The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Quick Tremolo in dialogues Effects Pedals, Units & Accessories
Old 2 weeks ago
  #1
Here for the gear
 

Quick Tremolo in dialogues

Hello!

A client has sent me an old movie to restore.

One of the problems that I have encountered is that a lot of dialogue in the film suffered from a "Quick Tremolo" effect, probably due to bad scanning and/or bad archiving conditions of the film material.

Rather than the pitch (wow & flutter issues), it is the INTENSITY that wobbles around.

Example picture:
(I cannot post the picture directly to this post because the site does not allow me to yet)

problem — imgbb.com


Is there a way to fix this? (if I could use iZotope RX Advance, better, but I am willing to solve this using other software if needed).


Thank you all! This site and its comunity are invaluable!
Old 2 weeks ago
  #2
Lives for gear
When I read your descriptioni, I thought it had to be related to gate flutter. But the pix doesn't look like 24Hz pattern... it looks even faster than 96Hz, which is related to 35mm's perf rate.

If the frequency is constant, and the only thing is irregular level but the softer signal never goes completely away, you could throw together a quick plugin in Sonic Birth or similar that uses an LFO (at the critical freq) to generate a sine that constantly fades between two gain paths. It would take some trimming of the levels and probably some LFO tweaking over the film's length. Getting the initial sync right would be hard, but a plugin should stay stable.

Or even put the film on two tracks. Hand-draw fader curves that alternate properly for the first second or so, then copy and paste... just turn off your controller's fader motors while it's running, to save wear and tear.

Or even do it on one track with hand drawn gain for the first second.

Or generate an LFO at the critical freq on one track, manually slip so it's in sync with the problem, and use it as sidechain for a compressor or expander on the critical track. That might run into a problem with time constants in the dynamic plug-in, however...

...or combine the first and last technique, and build a plug in that uses an external key to vary the gain on the main channel... like a synth VCA, only tuned to compensate for the problem.
Old 2 weeks ago
  #3
Here for the gear
 

Jay Rose, thank you very much for your ideas and for taking the time to help me!

I got in touch with the client and he told me that the source material is 16mm. My guess is that their scanner is more prepared to recieve a 35mm reel, and that he didn't make proper adjustments in order to prevent the film to rattle within the rails while being scanned, thus transfering that movement to the sound capture. That didn't occurred to me before you saying "35 perfs", so thank you very much!

If he can't fix this issue, I will try to do one of your methods, but I reckon it will be hard to pull off, since the "attack and release" (meaning the slope) of the sine wave (lfo) would have to match the cadence on which the tremolo increases and decreases the volume... otherwise I could be worsening the effect... with that and with sync problems.

It would be great if there were some module or software that tackled this tremolo issue... like a "celemony capstan" for intensity instead of pitch.

Thank you my friend!
Old 1 week ago
  #4
That looks very much like a film threading problem. You have to solve it at the source and repeat the transfer. Even if you manage to correct the amplitude, you won't be hearing the full contents of the material, because optical reproduction depends on track position and focus.
16mm optical is not a great recording media, that's why you can't afford to lose anything at the transfer.
There was a guy in London, specialised in 16mm optical, Martin Sawyer. He was capable of reading impossible subtleties from most damaged prints. Maybe you should try to find him, if he is still working...
Old 1 week ago
  #5
Lives for gear
General query:

Does anybody make (or made, last century) an optical audio transfer device without a gate?

This would be a smooth threading from one reel to the other, like a tape recorder...
no sprockets to move the film, just an idler and a capstan carefully milled to avoid the image and track areas...
no loops...
an optical sensor sees the perfs (or a film-driven sprocket with a tach, if need be) to regulate speed...
and an adjustable optical sensor to read the track.

(Well, maybe a motorized sprocket/tach to drive the film and regulate speed. But with a big flywheel. And rolling guides elsewhere.)

