[Matt Lyon]

[David Lunday] “When you say redundant frames are you referring to the 59.94fps footage? OR, are you saying to take all the original 23.98fps 720p60 footage and put it through compressor to encode it with 720p24 to remove the redundant frames you are referring to? If that were the case then it seems I would just put the entire sequence through compressor with the 720p24 encoding.”

David, try opening up two of your original clips in quicktime player. One should be a 23.98 clip and one should be a 59.94 clip. Step through both clips a frame at a time, using the right arrow key. You’ll see that in the 23.98 clip, every frame is a unique snapshot of a moment in time. For the 59.94 clip, you’ll notice a pattern of duplicate (what I am calling “redundant”) frames.

So there is no way to remove redundant frames from 23.98 footage, since they do not exist. It is the 59.94 clips we need to fix.

[David Lunday] “I simply took one of the 59.94fps shots (with it’s compression being 720p60) and encoded it with the 720p24…which of course turns it into 23.98fps…then just put it back into my sequence over the old clip so it matches all the other stuff…and of course I would do the same thing to all the other clips which are still 59.94fps. Is this what you were referring to? “

Yes, this is exactly what I meant. Convert the original, “bad” 59.94 footage to 23.98 and cut it back into the sequence.

But that being said, you need to make sure Compressor is doing its job properly. Open the new 23.98 clip in quicktime player and step through it one frame at a time. Is each frame unique? Or are there still duplicate frames? Use a section where these is lots of camera motion, because it is easier to spot duplicate frames. Also, watch out for new, or “interpolated” frames, which are a blend of two frames. This is also a problem. If you find Compressor is creating these interpolated frames, you should try setting the frame rate conversion method to “nearest frame.” Sometimes it takes several sets of test to find the right combination of settings. It sounds like Shane has got this to work, maybe he can elaborate on the settings he used.

HTH,

Matt Lyon
Editor
Toronto

[Matt Lyon]

Thanks for the info Dave, I was not aware of that. Cinema Tools probably doesn’t know what to do with that kind of footage. But is there no way to remove the redundant frames? Via Compressor maybe? Without having any footage to test, I can’t be of more help. But I still think the best course of action is to convert the original footage back to 23.98 (via proper redundant frame removal) and conform the shots back into the timeline.

Matt Lyon
Editor
Toronto

[Matt Lyon]

I’m not too familiar with this camera, but it sounds like the footage has “film style” 3:2 pulldown? (In other words “AA/BB/BC/CD/DD” field cadence).
If so, you should deliver at 23.98. Corral all the “wrong” 59.94 clips in the finder. Duplicate them and remove the pulldown using cinema tools. Many posts describing how to do this can be found on this forum, or in the cinema tools manual.
Verify that the pulldown has been successfully removed and import the new 23.98 footage into FCP. You will need to replace all the bad shots with the fixed media. Work on a new video track so you can A/B your work. Once you’ve replaced everything you can send a 23.98 QuickTime out to your post house.
HTH,

Matt Lyon
Editor
Toronto

[Matt Lyon]

Thanks Michael, and thanks for all the links to my article! Much appreciated.

Matt Lyon
Editor
Toronto

[Matt Lyon]

Hi Paul, this is a pretty common problem and one I discuss in my tutorial. You can skip to the second part, which describes how to set up FCP so audio imports correctly:

https://library.creativecow.net/lyon_matt/fixing-fcp-assets/1

HTH,
Matt

Matt Lyon
Editor
Toronto

[Matt Lyon]

Scott, I’ve also struggled with this over the years and you can get decent results using the compressor/limiter filter. But you won’t find a “one size fits” all solution, because how audio filters behave depends entirely on the nature of the source material. But hopefully I can give you a starting point:

1. nest your audio tracks (or, better yet: “bounce” them to disk, by exporting your timeline to a self contained AIFF file, using FILE>export>using quicktime conversion)
-You may want to do this step to your whole timeline, or just your DIALOG and SOUND FX, because music tracks are usually heavily compressed already.

