VideoHelp Forum

Try DVDFab and download streaming video, copy, convert or make Blu-rays,DVDs! Download free trial !
+ Reply to Thread
Page 62 of 62
FirstFirst ... 12 52 60 61 62
Results 1,831 to 1,847 of 1847
Thread
  1. Originally Posted by Selur View Post
    Did you try the dev version I send you via PM?

    Cu Selur
    Didn't see the notification until I saw your post. Just tried it with the same file I used before and the 24bit audio stream came through properly. I will continue with my encodes, I'll let you know if the problem resurfaces.
    Quote Quote  
  2. Happy that worked so far.
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  3. I use the option "create qp file for chapters" to place keyframes at where I want to cut a video. This is usually exact, the keyframe is right where I wanted it. But, occasionally, there are two anomalies.

    First, the keyframe is one frame after the point I set. There was no mistake in the time I set in the source, I checked. I've seen this in sequences where the keyframe is in the middle of a blank sequence, as well as where there's content on both sides of the keyframe.

    Second, there is a keyframe where I wanted it, but the frame after it is also a keyframe. At first, I thought it was because the minimum keyframe was set to 1. But this issue still appears after I changed the keymin to a larger number.

    I can work around these two anomalies by encoding a chapter specifically, but this is not always an option.

    These anomalies are very consistent. The same chapter time, in the same source video will always result in the same anomaly, even in several versions of Hybrid. So I can test this easily if you've got a fix.
    Quote Quote  
  4. [qutoe]First, the keyframe is one frame after the point I set[/Quote]
    Hard to tell which frame is placed by the encoder due to some rate control and which is placed due to the chapter/frametype file of Hybrid.
    Enlarging the lookaheads or using 2pass encoding might help.

    I don't thing Hybrid is here to blame.
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  5. Originally Posted by Selur View Post
    I don't thing Hybrid is here to blame.
    But Hybrid is the encoder I'm using. So if you can help, I'd appriciate it.

    By "lookaheads" do you mean the "Frame Lookahead" settings? Mine is currently 70/6/0, they're either default or I copied them from a source, I have no idea what the numbers mean. What do you suggest changing them to?
    Quote Quote  
  6. But Hybrid is the encoder I'm using.
    Hybrid isn't an encoder it's a graphical frontend for multiple encoders and you didn't care about mentioning what encoder you are using,....
    Hybrid is aimed as advanced users so it assumes quite a bit of knowledge or motivation to look stuff up from it's users.

    or I copied them from a source
    That usually is not a good idea when reencoding for 90% of all settings since the source of the original encode and the source you feed the encoder normally differs quite a bit. (assuming you use lossy encoding)

    I have no idea what the numbers mean.
    Then you should read up on them in the documentation of the encoder (https://x265.readthedocs.io/en/latest/) and you should also know at least some basics about the format you are encoding to.

    By "lookaheads" do you mean the "Frame Lookahead" settings?
    Assuming you are using x265 then those are the lookaheads for the different frame type I/P/B.
    And yes those are the setting I were referring to.

    What do you suggest changing them to?
    For the I frame lookahead the gop size as a value might help.

    You might have better luck asking of general help with x265 settings here in the forum than specifically in this thread.

    Some further thoughts about your 'problems':
    First, the keyframe is one frame after the point I set
    Since a frame is usually shown a bunch of ms and Hybrid will most likely have to round to decide which frame your time stamp refers to, you might want to take that into consideration when deciding on the times you use.

    Second, there is a keyframe where I wanted it, but the frame after it is also a keyframe.
    Where is the problem with that? At least from Hybrids point of view everything worked as it was asked for. You told it to tell the encoder to create a key frame at position X, it told the encoder to do so and it even decided to follow that instruction. That the encoder later also decided to create a key frame for the next frame too isn't really a problem is it. My guess is that making sure the min gop size > X should force the encoder to avoid this unless it is forced to use a key frame to abide vbv restrictions.

    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  7. Originally Posted by Selur View Post
    Assuming you are using x265 then those are the lookaheads for the different frame type I/P/B.

    For the I frame lookahead the gop size as a value might help.
    Yes, I am using x265. gop size would be referring to the keymax value, right?

    Since a frame is usually shown a bunch of ms and Hybrid will most likely have to round to decide which frame your time stamp refers to, you might want to take that into consideration when deciding on the times you use.
    I'll experiment a bit. To clarify, I use Aegisub to get the time of a frame. Its media player can step through a video frame-by-frame and it shows the time to the milisecond. I put that number as the chapter point.

    Where is the problem with that?
    Occasionally, I want to cut a video. If there are two adjacent keyframes, mkvmerge sometimes (not always) cuts at the wrong one. I got adjacent frames in my latest project, but the cut was at the correct frame, thankfully.

