Can anyone explain what’s causing this slight crackling sound in the audio on some scans? Link to example is below.
It’s an optical track? They pop and will never sound perfect.
Does the scanner you’re using have a dedicated optical track reader or is it doing the software track decode? If software, then the resolution you scan the picture at will come into play. Lower resolution picture means lower resolution scan of the optical track, which means lower quality sound (effectively lower sample rate).
If the scanner has the dedicated optical track reader, then you want to scan at 24fps or slower to maximize the resolution on the track image, and you want to use the “Optical Enhanced” track option (called something like that, I forget the exact wording). This removes the film grain and cleans up the waveform before the waveform is converted to audio. This gets rid of a lot of the high frequency noise in an optical track, which comes from the randomness of the film grain.
Software reads the optical track. I scan at the highest possible resolution with the Archivist. I did use the Optical Enhanced feature before, but I got told by someone to not use that because it adds unnecessary noise reduction
I don’t know how the optical enhanced feature works on a software-only scan, if it’s different from the hardware sound reader.
I can say that if you did it on the dedicated optical module, then that person is completely wrong. It’s not noise reduction in the traditional sense (where you’re removing noise from the captured audio signal). It’s noise reduction in the sense that it eliminates the stuff that causes the noise before the sound is interpreted from that image. It doesn’t affect the signal at all, in fact. I’ve had a blog post for our web site in the works for a while and need to get on that because what Lasergraphics is doing is pretty unique and really works well.
Part of the problem is that Lasergraphics is awful at naming stuff. It used to be called “Noise Reduced” before they renamed it “Enhanced” – neither of which inspires confidence. But it used to be that the “noise reduced” was optional and now that its “enhanced” it’s the default setting because it’s the better sounding option.
I’ve gone back and forth with them over how it works and we did extensive testing on our setup to confirm everything they’ve said, including capturing the sound on other machines. It’s pretty amazing what kind of quality you can squeeze out of a 16mm track - there’s a lot more dynamic range than people think, because traditional sound reproduction hardware low-pass filters out the high frequencies, to mask the noise. That takes a lot of signal with it, sound that most people don’t even know is there.
1 Like
No optical module on the Archivist I’m using, only software. It does a great job.
Concerning that “popping” in the optical track, I’ve noticed that it’s very intermittent. Some prints I’ve scanned don’t have any of that popping at all, so it may just be how the audio was printed. I set the audio level to -10 dB and I’ll occasionally still hear those sounds on some prints, though not during the entire print, just here and there, usually during loud sections, such as music. My concern was that the popping is actually audio clipping, but it seems that may not be the case if some prints don’t pop at all
Popping is generally dirt or scratches or some other physical issue on the film. Do you have the full overscan image with the soundtrack? If so then you can look back the right number of frames from the frame where you hear the pop and you should see it visually. For 16mm, it’s 26 frames offset from picture. I would start there to see if there’s an issue with the track that might explain what you’re hearing.
I believe your initial post about the intermittent popping being an issue with the optical track was correct. I scanned 3 prints today and all 3 do not have that, even during loud scenes. I looked at the soundtracks where the popping was present on some of the prints and I couldn’t see any noticeable issues. I’m glad to know it’s not a software issue
You can crop and resize at the time of scanning - 5.3K is completely uncropped and you’re wasting a lot of storage. Unless the archive is insisting that it be uncropped you can crop as close to the perfs as you want to. Most operators prefer to leave a modest amount of overscan because there’s a risk that a splice will jump the frame and with tight overscan that could result in needing to rescan the reel (or at least the part after the jump happened). If it doesn’t matter to you that you need to rescan on the odd occasion of a jumped frame then just do a tight crop in the scan itself.
This is probably not a reasonable suggestion for a professional environment, but I do recompress my raw files with a software called slimRAW. There’s a demo you can use to test the different conversion/compression results.
On the other hand, ~10MB(?) per single frame already sounds rather efficient at that resolution, so there’s probably not a lot to gain.
