegReader sometimes outputs a strange resource warning.

[Sora1123]

I am currently using pyvidplayer2 with pygame similar to the 'pygame integration' shown in the pypi website (https://pypi.org/project/pyvidplayer2/). When I tried to do 'vid.seek_frame(100, relative = True)', the new frame is not drawn, even when I put 'force_draw' True. When I tried on 'vid.seek(1)', it is working. Could you possibly modify this?

[anrayliu]

This is definitely a really interesting bug. From a quick glance, I see no reason why seek_frame would be any different from seek. Thanks for finding this, I'll get back to you soon regarding either a workaround or a patch.

[anrayliu]

I apologize for the long delay, but I finally made a patch. This behavior is actually intended, neither seek was ever supposed to generate frames. I've figured out why seek renders the new frame and can confirm that the logic checks out. That said, I did add a new method called buffer_current() that populates frame_data and frame_surf. You can call it after seeking to generate the current frame. Here's the link to the docs: https://github.com/anrayliu/pyvidplayer2/blob/main/documentation.md#:~:text=buffer_current()%20%2D%3E%20bool

[Sora1123]

Thank you for making a patch for this. I really appreciate this as my current project requires this. I tried buffer_current() after seek_frame(10), however I still could not generate/update the current frame on my pygame surface, and keep getting False for buffer_current(). Is it possible to manually change 'seek_frame()' into generating/updating the frames? I am sorry if I am missing something about this new patch.

[anrayliu]

Do you mind sharing your repository and video? Sharing a few lines of code should be ok too. I’m sorry it’s not working, but I have a bit of time this week so I’ll try and fix this problem asap.

…

On Tue, Jun 24, 2025 at 11:05 AM Sora1123 ***@***.***> wrote: *Sora1123* left a comment (anrayliu/pyvidplayer2#53) <#53 (comment)> Thank you for making a patch for this. I really appreciate this as my current project requires this. I tried buffer_current() after seek_frame(10), however I still could not generate/update the current frame on my pygame surface, and keep getting False for buffer_current(). Is it possible to manually change 'seek_frame()' into generating/updating the frames? I am sorry if I am missing something about this new patch. — Reply to this email directly, view it on GitHub <#53 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ARVZARDQBDDNRVGPAI3YT333FFSMZAVCNFSM6AAAAABU6YJZGGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTAMBQHA3TIMJRHA> . You are receiving this because you were assigned.Message ID: ***@***.***>

[Sora1123]

Of course! Thank you for spending your precious time on this. I started coding around a year ago, so my code would be really hard to read. vid.seek_frame(10) lies at line 133. 'Angle_check.py' is the file that has the code, and 'video' file stores a video https://drive.google.com/drive/folders/1yBOPJtiNnTWAfV1u8Ie4XtZOGvGItYbZ

…

On Tue, Jun 24, 2025 at 11:09 PM Anray Liu ***@***.***> wrote: *anrayliu* left a comment (anrayliu/pyvidplayer2#53) <#53 (comment)> Do you mind sharing your repository and video? Sharing a few lines of code should be ok too. I’m sorry it’s not working, but I have a bit of time this week so I’ll try and fix this problem asap. On Tue, Jun 24, 2025 at 11:05 AM Sora1123 ***@***.***> wrote: > *Sora1123* left a comment (anrayliu/pyvidplayer2#53) > < #53 (comment)> > > Thank you for making a patch for this. I really appreciate this as my > current project requires this. I tried buffer_current() after > seek_frame(10), however I still could not generate/update the current frame > on my pygame surface, and keep getting False for buffer_current(). Is it > possible to manually change 'seek_frame()' into generating/updating the > frames? I am sorry if I am missing something about this new patch. > > — > Reply to this email directly, view it on GitHub > < #53 (comment)>, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/ARVZARDQBDDNRVGPAI3YT333FFSMZAVCNFSM6AAAAABU6YJZGGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTAMBQHA3TIMJRHA> > . > You are receiving this because you were assigned.Message ID: > ***@***.***> > — Reply to this email directly, view it on GitHub <#53 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BIMEMSJ3YWJEXCPH3RQCYBD3FFSZFAVCNFSM6AAAAABU6YJZGGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTAMBQHA4DKNJZGY> . You are receiving this because you commented.Message ID: ***@***.***>

[anrayliu]

Of course! Thank you for spending your precious time on this.

No problem at all, in fact you're helping me by finding bugs.

I've figured out the issue here, and it's a matter of semantics. By definition, buffer_current will only populate frame_data and frame_surf if any of them are None. When you are seeking frames, you are doing so in the middle of playback, when frame_data and frame_surf are both populated (albeit with the wrong frame). Therefore, buffer_current will fail and return False.

The obvious workaround is to manually set either frame_data or frame_surf to None before calling buffer_current. I tried this in your code and was able to successfully step through the frames. However, I'm not sure if this is intuitive, and I'm thinking about changing the behaviour of buffer_current entirely to accommodate for this. What do you think?

Also, just a side note, seeking frames is not very efficient. If performance ever becomes an issue, feel free to let me know again so we can optimize it. This seems like a very cool project :]

Let me know if this fixes your problem. Cheers!

[Sora1123]

It worked as expected! Thank you so much. I agree with changing the behaviour of buffer_current to include setting frame_data or frame_surf to None. For now, I am satisfied with the 'seek_frame' performance, but I will let you know if I need a better performance. Best wishes!

…

On Wed, Jun 25, 2025 at 12:23 PM Anray Liu ***@***.***> wrote: *anrayliu* left a comment (anrayliu/pyvidplayer2#53) <#53 (comment)> Of course! Thank you for spending your precious time on this. No problem at all, in fact you're helping me by finding bugs. I've figured out the issue here, and it's a matter of semantics. By definition, buffer_current will only populate frame_data and frame_surf if any of them are None. When you are seeking frames, you are doing so in the middle of playback, when frame_data and frame_surf are both populated (albeit with the wrong frame). Therefore, buffer_current will fail and return False. The obvious workaround is to manually set either frame_data or frame_surf to None before calling buffer_current. I tried this in your code and was able to successfully step through the frames. However, I'm not sure if this is intuitive, and I'm thinking about changing the behaviour of buffer_current entirely to accommodate for this. What do you think? Also, just a side note, seeking frames is not very efficient. If performance ever becomes an issue, feel free to let me know again so we can optimize it. This seems like a very cool project :] Let me know if this fixes your problem. Cheers! — Reply to this email directly, view it on GitHub <#53 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BIMEMSPCO4UUPDYHVVDZQPT3FIP3NAVCNFSM6AAAAABU6YJZGGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTAMBTGA4DOMJZGI> . You are receiving this because you commented.Message ID: ***@***.***>