    My guess is that making sure the min gop size > X should force the encoder to avoid this unless it is forced to use a key frame to abide vbv restrictions.
    I tried that already. The keymin was set at 1 in the settings I initially used. Nowadays, it's 6, and I still see adjacent keyframes.
    Quote Quote  
  8. gop size would be referring to the keymax value, right?
    Assumign you used closed gops: yes.
    To clarify, I use Aegisub to get the time of a frame. Its media player can step through a video frame-by-frame and it shows the time to the milisecond. I put that number as the chapter point.
    You best would be using the decoder that is used during your conversion, depending on the decoder used times can easily be off by 1-2 frames.
    Occasionally, I want to cut a video. If there are two adjacent keyframes, mkvmerge sometimes (not always) cuts at the wrong on
    Then you might want to look into how you use mkvmerge.
    The keymin was set at 1 in the settings I initially used. Nowadays, it's 6, and I still see adjacent keyframes.
    then the encoder is probably forced to act that way due to vbv constrains or 1pass rate control.
    You would have probably to ask the x265 dev why this exactly happens in a given example.

    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  9. Originally Posted by Selur View Post
    Assumign you used closed gops: yes.
    I do. I'll give it a try and see if that does anything.
    You best would be using the decoder that is used during your conversion, depending on the decoder used times can easily be off by 1-2 frames.
    I didn't set a preferred decoder for Hybrid when encoding. What would it use in that case, and how would I be able to check the video time using it?
    Quote Quote  
  10. I'm looking for some help in working with mpls playlists because I am doing something wrong. There's a video I want to encode that's in two pieces. It's a BD raw so I thought to put the mpls that combined them into the input and see how it came out. Since it would be a lengthy encode, I thought I'd test by encoding a short mpls (of 4 seconds) first.
    At first, Hybrid rejected the input and suggested I clear the "Ignore playlists with..." options, which I did. Then, rather than just the 4-second video that played when I opened that mpls with MPC-HC, Hybrid proceeded to encode the whole video on the disk, a 90-plus minute item. Since that wasn't what I wanted, I aborted the process.
    So, I would like to ask, what's the right way to use an mpls playlist as input in Hybrid?
    Still, I saw that Hybrid combined the audio from the various video segments properly. If I can figure how to get just the segments I want, this will do.
    Quote Quote  
  11. Hybrid should read all the m2ts files that are refereed to in an mpls if the mpls if read as input.
    No clue why (or how) Hybrid would read more m2ts files than those listed.
    If the segment is a 4 second file simply opening the m2ts file might work.

    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  12. Originally Posted by Selur View Post
    If the segment is a 4 second file simply opening the m2ts file might work.
    Cu Selur
    Yes, but I wanted to test using mpls as input.

    Anyway, I figured out an even easier way to test, I just used the passthrough option for the video. This lets me see what Hybrid considers to be the input. And the result is strange. I've tried 3 different mpls from this disc and Hybrid only compiled the "default" mpls of the disc, which wasn't one of the three I used, in every case. In one case, the mpls had totally different chapter times from the default (because it was a different video) but the resulting video (once again, the default) had the chapters from the selected mpls, which was totally wrong.
    I'm going to try with some other discs I have later and see what happens. In the meantime, I might be able to work with this. I have the joined audio streams of the two segments I wanted. I have to make sure there are no timing discrepancies, but I can encode the videos seperately and combine them afterwards.
    Quote Quote  
  13. I'm looking at the mkv that was produced by using the mpls as input from yesterday, and I've got results I don't understand. My impression is the "passthrough" option does no encoding, correct? And keyframes are created during encoding, correct? Therefore, passthough video should not change the keyframes on a video, correct?
    But this mkv has fewer keyframes than the source it came from, I compared it to the source. The distribution of keyframes look more like that of an encoded video than a BD source. Can passthrough drop keyframes without reencoding?

    Update: It's got something to do with using mpls as input. I joined the segments using mkvmerge and then used passthrough on the result, the two videos had the same set of keyframes.
    Last edited by Compositor; 23rd May 2020 at 23:49.
    Quote Quote  
  14. My impression is the "passthrough" option does no encoding, correct?
    Correct, otherwise it would take a lot longer processing the streams.

    And keyframes are created during encoding, correct?
    yes

    Can passthrough drop keyframes without reencoding?
    No, dropping key frames would break the stream since the reference of all other frames would be missing.

    -> Are you sure the tool you use to check the key frames differs between I- and IDR- frames?

    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  15. Originally Posted by Selur View Post
    -> Are you sure the tool you use to check the key frames differs between I- and IDR- frames?

    Cu Selur
    Considering the tool I used is Aegisub, I doubt it can. But it's as I said, the keyframes are there in the source when I look at the video in Aegisub. When I look at the passthrough created video in Aegsiub as well, many of the keyframes are no longer there. They've become non-keyframes. Even if Aegisub didn't consider one of those types of frames to be keyframes, the passthrough process would have needed to change something about those frames so that Aegisub would no longer recognize them as keyframes, right?
    Quote Quote  
  16. May be something in the container changed, no clue. I do not see a way how Hybrid could change the key frame info.
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  17. Maybe it has something to do with how mpls presents the video segments to other applications.

    Considering the weird results, I'm putting aside working with mpls for now. I'll just stick to joining the m2ts with mkvmerge and deal with any discrepancies.
    Quote Quote  



Similar Threads