Timecode Discrepancy

RE: Timecode Discrepancy (soundfreely)

Mon, 27 Jun 2005 04:48:38 GMT

ORIGINAL: SonicPadawan

ORIGINAL: soundfreely

You may want to try working to an AVI or Quicktime using either DV, Cinepak, or Sorenson3 (or something else I am forgetting about). Sometimes MPEGs can cause sync problems.

Hmm, I didn't consider compression to be an issue but your description makes a bit of sense. But really, isn't a frame just a frame regardless what method is used to arrive there? FCP seems to prefer Quicktime so that's where we're working with, it's been good to us so far. Now, the real trick for me has been deciding what format to use. At first I insisted on DV/DVCPRO-NTSC because of Sonar's key frame need. Is DV/DVCPRO-NTSC what you meant by DV? It worked well enough in the video portion but the image quality seemed to suffer and the file size was insane. I then requested MPEG4 with key frames every 3 frames. The image quality improved (visual quality for my benefit only), the file size improved, and Sonar didn't choke on the video.

You may want to check if MPEG4 uses a similar I and P frame compression method that MPEG2 uses. Let me know what you find as I'd like to know too! And, yes, I did mean DV-NTSC (to others who may read this: it would be PAL if it were 25 fps).

PS As I assume you and your video editor already know, changing between 24 fps and 29.97 fps requires a 3:2 pulldown (same as the telecine process when refering to transfering film to video). You may need to slow the audio clock by .1% if the frames are converted to 30 fps as opposed to 29.97.

While we do know that, I didn't know about the audio clock adjustment. We are using 29.97 though, so this adjustment isn't necessary if I read this correctly.

From my point of view, it seems the video I received from the other editor skipped a frame. With that visual evidence it makes me wonder if it was dropped during render/export. When my editor returns next week I'm going to ask to get a timecode reader imprinted on the video and see if sync is lost there. Thanks for the insights.

When converting film to video, 24 fps is converted to 30 fps. The 3:2 pulldown means that there are 3 fields of one 24 fps frame going to the 30 fps and then 2 fields of the next 24 fps frame going to the next 30 fps frame (I wish I could have worded that one better). However, the 30 fps still needs to be slowed to 29.97 for NTSC spec and that is why the audio must be .1% slower. You may want to check if FCP is slowing the audio for you during the 24 fps to 29.97 fps conversion.

-Erik

Edit: better explanation of 3:2 pulldown- first 24 fps frame = 3 NTSC fields, 2nd 24 fps frame = 2 NTSC fields, 3rd 24 fps frame = 3 NTSC fields, and so on. Also, 2 fields = 1 frame of interlaced video.

RE: Timecode Discrepancy (tarsier)

Mon, 27 Jun 2005 02:39:52 GMT

ORIGINAL: SonicPadawan

ORIGINAL: tarsier
I agree with soundfreely. While I havn't had much experience with mpeg4 videos yet, we do a lot of transfers back and forth with FCP and Sonar here. Sorensen 3 seems to work ok, and my favorite is MJpeg.

Is that Motion JPEG A/B or Photo-JPEG? I recently read that Photo-JPEG is used by the sound editors, but I'm not sure what the difference between Motion and Photo is.

I'm not sure either." /> Try 'em all and see what works.

As far as the timecode goes, why are you (or your editor) exporting from 24 to 29.97fps? That only confuses the issue. (even though that is the scenario that is workig for you.)

24 fps is our final output. We've read (I'm not sure where) that it's best to edit sound at 29.97. So yes we remove the advanced pulldown for video editing then add pulldown for sound editing. So sayeth the article(s) we've read. I don't believe it gave a reason though. I then give a stereo mix to my editor to be placed back with the video. I honestly don't know if this is the best method but we haven't *noticed* any issues.

