Handbrake is "adding" a few seconds to the compressed video

[hshadow]

Hi, everyone!

I've been using Handbreak to shrink the size of my videos for a few years now. But from a couple of years till now something's happening and when I shrink a video, the exported one is a bit "longer" than the original (by some seconds up to a couple of minutes depending on the duration of the original). Everything's OK with the size (30 to 60% the size of the original, awesome compression rate without any loss as always).

I'd like to know why this happens and how to fix it. I make longplays and compilations for my YouTube channel so these videos are expected to be extremely long (by hours I mean). The thing is that when I'm editing the videos in Premiere I take note of the timestamps when different things happen (common text file, nothing odd about it) so when I upload the video I can simply paste them into the description so they become clickable.

But, once the video is online I have to check all the timestamps one by one (they can be a lot sometimes) to make sure they are pointing to the correct time in the video's timeline. It's a bit tedious to have to do it every single time. The more advanced I am in the video, the greater the difference in seconds that I have to correct.

For instance, a video of 1:41:53 exported by Premiere, turns out to be of 1:42:00 after passing it through Handbrake. (All-Bosses Compilation, +7 seconds)
And a video of 23:35:20 exported by Premiere, turns out to be of 23:36:45 after passing it through Handbrake. (100% Longplay, +1min 25 seconds)

I use a preset I created based on the "Vimeo YouTube 720p60" one. The only modification I made was to set the cropping to none (0px all sides) because I like the black bars when recording classic games.

The other modification I made was to set the "Peak Framerate" to "Constant Framerate" because I thought some frames where "missing" or "added" and that made the compressed version longer. But it wasn't the case, same results.

Thanks!

[davexnet]

Check the mediainfo, text report for full details, about the frame rate and related details of the source.
Try setting framerate (same as source)/variable frame rate

[poisondeathray]

Why don't you export it from Premiere directly , just once ? You can use voukoder (using the same encoding libraries as handbrake - such as libx264 or libx264 or nvenc) .

Or are you on a Mac ?

[hello_hello]

What's the video frame rate?

I never deal with SMPTE timecode, and I don't use Premier, but a little tingle in my brain made me check, and it seems mighty co-incidental that SMPTE Non Drop Frame Timecode, applied to what'd normally be a Drop Frame frame rate, is out compared to people time by the same amount as the examples you gave.

The way I remember it, SMPTE Non Drop Frame Timecode would assume 23.976fps video is 24fps, or 29.97fps is 30fps, even though it isn't, because it can only count a whole number of frames per second. Hence the time difference for those frame rates.

People time is hours, minutes, seconds, milliseconds.
01:01:01.083
SMPTE time is hours, minutes, seconds, frame number.
01:01:01:02

https://en.wikipedia.org/wiki/SMPTE_timecode#Drop-frame_timecode
http://avisynth.nl/index.php/ShowFrameNumber#Drop-Frame_versus_Non-Drop-Frame_Time_Code


[Attachment 62786 - Click to enlarge]


[Attachment 62787 - Click to enlarge]

PS If you happen to be an Avisynth user, the screenshots above came from this little script. A link to the Position function is in my signature.

BlankClip(length=2400000).KillAudio()
AssumeFPS(24000, 1001)
Position(All=true, DropF=false)

[hshadow]

Check the mediainfo, text report for full details, about the frame rate and related details of the source.
Try setting framerate (same as source)/variable frame rate

I'll try this on my next longplay next week. Thanks!

[hshadow]

Why don't you export it from Premiere directly , just once ? You can use voukoder (using the same encoding libraries as handbrake - such as libx264 or libx264 or nvenc) .

Or are you on a Mac ?

Nope, I'm on Windows. The thing is that I don't know how to use those things you mention with Premiere. I use defaults mostly. But I'm willing to learn! Where can I read or watch how to use them? If there's a way to DIRECTLY export from premiere a "final" already compressed with no loss version I want to know it of course!

Thanks!

[hshadow]

By combining what @davexnet and @hello_hello said, I realized that using "same as source" maybe I can fix it. I forced it to 60fps because I thought that being videogames it would've had more quality (rounding up any posible ".97" or that sort of "odd" fps count, generating that difference in length). I will try it in a couple of weeks with the next longplay. Thanks!

[hello_hello]

By combining what @davexnet and @hello_hello said, I realized that using "same as source" maybe I can fix it. I forced it to 60fps because I thought that being videogames it would've had more quality (rounding up any posible ".97" or that sort of "odd" fps count, generating that difference in length). I will try it in a couple of weeks with the next longplay. Thanks!

