Gareth Randall
Forum Replies Created
-
CT changes the sample rate to 50.4 KHz. If it causes an audio sync problem, which as I say is about 50% of the time, it happens in FCP and any media player you try and play the file in.
FWIW the workaround is simple but still a chore. I export the audio from the 23.98 version, note the exact duration of the 25fps version, and then use Audacity to speed up the 23.98 audio to match (with pitch change). Then I just paste it back onto the 25fps version in QT Player 7.
I did try doing the speed change in Soundtrack Pro, but that seems to insist on chopping out samples to retain the original pitch, which causes some very audible artefacts. I haven’t found a way of doing an Audacity-style speed-and-pitch change in STP.
-
Just tried changing the framerate within Compressor – predicted transcode time went up from 4 hours to 26 hours on a 2-machine cluster. Ouch.
-
Just to be clear, you want to make the word “ways” sound like it’s the natural end of a sentence?
Try and find a sentence from your talent that ends on the same sort of “zzz” sound as “ways” does. It doesn’t have to be the same word at all, just a word that ends in the same sound – you could steal the “zzz” ending from “guns” or “cars” or anything like that – and which also ends the sentence so it has a natural downward inflection.
As has been said, you’ll probably need to use an audio editor for the job because you’ll probably need to work at much greater level of accuracy than FCP will allow.
I worked in radio production for years before moving to TV, so this sort of thing cropped up fairly regularly. There are lots of little tricks you can use; for example, some sounds are unvoiced (such as the “zzz” we’re talking about here), so you can actually take a man’s voice saying “guns” and a woman’s voice saying “cars” and cut the ending of “cars” onto “guns” and it will still sound completely natural and not like two different-gender voices.
-
There’s no plugin that will do this for you automatically (and that goes for *any* compositing app). You’ve got the basic technique right, you just have to create a freeze-frame and cut your guy out of it each time, but you don’t have to export anything to Photoshop, you can do the whole thing in Motion.
Stephen Smith has a tutorial that covers very similar ground here: Time After Time tutorial
-
I’d like to clarify something if possible: when you start a processing job, does it immediately take priority control of all the computers in the cluster, or can the Macs be used for other things while processing happens in the background?
What I’d like to be able to do is set a job going, and then go home – my iMac would immediately be idle and available to transcode, but other Macs in the cluster may still be in use for FCP work for some hours after that. Do all the computers have to be idle before you start a job, or does Qmaster manage them so that the ones that are busy doing other things aren’t brought fully online for the processing until they become idle?
-
Fantastic, thanks 🙂 I’ve found a Vimeo tutorial that covers setting up QMaster, looks pretty straightforward.
-
There’s nothing that will do it automagically, but check out this post for a typical way of doing it manually.
-
I seem to have fixed it, but I’m not entirely sure why it’s worked!
I applied the Threshold filter to the original card to create a hicon B&W clip, rendered that out and reimported it and did a 3-point track on the text. The fact that it was quite sharp-edged B&W helped the tracker retain a better lock during the frames where the card was moving the most and the text was motion-blurred.
Then I just applied Match Move to the text and played with the sources of the Anchor and Rotation-Scale parameters until I got what I wanted. Turns out that to make it work, Anchor needed to follow Track 3 and Rotation needed to follow Track 1. Any other combinations gave me wrong results.
-
I’m just trying to understand the ultimate difference – if any – between conforming a 23.98 clip to 25fps in CT, and dropping the “raw” 23.98 clip into a 25fps FCP sequence.
If I do the latter, I get one repeated frame every second – which makes sense as being FCP’s quick’n’dirty way of converting 23.98 to 25. It maintains the original speed, more or less, but at the expense of smooth motion.
If I go the CT route, I’m guessing that I’ll get smooth motion with no repeated frames, but at the expense of the clip running 4% faster (which isn’t really a problem in most circumstances)?
-
If the image isn’t moving, then either you haven’t properly associated a music track with the audio behaviour, or you haven’t associated a transformation to the behaviour, or the transformation scale hasn’t been set high enough.