Jump to content

Is FilmFabriek aware that their sound scanner has audio problems?


Recommended Posts

  • Site Sponsor

There are two ways to really do this...

1. A sound speed crystal locked capstan that will give consistent sync results from analog capture with no wow or flutter.

This is how it has been done with telecine and resolvers like the MagnaTech for years.

1. A digitizer like a Nat Instruments card that treats the incoming audio as data and then a servo capstan with a high resolution quadrature encoder that the scanner system can correlate pulses per frame to the incoming audio data to keep them in sync and matched to the proper resultant frame rate at whatever scan speed the scanner is running at.

Xena Scan Station and Kinetta do the latter I believe, so they can scan and digitize audio at any scan frame rate.

  • Like 1
  • Upvote 1
Link to comment
Share on other sites

21 hours ago, Tyler Purcell said:

I've had no problems outside of the drive motor being a bit inconsistent, so the audio has a bit of wow in it. 

That is what I mean. Either audio is right or not.

They can make a $100 cassette machine play right, or at least they used to. Why not a machine for $45k?

Have you talked to FF about it? 

Link to comment
Share on other sites

2 hours ago, Robert Houllahan said:

There are two ways to really do this...

1. A sound speed crystal locked capstan that will give consistent sync results from analog capture with no wow or flutter.

This is how it has been done with telecine and resolvers like the MagnaTech for years.

1. A digitizer like a Nat Instruments card that treats the incoming audio as data and then a servo capstan with a high resolution quadrature encoder that the scanner system can correlate pulses per frame to the incoming audio data to keep them in sync and matched to the proper resultant frame rate at whatever scan speed the scanner is running at.

Xena Scan Station and Kinetta do the latter I believe, so they can scan and digitize audio at any scan frame rate.

Thanks Robert. So what is wrong with FF's scanner? Is the optical reader designed wrong?

Link to comment
Share on other sites

  • Premium Member
12 minutes ago, Daniel D. Teoli Jr. said:

That is what I mean. Either audio is right or not.

They can make a $100 cassette machine play right, or at least they used to. Why not a machine for $45k?

Have you talked to FF about it? 

I have talked to FF about it, they didn't understand what I was talking about really. 

I think the problem is the tension motors, not the capstan itself. I think the capstan isn't strong enough to remove the tension motors which are not consistent. It needs a larger diameter capstan which can have more control over the film. Also, a cassette tape machine has a huge flywheel and pinch roller which holds the tape to the drive motor. There is no real way to do that on these machines. 

Magnatech gets away with this by having two big capstans. 

Link to comment
Share on other sites

2 hours ago, Dan Baxter said:

To be fair optical audio off a Lasergraphics is unmitigated shit. Professional companies just transfer optical audio separately.

That's completely ridiculous, and also untrue. It is in fact, among the best optical soundtrack reproducers currently available.  

We've had filmmakers who are intimately familiar with their own soundtracks tell us they've never heard their optical tracks sound as good as they do after we've scanned them.

Honestly, what is up with you and spewing information about a machine you have apparently never even used? Why do you insist on spreading misinformation like this?

Edited by Perry Paolantonio
  • Upvote 3
Link to comment
Share on other sites

8 hours ago, Perry Paolantonio said:

That's completely ridiculous, and also untrue. It is in fact, among the best optical soundtrack reproducers currently available.  

No it's not - what are you comparing it against? When I get a chance I'll put a piece of sound film through a ScanStation followed by a better audio machine. The Lasergraphics is very hissy. I personally think the software audio extraction (which is a free inbuilt software feature) does a better job than the Optical/Keykode reader, but I'll admit I haven't done a side-by-side test on that. It just seems obvious because of how hissy the optical audio reader is.

Why not do your own side-by-side test sometime instead of accusing me of spreading misinformation?

Most of the FT_Depot Youtube videos have the audio straight off their ScanStation with no cleanup, people can judge for themselves if the audio sounds right to them or not.

Link to comment
Share on other sites

  • Premium Member
9 hours ago, Perry Paolantonio said:

Magnatech doesn't use capstans. They're sprocket-drive, with stepper motors.

They have double flywheels, that's what I was referring to, that's what helps smooth out the playback. 

Link to comment
Share on other sites

I’ve been happy with the sound from my HDS. I try to adjust the azimuth of the sound head to match the original recording head, just like when I digitise magnetic audio tape. 
 

the HDS does have a flywheel behind the capstan, it’s about 5 inches in diameter. I’d guess the audio quality could be improved by adding dancer arms. I’m no expert, but I would assume the dancer arms would take out inconsistencies with reel tension etc to reduce speed changes affecting sound. But who knows! 