What is your souce format? Film? HD video? CG animation? Something else? 24 fps film will get transferred to video for editing purposes, but with FCP that's not really necessary... I try to avoid 29.97 like the plague and only use it when dealing with video. We're about half and half CG and video. If you're final output is 24 fps, I don't understand why any article would suggest editing at 29.97 (unless it's been converted to 29.97 video.)

... Also, make sure you are getting two-pops at the head and tail of your video and omf files. They can be a lifesaver.

... Now, the two-pop idea makes a lot of sense and I'm embarassed I didn't consider that myself. Would that be generated by something such as FCP or would I need to lift that from somewhere? Great suggestion, thank you!

Have your editor put in one frame of colorbars in the video track and one frame of a 1kHz sine at -20dBFS (you can provide the soundfile for that) two seconds before the start of the material and two seconds after the end of the material:

<beep><2sec. of video black><program start>...<program end><2sec. of video black><beep>

RE: Timecode Discrepancy (SonicPadawan)

Fri, 24 Jun 2005 11:41:34 GMT

ORIGINAL: soundfreely

You may want to try working to an AVI or Quicktime using either DV, Cinepak, or Sorenson3 (or something else I am forgetting about). Sometimes MPEGs can cause sync problems.

Hmm, I didn't consider compression to be an issue but your description makes a bit of sense. But really, isn't a frame just a frame regardless what method is used to arrive there? FCP seems to prefer Quicktime so that's where we're working with, it's been good to us so far. Now, the real trick for me has been deciding what format to use. At first I insisted on DV/DVCPRO-NTSC because of Sonar's key frame need. Is DV/DVCPRO-NTSC what you meant by DV? It worked well enough in the video portion but the image quality seemed to suffer and the file size was insane. I then requested MPEG4 with key frames every 3 frames. The image quality improved (visual quality for my benefit only), the file size improved, and Sonar didn't choke on the video.

PS As I assume you and your video editor already know, changing between 24 fps and 29.97 fps requires a 3:2 pulldown (same as the telecine process when refering to transfering film to video). You may need to slow the audio clock by .1% if the frames are converted to 30 fps as opposed to 29.97.

While we do know that, I didn't know about the audio clock adjustment. We are using 29.97 though, so this adjustment isn't necessary if I read this correctly.

From my point of view, it seems the video I received from the other editor skipped a frame. With that visual evidence it makes me wonder if it was dropped during render/export. When my editor returns next week I'm going to ask to get a timecode reader imprinted on the video and see if sync is lost there. Thanks for the insights.

RE: Timecode Discrepancy (SonicPadawan)

Fri, 24 Jun 2005 10:59:13 GMT

ORIGINAL: tarsier

I agree with soundfreely. While I havn't had much experience with mpeg4 videos yet, we do a lot of transfers back and forth with FCP and Sonar here. Sorensen 3 seems to work ok, and my favorite is MJpeg.

Is that Motion JPEG A/B or Photo-JPEG? I recently read that Photo-JPEG is used by the sound editors, but I'm not sure what the difference between Motion and Photo is.

As far as the timecode goes, why are you (or your editor) exporting from 24 to 29.97fps? That only confuses the issue. (even though that is the scenario that is workig for you.)

24 fps is our final output. We've read (I'm not sure where) that it's best to edit sound at 29.97. So yes we remove the advanced pulldown for video editing then add pulldown for sound editing. So sayeth the article(s) we've read. I don't believe it gave a reason though. I then give a stereo mix to my editor to be placed back with the video. I honestly don't know if this is the best method but we haven't *noticed* any issues.

One thing that can help is that Sonar 4 can display a framenumber for each frame in the video track. If you're handy with timecode conversions, you can see if or when the discrepancy is happening. Also, make sure you are getting two-pops at the head and tail of your video and omf files. They can be a lifesaver.

I'm not handy with timecode conversions at this time, but that's a great method I wouldn't have considered. I'll do a little research on the method and see what that gives me. Now, the two-pop idea makes a lot of sense and I'm embarassed I didn't consider that myself. Would that be generated by something such as FCP or would I need to lift that from somewhere? Great suggestion, thank you!