I'd be astounded if it's anything to do with Handbrake.
I strongly suspect that Premier is displaying SMPTE Non Drop Frame timecodes for fractional frame rates and you're making note of them, but they don't match people timecodes. If I'm correct and the frame rate is fractional (29.97fps etc) and you tell Premier to use Drop Frame Timecode instead, it should match up to people timecode later on. As it does in this screenshot. I don't use Premier but I can't imagine what else it could be.


[Attachment 62790 - Click to enlarge]

Forcing a frame rate for Handbrake shouldn't change anything. If the video happened to be 30fps and you told it to output 60fps, it'd simply duplicate every frame to double the frame rate, but if something happens at a particular time in the video that won't change. Changing the frame rate in Handbrake will cause it to add and/or drop frames as required to output the specified frame rate but the video duration won't be any different so your times still won't match and you'll just have a video with a different number of frames than before.

[poisondeathray]

Nope, I'm on Windows. The thing is that I don't know how to use those things you mention with Premiere. I use defaults mostly. But I'm willing to learn! Where can I read or watch how to use them? If there's a way to DIRECTLY export from premiere a "final" already compressed with no loss version I want to know it of course!

Thanks!

https://www.voukoder.org/
https://www.voukoder.org/forum/thread/783-downloads-instructions/

There are video tutorials on sites such as youtube

[hshadow]

By combining what @davexnet and @hello_hello said, I realized that using "same as source" maybe I can fix it. I forced it to 60fps because I thought that being videogames it would've had more quality (rounding up any posible ".97" or that sort of "odd" fps count, generating that difference in length). I will try it in a couple of weeks with the next longplay. Thanks!

I'd be astounded if it's anything to do with Handbrake.
I strongly suspect that Premier is displaying SMPTE Non Drop Frame timecodes and you're making note of them, but they don't match people timecodes. If I'm correct and the frame rate is fractional (29.97fps etc) and you tell Premier to use Drop Frame Timecode instead, it will match up to people timecode later on. As it does in this screenshot. I don't use Premier but I can't imagine what else it could be.


[Attachment 62790 - Click to enlarge]

Forcing a frame rate for Handbrake shouldn't change anything. If the video happened to be 30fps and you told it to output 60fps, it'd simply duplicate every frame to double the frame rate, but if something happens at a particular time in the video that won't change. Changing the frame rate in Handbrake will cause it to add and/or drop frames as required to output the specified frame rate but the video duration won't be any different so your times still won't match and you'll just have a video with a different number of frames than before.

Thanks, man! It was a Premiere thing. I went to "Sequence settings" and changed the "Display Format" from "Non-Drop-Frame" to "Drop-Frame" and now the timer is showing the "real" length. Thanks again!

[hshadow]

Nope, I'm on Windows. The thing is that I don't know how to use those things you mention with Premiere. I use defaults mostly. But I'm willing to learn! Where can I read or watch how to use them? If there's a way to DIRECTLY export from premiere a "final" already compressed with no loss version I want to know it of course!

Thanks!

https://www.voukoder.org/
https://www.voukoder.org/forum/thread/783-downloads-instructions/

There are video tutorials on sites such as youtube

I haven't forgot about you, I'm going to check this stuff right away! If I can use this thing to export my videos directly compressed I'll kill two birds with one THREAD XD!

Thank you too!

[hshadow]

Nope, I'm on Windows. The thing is that I don't know how to use those things you mention with Premiere. I use defaults mostly. But I'm willing to learn! Where can I read or watch how to use them? If there's a way to DIRECTLY export from premiere a "final" already compressed with no loss version I want to know it of course!

Thanks!

https://www.voukoder.org/
https://www.voukoder.org/forum/thread/783-downloads-instructions/

There are video tutorials on sites such as youtube

Well, I tried using it but something's not right. The 1:40:00 video takes from 4 to 6 hours to render using Voukoder (I canceled it after some minutes). With the "normal" method it takes around the same amount of time the length of the video plus 20% more or less (for instance this 1:40:00 video takes 2 hours to render approx.). And less than that for the Handbrake process. (Around an hour and a half for this same example).

Thanks, anyway!

[poisondeathray]

You're probably not using comparable settings

The voukoder defaults are higher quality, slower than handbrake's defaults

(Or if you use NVEnc GPU, it will be much faster, but lower compression)

If you use comparable everything, it will be faster because you omit the 1st premiere export step. It should be about 30-60% faster overall