Link to comment
Share on other sites

  • Premium Member
6 hours ago, Andrew Wise said:

I’ve been happy with the sound from my HDS. I try to adjust the azimuth of the sound head to match the original recording head, just like when I digitise magnetic audio tape. 

The quality of the sound head is fine, it's the inconsistencies of the film being pulled through it, which is the problem. 

It can be done as many other scanners use a very similar design. I have not really investigated what's wrong, but it's pretty serious, not something I feel confident in giving clients as a high quality final sound. Plus, it does not stay in sync from the first frame to the last frame, if you capture in DPX and capture the audio on another pass. 

Link to comment
Share on other sites

17 hours ago, Dan Baxter said:

The Lasergraphics is very hissy

The Lasergraphics is not "hissy."

Your implication with this statement is that it is the scanner that is introducing noise that wasn't there, and this is incorrect.

What you are hearing is an accurate representation of the soundtrack as it exists on the film - that white noise is there. It is a part of the soundtrack that has always been there. In an analog playback system (say, theatrical projection), that would get cut out in the sound reproduction path so you wouldn't hear it. But you'd also be losing some information in that process. 

And as audio isn't my area of expertise and I don't want to be speaking about things I don't know about, I did a test this morning: I captured the 16mm mag mix, the 16mm optical neg and 16mm Print of a short film I made 30 years ago. I sent the files to a friend of mine who is an audio mastering engineer. Here's what we found:

DD_Mag-OTN-Print.thumb.jpg.0f18ceb557460fd19a1dd94cd266fad5.jpg

On the left you see the mag. This is the mix, which is obviously the cleanest version with the most dynamic range. This mag is a dub of the same mix as the element that was sent to the lab that made the OTN (which in this case was a Nagra tape), so the sound hasn't been compressed into the frequency range required for 16mm audio (about 100Hz-6kHz). That was done by the lab when the Nagra master was played into the optical track recorder. The Track neg and the Print are in the middle and right, respectively. 

Horizontal axis is time. Vertical axis is frequency. The audio is the same ~2 minutes from the beginning of the film. All three were captured at 24bit/96kHz to WAV files. The brightness of the color indicates the level, and as you can see the optical tracks have brighter backgrounds, which we perceive as hiss. the line at about 5kHz in the optical tracks is an artifact of the Nagra tape (the pilot tone that's reference below). Here's their explanation:

Quote

 

White noise has equal energy per linear band, pink noise has equal energy per log band. Since we perceive frequency on a log scale, pink noise sounds flat and white noise sounds bright, ie "hissy." In the optical path, there would be a low-pass filter that would remove the pilot tone and everything above it. So the noise above 5k would go away, but the stuff below would remain

In other words, because of the limited frequency range of the optical track, most everything above about 5kHz is noise. And because that's all higher frequency, and because there's little to no signal there, we perceive it as extraneous hiss.

The ScanStation is capturing the sound as it exists on the soundtrack, and not processing it further to remove that. As an archival scanner, this is the correct way to do it, and I wouldn't want it to behave differently. If you don't want the hiss to be there in the captured audio, you have the option of applying noise reduction to the capture. We don't do this, because from an archival perspective it's the wrong way to approach this problem. The right way is to do it post-scan, on a copy. Again, I defer to the audio expert for an explanation:

Quote

As you can see, the mag has signal up to 20k. So if you are working off optical, you are better oversampling, notching out the pilot tone, and denoising the rest. You aren't going to get signal to 20k off it, but you can get another octave of real information pretty easily.

This is trivially easy to do even with free software. Resolve has a pretty decent de-noise filter and it does a nice job most of the time with just the default settings. 

Again, I would ask you to please stop spreading misinformation. The scanner is doing exactly what it's designed to do, and if that isn't what you want, then there is the option of applying noise reduction at the time of scanning.

Though, anyone thinking they can use a modern film scanner to create a final product directly off the machine, is using that machine in the wrong way, 

 

 

  • Like 2
Link to comment
Share on other sites

6 hours ago, Perry Paolantonio said:

What you are hearing is an accurate representation of the soundtrack as it exists on the film - that white noise is there.

Fine Perry if you want to argue I'm going to call out your BS on this. You're a service provider and you're in a position to know better, and telling people that LG audio is "among the best optical soundtrack reproducers currently available" is false, deceptive, and misleading advertising and no different IMO to a company with a Retroscan or a Tobin or a Ventura Images scanner that claims their scans are top-shelf off the most professional scanning systems etc. Do you want to know how many times I've heard that claim in the past 12 months - half of it from the mom-and-pop companies and half of it from their customers that don't know any better that say things like "I don't think there's a problem with the scan I think the problem is with the film"? Too many times, that's how many. And I mean no disrespect to anyone, and especially the end-clients that just don't realise how their film should look or what the scan is missing. I might add here that MOST of the time someone makes a claim like that they're literally comparing downstream not upstream - i.e. comparing against a Wolverine rather than against a ScanStation.