Film equivalent of the infamous MCI 1" 9.6 ips Type C audio deck. Which IMHO didn't catch on not because it wasn't a good idea, but because every post house that could afford one would have a spare Sony 1" deck somewhere for laybacks. And the audio department would need access to one of the video decks anyway, to dub into their HDVR or 3/4". So for most of the market, it just wasn't a great investment.

Somebody must have done something like this.
Old 1 week ago
  #6
Gear Nut
 

Sondor used to. Their technology has changed owners:
dft acquires Sondor Technology
Old 1 week ago
  #7
Quote:
Originally Posted by Jay Rose View Post
General query:

Does anybody make (or made, last century) an optical audio transfer device without a gate?

This would be a smooth threading from one reel to the other, like a tape recorder...
no sprockets to move the film, just an idler and a capstan carefully milled to avoid the image and track areas...
no loops...
an optical sensor sees the perfs (or a film-driven sprocket with a tach, if need be) to regulate speed...
and an adjustable optical sensor to read the track.

(Well, maybe a motorized sprocket/tach to drive the film and regulate speed. But with a big flywheel. And rolling guides elsewhere.)
Somebody must have done something like this.
Several manufacturers did it and Blackmagic's (Cintel) telecine still does it. The best transport system , imo, was made by Albrecht, aka MWA. Their MB-51 followers and recorders didn't use the capstan and pinch roller, controling the speed by tension of the supply and take-up reels.
Attached Thumbnails
Quick Tremolo in dialogues-albrecht115-2.jpg  
Old 1 week ago
  #8
Lives for gear
 

Yep we gave away our last MB51 a few years ago to the Swedish film institute to use for exactly this purpose. We had mounted a optical reader that could run both red and white light.
When I started in film in 93-94 we had six MB51 and five older MB41 we even still had even older MB models (they used to require a "roto syn engine" to run them in sync.we also had a bunch of Sondor units we dubbed them "sönder" which means broken as they broke repeatedly.

And then there were computers (and a few tapes in between).
Old 1 week ago
  #9
Lives for gear
 
iluvcapra's Avatar
Quote:
Originally Posted by Jay Rose View Post
General query:

Does anybody make (or made, last century) an optical audio transfer device without a gate?
Totally spitballing here but it seems like if you had a full-aperture 4k scan of a frame with standard 4k film scanning equipment, you'd be able to reconstruct the audio from a scanned image... Vertically you have 2160 samples per frame, 2160 * 24fps = ~51 kHz sampling rate, and that significantly exceeds the performance of the film originally. The challenge would be getting much dynamic range out of it I suppose, the sound track is only about 11% of the aperture area, so that's 4096 * 0.11 = 450 dynamic levels or about 53 dB total dynamic range. (I dunno how SVA affects this, in the worst case it'll knock ~6 dB off this number). Then again I'm pretty sure I've read 48ish dB as being the design performance of un-Dolby optical audio.

Okay someone go build it!
Old 1 week ago
  #10
Quote:
Originally Posted by iluvcapra View Post
Totally spitballing here but it seems like if you had a full-aperture 4k scan of a frame with standard 4k film scanning equipment, you'd be able to reconstruct the audio from a scanned image... Vertically you have 2160 samples per frame, 2160 * 24fps = ~51 kHz sampling rate, and that significantly exceeds the performance of the film originally. The challenge would be getting much dynamic range out of it I suppose, the sound track is only about 11% of the aperture area, so that's 4096 * 0.11 = 450 dynamic levels or about 53 dB total dynamic range. (I dunno how SVA affects this, in the worst case it'll knock ~6 dB off this number). Then again I'm pretty sure I've read 48ish dB as being the design performance of un-Dolby optical audio.