2. Apply the “compressor/limiter” filter.

3. IMPORTANT: Turn OFF the “preserve volume” option. This is critical. If you leave this option on, then FCP will basically just BOOST your quiet sounds to achieve dynamic range compression. If you turn this option off, then FCP will lower (aka attenuate) the loud sounds in order to achieve dynamic range compression. With the “preserve volume” on, FCP is doing what other programs call “auto gain”; the result is your levels are just as hot, so no audio limiting is effectively taking place.

4. Now you have to dial in the settings to get the proper amount of attenuation. Start with a very low threshold (-6 to -18). You only want the filter affecting the loudest sounds. The exact setting you use depends on the peak levels of your source material, so every program will be different.

5. Now set your “Ratio.” This is the amount of attenuation applied. I don’t know exactly the relationship between this number and dB, but I /think/ that “-2” equals 6 dB of attenuation. So that is a good starting point.

5. Now adjust the attack time and release time. I start with 0 attack and maximum release (4000) (because you want the filter to be “on” all the time). But this can sometimes create too much signal limiting. Use your ears and find the settings that suit your material. But I usually stay with “0” attack, and never go below “200” for release, but YMMV.

6. Now that you’ve tamed the dynamic range of your program, you may want to apply “normalization gain” (Modify > Audio > Apply Normalization Gain). This will let you actually set a “hard maximum” for your audio level. I’ve found the number I enter in the option box doesn’t seem to create the expected results, so some fudging may be required. Normalization gain works MUCH better after you’ve compressed your dynamic range, by the way.

7. If you’ve left your music tracks unfiltered, you may now need to give them a global volume adjustment in order to “sit” better with the rest of the material.

Matt Lyon
Editor
Toronto

[Matt Lyon]

It’s been a long time since I captured any minDV footage, so I may be off base here, but: did you shoot the original footage on a consumer camera? I remember that some of those cheaper cameras could create some funky audio issues. The solution unfortunately might be that you need to capture the footage from the same camera it was shot with. Putting those tapes in a professional deck could be what’s creating the problem. I’m speculating, but this is my initial hunch.

If you can’t find the original camera, it might be worth a test using a “consumer” grade device, instead of a DVCam deck (I’m guessing this is the device you are currently using to capture?)

hth,

Matt Lyon
Editor
Toronto

[Matt Lyon]

Thanks for the update Lars,

That makes a lot of sense. I was re-reading the thread and was thinking that since you also had the sync problem with Adobe Premiere, that means the issue is likely with the source material, not the software. And given that your thorough testing could not fix the problem, it all makes sense now.

I love Nagras too! But have you tried the Zoom h4n recorder? It’s pretty good for the price! (Although I haven’t tried running it for a 70 minute take).

Matt Lyon
Editor
Toronto

[Matt Lyon]

I agree that the compressor filter isn’t great, but it should work better than the original poster describes. In my experience, turning off the “preserve volume” setting yields more predictable results. Once you turn off this check box, the track will no longer sound louder. Instead, FCP will just limit the maximum amplitude of the clip in question. Now you can apply the normalization gain filter to bring the levels back to where you want.

Although in my experience, the normalization gain seems to always set thing LOWER then where I set the slider (although I’m only judging the results using the built in meters).

I also like to nest the tracks in question that I want to compress and normalize, so I can apply the effects globally (this works well for reverb too). If you don’t like nesting, you can also export audio stems of select tracks and bring them back into the timeline.

Matt Lyon
Editor
Toronto

[Matt Lyon]

Lars, the 100.1% speed change you are applying is consistent with the difference between 24.00/23.976 fps … in other words, standard audio pulldown that is used for film workflows. As described in my tutorial, this is happening because something is preventing the audio from importing properly, and FCP is applying the incorrect NTSC flag. Look for the telltale orange bars I describe overlayed on your audio files…

If you’re audio was being imported properly, the NTSC flag would be set to FALSE! So you need to revisit the steps you are taking when importing the audio, and figure out what is going wrong exactly. Your sequence preset should be set to a 23.98 frame rate. Try importing the files in a new, empty project and then dumping out a quick XML to verify the import worked.

I can’t imagine that disk speed is an issue, since you are doing a completely file based workflow, no? There is no “digitizing” to speak of.

I don’t have direct experience with the camera and audio recorder you mention, so I can’t say whether something specific to those devices is creating issues too.

Hope this helps,

Matt Lyon
Editor
Toronto