The professionals that I know transfer optical audio separately off the best machine they have available. In the past the go-to machine used to be the Sondor and in fact when I first heard about it I just thought/assumed the Sondor was literally an audio machine like a dubber (to be fair, they look like a dubber!) DFT even advertises in the Scanity brochure that they use the Sondor audio components (DFT bought-out Sondor). You will also hear from time-to-time people talking about "Sondor audio units" and what I think they usually mean is old machines that no longer capture video at all that are just set up for professional audio transfer.

That is the point I was making regarding why Filmfabriek doesn't perfect theirs - the R&D to do it and the fact that it's 16mm-only would make it a fool's endeavour when you can literally buy the best 35/16 optical audio machine for ~$35K and half that on the used market (the Cintel that is - a used Sondor would be even less but either option will handle both formats).

11 hours ago, Perry Paolantonio said:

The ScanStation is capturing the sound as it exists on the soundtrack, and not processing it further to remove that. As an archival scanner, this is the correct way to do it, and I wouldn't want it to behave differently.

Really - is that how it works? So what about the base damage - would you prefer that the backlight is not designed to minimise the visibility of the damage in-camera so that you can then try to remove all the scratches after the scan artificially because the scan is preserving the film exactly as it is? You don't get to have it both ways there. I don't pretend to understand exactly how the audio capture is different - but whether the hiss is on the film or not doesn't change the fact that you don't need to pick it up with a well designed audio capture device. My understanding on it (and I may be mistaken) is that just like with the actual scan in the gate you need the film to be perfectly flat to get a perfect audio capture - it's doing the same basic thing that the image-scan does which is run the film under a sensor (imager) with a backlight. If you can get the film perfectly flat you get a perfect capture, and if not you get hiss - again I'm sure that's not the whole story but part of it that contributes towards getting perfect audio.

You've absolutely no evidence backing up your claim that the LG optical audio reader has been engineered perfectly - please do a proper comparison sometime then report back.

11 hours ago, Perry Paolantonio said:

If you don't want the hiss to be there in the captured audio, you have the option of applying noise reduction to the capture.

That's the same thing that people with Retroscans etc say about the picture - whether that's about noise or dynamic range or anything else "you can just fix it up in post - our scanner gets you exactly what's on the film". Again do you want me to tell you how many times I've heard people say in one variety or another "oh it's not worth doing 8mm on something better because you're not going to get any more detail out of the scan anyway"? Honestly it's the same claim except made about audio instead of video - you should not need to do any digital de-noising either way with a good capture.

Link to comment
Share on other sites

17 hours ago, Dan Baxter said:

you should not need to do any digital de-noising either way with a good capture.

..then the noise reduction is applied during capture, maybe you just don't know it. What about clicks and pops, should the scanner also eliminate that?

  • Like 1
Link to comment
Share on other sites

37 minutes ago, Robino Jones said:

..then the noise reduction is applied during capture, maybe you just don't know it.

I do not think so. The audio is not an audio file on the print, it's a printed picture it's all about how you capture the picture and convert it to audio. The hiss is noise, and whether the noise is due to silver particles in the soundtrack or something else I'm not arguing about as I don't know exactly. One thing I can say though from my limited knowledge of electronics is that the power supply itself can introduce noise into an analogue signal, so one of the improvements you can do to projectors is change their power supply to something better. Some of my friends at the moment are working on getting perfect audio out of their projectors. Whatever the source of the hiss is though there's a science behind capturing the audio perfectly and it certainly not as simple as capturing hissy audio and then removing the hiss afterwards.

If it was doing noise reduction in the scan it would sound deformed. Try this - take some audio of a ScanStation and run it through free software like Audacity and see how well it cleans up. It'll sound deformed. Two of my mates bought their ScanStations last year, one has experience with every version going back to 2013, and so for my other mate I confirmed with him whether it's worth buying the Optical/Keykode reader right away or not. He said you don't need it, and he's right the current ScanStations come with perfectly good software sound extraction that works better than the expensive Optical reader which you would only really want for Keykode (which you can always buy later if you land a job that requires one). I may sound like I'm bashing the scanner, but I'm really not - it's a great design overall you can bypass the useless P/T rollers, they've put the capstan on the take-up side on the new ones which probably makes threading easier compared with having it next to the sound modules, and it's got a ton of good software features that are all free like in-scan stabilisation, optical audio extraction, and failed splice recovery.