Okay someone go build it!
You can use any scanner that captures full frame and has a sufficient overlap. After scanning, you can process it with something like AEO-Light
With this software, you can decode both Variable Area and Variable Density tracks, the latter depending heavily on the print quality. SVA tracks can be easily enhanced in any Photoshop-like app (if you have the patience!), removing scratches manually and improving the s/n by removing the grain in white (transparent) areas. You can then experiment with contrast to get the best distortion figures...
Old 1 week ago
  #11
Here for the gear
 

This is a title

Quote:
Originally Posted by Branko View Post
That looks very much like a film threading problem. You have to solve it at the source and repeat the transfer.
I was lucky enough to have that luxury and I got another scan.

But this issue that I had to deal with made me think about those times when you cannot get another transfer and you have to work with what you have...

restorers find ourselves in this situation quite a lot.

I find rather weird that there is not plug in or module that fixes high speed fluctuations in amplitude (a plug in that could "learn" the fluctuation frequency and adjust the gain accordingly).

The same problem is quite common in imaging restoration... fluctuations in the exposure of a film, and the alogithms that deal with that are quite common... I suppose that doing the same for audio wouldn't be much more difficult.
Old 1 week ago
  #12
Lives for gear
 
iluvcapra's Avatar
Quote:
Originally Posted by mfzb10 View Post
I was lucky enough to have that luxury and I got another scan.

But this issue that I had to deal with made me think about those times when you cannot get another transfer and you have to work with what you have...

restorers find ourselves in this situation quite a lot.

I find rather weird that there is not plug in or module that fixes high speed fluctuations in amplitude (a plug in that could "learn" the fluctuation frequency and adjust the gain accordingly).
What was the rate of the tremolo finally? If you were to low pass filter it at, say, 10 Hz, would you get a signal that was just the gain variation? I can't really tell from the iZotope image what the rate is, the image might be aliasing.

There are ways of extracting cyclical signal variations, but it's more of a statistical thing, like by using moving averages.
Old 1 week ago
  #13
Lives for gear
 

At this point, most of the big restorations have been done. I am not sure that there is the money in it to be worthwhile to develop.

Randall


Quote:
Originally Posted by mfzb10 View Post
I was lucky enough to have that luxury and I got another scan.

But this issue that I had to deal with made me think about those times when you cannot get another transfer and you have to work with what you have...

restorers find ourselves in this situation quite a lot.

I find rather weird that there is not plug in or module that fixes high speed fluctuations in amplitude (a plug in that could "learn" the fluctuation frequency and adjust the gain accordingly).

The same problem is quite common in imaging restoration... fluctuations in the exposure of a film, and the alogithms that deal with that are quite common... I suppose that doing the same for audio wouldn't be much more difficult.
Old 1 week ago
  #14
Here for the gear
 

This is a title

Quote:
Originally Posted by quadraphonics View Post
At this point, most of the big restorations have been done. I am not sure that there is the money in it to be worthwhile to develop.

Randall
Maybe, but I still see some use to it. I've restored more than 60 titles so far, and counting.

A module like that could be used for forensics... maybe to lower the amplitude interval of a recorded tremolo?

Who am I kidding... this module is not that needed
But it would make the hell of a fun project to program!
Old 1 week ago
  #15
Here for the gear
 

Quote:
Originally Posted by iluvcapra View Post
What was the rate of the tremolo finally? If you were to low pass filter it at, say, 10 Hz, would you get a signal that was just the gain variation? I can't really tell from the iZotope image what the rate is, the image might be aliasing.

There are ways of extracting cyclical signal variations, but it's more of a statistical thing, like by using moving averages.
I don't think the image was aliasing, and if it was, I still could count 15 fluctuations by hand in 0.1 seconds.

15 in 0.1 is 150 hz.

My educated guess is that the film wasn't properly secured to the scanner's rails, thus moving (ratling) 150 times per second.

Either that, or the scanner's speed was decalibrated for this kind of material by an interval resulting in a 150hz artifact.
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
xecutionist007 / Post Production forum
6
Hank_Chinaski / Post Production forum
7
mark007 / Post Production forum
9
Neil Whippey / Post Production forum
13

Forum Jump
Forum Jump