RE: Timecode Discrepancy (tarsier)

Wed, 22 Jun 2005 09:35:47 GMT

I agree with soundfreely. While I havn't had much experience with mpeg4 videos yet, we do a lot of transfers back and forth with FCP and Sonar here. Sorensen 3 seems to work ok, and my favorite is MJpeg.

As far as the timecode goes, why are you (or your editor) exporting from 24 to 29.97fps? That only confuses the issue. (even though that is the scenario that is workig for you.) One thing that can help is that Sonar 4 can display a framenumber for each frame in the video track. If you're handy with timecode conversions, you can see if or when the discrepancy is happening. Also, make sure you are getting two-pops at the head and tail of your video and omf files. They can be a lifesaver.

A while back I discovered a timecode discrepancy of one frame in Sonar, but I reported it to Cakewalk support and they fixed it. But now I'm investigating Soundtrack Pro, since it has some terrific FCP integration features. We may just drop Sonar for soundtrack work.

RE: Timecode Discrepancy (soundfreely)

Wed, 22 Jun 2005 06:14:29 GMT

You may want to try working to an AVI or Quicktime using either DV, Cinepak, or Sorenson3 (or something else I am forgetting about). Sometimes MPEGs can cause sync problems. I don't know too many of the details of MPEG4, but I do know that MPEG2 is a nightmare to sync with due to its GOPs (Group of Pictures). This is because every frame is not discrete but rather a P Frame surrounding several I frames. The data for the I frames is only the difference between the nearest P frame and that makes working in a non-linear fashion a bit tricky. I wish I knew if MPEG4 uses a similar compression method. However, I can say that I've worked to both AVI and Quicktime using DV and I have had no sync issues inside or outside of Sonar.

HTH,
Erik

PS As I assume you and your video editor already know, changing between 24 fps and 29.97 fps requires a 3:2 pulldown (same as the telecine process when refering to transfering film to video). You may need to slow the audio clock by .1% if the frames are converted to 30 fps as opposed to 29.97.

Timecode Discrepancy (SonicPadawan)

Wed, 22 Jun 2005 01:38:14 GMT

Good day,

I'm using Sonar 4 PE for audio post production for video. In general I'd sure love to hear from others who are doing the same as it seems these forums are a musicians only club. However, I have a timecode question that I hope will prove to be a simple fix. This crosses the boundry between video and audio suites, please bear with me.

Normal situation:
My editor edits the video in 24 fps with 48kHz/16bit audio in Final Cut Pro HD. He exports the video for me to work as an MPEG-4 video, no audio, at 29.97 fps with a keyframe every 3 frames. He also exports an OMF at 48kHz/16bit, it's with this which I built my workspace. I'll open the OMF in Sonar then import the video, naturally the timecode is present with the process. I haven't experienced any difficulties with this workflow.

Current situation:
An outside editor edits the video project in 29.97 fps with 48kHz/16bit audio in Final Cut Pro HD. A Timecode Reader is rendered into the video (by FCP and not external hardware) then exported for my use as an MPEG-4 video, no audio, at 29.97 fps with a keyframe every 3 frames. I'm also provided an OMF in the same manner as described above with both being assembled as already mentioned.

The trouble:
Since I normally don't have a timecode in my video I can't say if there's a "sync" issue, but with the TWO projects I've been sent from this outside editor, I've noticed that our timecodes (generated by Sonar and by FCP) begin to show a discrepancy around the 1 minute mark. One project was around 00:00:55:10 and the other around 00:01:12:12. If I scrub frame by frame I see my timecode advance one frame while the video embeded timecode lingers at the same number for that frame. How many frames its off varies but it can be up to 12 seconds by the end of the 6 minute piece.

Now, I don't see this as a serious issue since when I watch the video my audio matches precisely, but the outside editor is getting quite nervous. Does anyone else have experience with this issue?