Link to comment
Share on other sites

On 3/12/2022 at 2:55 AM, Dan Baxter said:

You're a service provider and you're in a position to know better, and telling people that LG audio is "among the best optical soundtrack reproducers currently available" is false, deceptive, and misleading advertising

Honestly, Dan.This is the second time you've accused me of something like this (once in a private message on another forum) and it's getting really old, bordering on libel. I use this machine day in and day out. We've run well north of 3 million feet of film through our ScanStation. And you're making claims about things that seem to be primarily based on third-party information rather than first hand experience. I think I'm in a better position to speak about the quality of the audio on this scanner than you. 

The bottom line is: If you're using an archival film scanner like the Lasergraphics machines, as much as is possible, you don't want the scanner affecting the image or the sound when capturing.

I'm not going to argue that the audio from a Sondor is likely to sound better directly off the machine . Of course it is, because it's filtering out that hiss on the film. I do not know what methods it uses, but it's likely (at least in part) a low pass filter, which means it's cutting off any frequencies above a certain value. And that's a very effective way of dealing with the hiss. But it's also a very effective way to remove any signal that might be above that cutoff value as well. If you capture it all, you can fine-tune later to recover that signal. 

What I have showed you in the image above is the sound captured from a 16mm B/W film that I happened to have the elements for, so we could quickly take a look at what's happening. You can clearly see in the two optical tracks that there is signal above 6kHz. It's not much because it's buried in the hiss. But it's there. A low-pass filter will cut that off and it will be gone. But capturing it gives you the chance to recover that later using other tools.

How is that conceptually different a flat scan of the pix with no grading? The whole idea there is to not affect the image when scanning, such that there's as much to work with as possible after you scan. 

On 3/12/2022 at 2:55 AM, Dan Baxter said:

would you prefer that the backlight is not designed to minimise the visibility of the damage in-camera so that you can then try to remove all the scratches after the scan artificially because the scan is preserving the film exactly as it is

This argument about the lighting on the picture side is a red herring and I'm not going to get into that.

On 3/12/2022 at 2:55 AM, Dan Baxter said:

You've absolutely no evidence backing up your claim that the LG optical audio reader has been engineered perfectly - please do a proper comparison sometime then report back.

Nor have I said this is the case, so I'm not really sure what you're getting at here. 

On 3/12/2022 at 9:31 PM, Dan Baxter said:

The audio is not an audio file on the print, it's a printed picture it's all about how you capture the picture and convert it to audio.

It's more about what you do with it (via filtering, etc) than the method of converting the image of sound to an actual sound. Applying noise reduction or a low pass filter during the capture is permanent, and you will never be able to get at the signal that may have been captured, but was filtered out. Once you've done that, it's gone forever. It's fine to run those filters, but after you've captured the sound, where you can see the effect it's having and maximize the signal that's there. 

On 3/12/2022 at 9:31 PM, Dan Baxter said:

Whatever the source of the hiss is though there's a science behind capturing the audio perfectly and it certainly not as simple as capturing hissy audio and then removing the hiss afterwards.

A projector is a different thing, with a completely different purpose and a totally different method of capturing the sound. When you project, you are watching the film and hearing the sound immediately. You make a compromise  and apply a low pass filter or EQ the audio to get good sound because there is no time to process it more than that. A low pass filter can be implemented in hardware with basically no latency. It's also a 100% analog signal path so yes things like hum from the power supply can manifest in the sound. This isn't a problem with a digital capture of the track, though, so that's kind of irrelevant. 

But you don't use a scanner to project film, you use a scanner to capture the picture and sound, with the intention of doing more work on it afterwards, in the digital realm. 

And even so, as I've said above, if what you're looking for is an immediately viewable file from a scanner, you have the option to apply a grade while capturing, and to capture the audio with noise reduction. But we wouldn't do that for an archival scan because it's eliminating data during the capture that can never be recovered. And the scanner you're saying produces terrible audio is an archival scanner, doing what it's supposed to do.

It's not a telecine. if you want a graded scan with filtered audio, use a telecine. 

Link to comment
Share on other sites

18 minutes ago, Perry Paolantonio said:

Honestly, Dan.This is the second time you've accused me of something like this (once in a private message on another forum) and it's getting really old, bordering on libel. I use this machine day in and day out. We've run well north of 3 million feet of film through our ScanStation. And you're making claims about things that seem to be primarily based on third-party information rather than first hand experience. I think I'm in a better position to speak about the quality of the audio on this scanner than you.