1 Like
ProRes is not RAW (well, ProRes RAW is, but that’s not what’s being made here). So there’s no real benefit to recompressing it. It’s already a highly efficient visually lossless format. We have been slowly convincing many of our archive customers to scan directly to ProRes 4444 because it’s that good, and so much easier to manage long term than DPX or similar uncompressed files.
Scanners like the ScanStation are always doing some amount of image processing before the file is written - they have to, in order to do the cropping, perf detection, and any number of other things. So most do not output a true RAW format in the way a DIY system might.
2 Likes
Hi Perry, quick questions…
Any used commercial scanner or file format that provides sensel-data (sensor values), particularly for bayered sensors?
For example, the raw image in a DNG it is not the same as sensel-data (offet was already removed).
I ask because there are some interesting presentations on the subject of raw data being cooked.
It is actually what I am doing with my processing pipeline for full resolution, more out of hardware necessity than for trying to be an archival purist 
.
Another question… how widely used is OpenEXR in the commercial scanning world?
Thanks!
The only commercial scanner I know of that use(d?) RAW camera data is the Kinetta. Certainly that’s how the early versions worked. The problem with this is that it means you really can’t do any processing in-scanner, as I understand it. The early Kinetta versions didn’t do any frame registration, so you had to repeatedly scan problematic sections, sometimes forwards and backwards, then debayer, piece it all together later, and run a stabilization pass on it in post. I haven’t worked with footage from one in years, but I know that Tommy from Colorlab did some work on the software that controls it in recent years, and I think that stabilization and other features that are more common on more well known machines are now there, which would suggest that they’re not using RAW in the same way. But this is a guess.
Actually, I take that back - the Cintel used a RAW format that required Resolve to do the debayering, etc. Again, I don’t know how raw that truly is though, since I’ve never used one personally.
I don’t think very. If I ever have any time to get back on Sasquatch, this is what we’re planning to use though. You can store the separate R,G,B channels in the OpenEXR file and then merge them in Resolve, which cuts all the color science work out of the scanner software, something I’m perfectly ok doing. So we would only be doing the basic stuff like making sure it’s registered in the scanner, and then dealing with the rest afterwards.
Too many other projects right now, and I’m currently up to my neck in Umatic decks and learning how to fix and maintain them. But OpenEXR makes a ton of sense as a good bridge format from scanner to Resolve or similar apps.
1 Like
Thanks Perry, great information.
I did some of that about 35+ years ago. It was not easy then; it would probably be a nightmare with 35+ year old electrolytic caps everywhere. Have a colleague that stuck a bit longer with Sony Broadcast, he is based in Davie, Florida, but he is now overseas.
About U-Matic, check out this project… very cool. I made a technical pitch to Memnon US regarding doing digital demodulation for C-Format (around 2014 NAB), and they thought I was on crack :). Glad someone validated the idea and made it work.
Take care.
Yeah I’m familiar with that project. Though we’re using BVU-950 decks, which have pretty great composite video out (significantly better than lower end decks like the VO-5000 series), we also use TBCs with dub connector inputs to get the same effect as this without all the hacking. That method makes much more affordable VO-9850 decks workable too. Hard to find the TBCs though. Someone should figure out how to make that project work as a standalone box without having to solder anything onto the board. There has to be a way.
These machines are built like tanks. The main problems we’ve found haven’t been electronic as much as mechanical - stale grease gumming things up, and rollers that have dried out and need to be replaced. At a recent auction I wound up with half a dozen brand new pinch roller assemblies, which are worth as much as I paid for the 500+ other parts we got. There’s a guy who refurbishes these rollers and he quoted me something like $150 per roller to rebuild, so this was a major score. We’re about to start a project later this month to capture 600+ Umatic tapes, so I’ve been gearing up to handle that.
We’re also doing a bunch of 1” - recently got this beast working: https://youtu.be/-Lrp02Is2hs
I still have my Sony training certificate on the BVH2000! That machine is well built.
1 Like
that was fun to work on. We recently picked up a second one from a retired engineer that he says works. Just wanted it out of his living room, so it was free! I haven’t had a chance to test it yet.
1 Like