Fine.

Me: "What do you do to clean up the hiss of the scanstation audio (aside from use the cintel)?"

Professional: "I use Izotope Rx if I have to clean up, but really we just use the Sondor or the Cintel to get better audio in the first place."

That is quoted straight out of my email inbox word-for-word.

You can message me in private if you really need me to tell you who said that. But what I would point out to you is that you are literally the only person claiming that the ScanStation audio reader is perfect - no one else is claiming that. No one.

I would really like to know what you're comparing against when you claim the Lasergraphics audio reader is "very good"?

35 minutes ago, Perry Paolantonio said:

What I have showed you in the image above is the sound captured from a 16mm B/W film that I happened to have the elements for, so we could quickly take a look at what's happening. You can clearly see in the two optical tracks that there is signal above 6kHz. It's not much because it's buried in the hiss. But it's there. A low-pass filter will cut that off and it will be gone. But capturing it gives you the chance to recover that later using other tools.

If you want to actually identify where the hiss is coming from you need to scan the same reel of film twice on the same settings and then check if the audio is bit-perfect and whether there's deviation between each capture (and then whether the hiss/noise sounds different on each capture or not).

Link to comment
Share on other sites

8 minutes ago, Dan Baxter said:

you are literally the only person claiming that the ScanStation audio reader is perfect - no one else is claiming that. No one.

No i'm not. Because I never said that. You keep putting words in my mouth in order to further your argument, as you have done on other forums. Please, show me where I said that. 

You are doing what you always do - deflecting and ignoring facts. One last time, for the record: I would rather have a scan that captures everything than one that removes stuff while capturing. Scanning the sound a second time on another machine takes a lot longer then applying a filter and rendering out the sound. And with the method we're using, you have the original scan to go back to if you need it for fine tuning. That is a fundamental difference between archival scanning and scanning for quick release of a product on youtube 

 

  • Upvote 2
Link to comment
Share on other sites

5 minutes ago, Perry Paolantonio said:

One last time, for the record: I would rather have a scan that captures everything than one that removes stuff while capturing.

And what about the backlight then reducing the visible base-damage in-camera would you prefer it didn't do that?

Link to comment
Share on other sites

1 minute ago, Dan Baxter said:

And what about the backlight then reducing the visible base-damage in-camera would you prefer it didn't do that?

 

1 hour ago, Perry Paolantonio said:

This argument about the lighting on the picture side is a red herring and I'm not going to get into that.

 

Edited by Perry Paolantonio
Link to comment
Share on other sites

Oh okay so that's a red-herring.

Yet on Audio I think we both agree we're not audio engineers, and yet you're the one throwing out a bunch of assumptions about how the better audio capture works and why you wouldn't want it. "It must be doing this, it must be doing that", "digital manipulation in post would do better", "the Lasergraphics is capturing exactly what's on the film".

Link to comment
Share on other sites

 

4 minutes ago, Dan Baxter said:

Oh okay so that's a red-herring.

it is.

 

4 minutes ago, Dan Baxter said:

Yet on Audio I think we both agree we're not audio engineers, and yet you're the one throwing out a bunch of assumptions about how the better audio capture works and why you wouldn't want it.

Yes, we agree that we are not audio engineers. Which Is why I consulted one before posting. 

The *only* assumption I've made is that a sound reproducer that is outputting audio without the hiss is likely doing some level of filtering after the image of the soundtrack is picked up, probably a low-pass filter, but possibly more. This is a reasonable assumption. 

The hiss is the grain of the film, which manifests as white noise when quantized. The way to get rid of it is to filter it. Some machines do this while capturing. Some machines don't. I'd prefer the former as much as possible, so that we can control it ourselves. To me, that is certainly the better option. Just as capturing a flat scan without grading is a better option. We don't want to bake in any assumptions to the capture.

If you would like to keep ignoring what I've been saying, that's fine. But just to reiterate one more time, we would rather capture everything and then tease out the details, then cut off those details when capturing in order to get something that sounds better immediately. That's not a good trade-off, if you ask me. 

 

Link to comment
Share on other sites

On 3/15/2022 at 2:40 AM, Perry Paolantonio said:

This is a reasonable assumption. 

And it's reasonable you should do some proper tests.

On 3/15/2022 at 2:40 AM, Perry Paolantonio said:

Yes, we agree that we are not audio engineers. Which Is why I consulted one before posting. 

As did I.

The audio reader does Keykode as well so it's not completely worthless don't get me wrong there. I only just found out that Blackmagic's one is optical + keykode + mag (16mm) and that actually makes it incredible value for what it costs.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...