Perry Paolantonio

I saw the parts list for a Universal build. Someone posted it on facebook. the machine consists mostly of screws and springs and off the shelf bits. Inexplicably, several copper roofing nails too. The bankruptcy filing lists IP and patents as having a value of about $100. I don't think there's a single thing in this machine that's especially innovative or special, or patentable.

It's more than a "piece of metal," it's a precision machined part consisting of several pieces. You get the gate, which is a metal frame with mirrors inside to reflect the light, its polished skid plate, and a pressure-plate that closes down on top of the film. There's also a set of flat rollers that help to flatten the film a bit before it gets to the gate. It's expensive yes, but people with them have reported good results. I've been looking at modding ours to take a pressure plate like the one I built for our 70mm scanner but haven't really had time yet. These are some photos I took of it at NAB last year.

We have a Lipsner Excel 1100 - non immersion cleaner that uses isopropyl alcohol. Works well. it's not as good as an ultrasonic cleaner for really caked on gunk but it gets most stuff off the film in one pass. Two for really bad film usually does the trick. It's cheap to run and doesn't use any nasty solvents. We just vent it to the outside through a standard clothes dryer vent cap. But we only use that for 16mm and 35mm that's in decent shape. The rest we do by hand (8mm/S8 and the more delicate 16mm and 35mm)

I believe it's in the $50-$60k range. it's a nice machine. those rollers were also used on the Lipsner-Smith machines and yes, they're basically paint rollers. The reason this machine has such a complex threading system is that it's not full immersion, so there needs to be enough exposure to air to allow the film to dry before it hits the takeup roller. On the Lipsner machines the film went through a heated air knife to dry the film.

it takes about 10 seconds between reels to clean the dust off the roller with some packing tape. It takes about 5 minutes to wash them, which we do periodically, but not after every job. PTRs are not silicone. And they're not rubber. They're made of urethane. You can't make a blanket statement about them compared to phone cases or rollers in other environments. Different materials. And yes, alcohol will ruin rubber over time too.

you can buy them from Kodak. this is not an issue.

It's multi-spectral in the sense that they're using many narrow-band light sources to compose the image. think 16-18 exposures per frame, each with light with a different spectral range. I'll be honest, the credibility of the person behind this is shot as far as I'm concerned, since she was the one who was responsible for that ridiculous scanner comparison paper from a few years ago that was riddled with basic factual and methodological errors. This whole Multispectral scanning thing sound like a solution to a problem that doesn't exist, if you ask me. it's aimed at archivists, not at people who need to work with the film in a day-to-day way. The theory is that it's a better representation of the color on the film. But when you think a scanner can't scan a specific hand-tinted print color because you set up the scanner incorrectly, and you "prove" this by taking a picture of the film with your cell phone and you see the color, but you assume that's because the $500 phone has a better imager than the $500,000 scanner, then you start a company to solve this "problem," you're clearly not operating on the same wavelength as the rest of us. pun intended.

Blu-rays are made from whatever sources are available. If there's no limit to the budget, and original elements (neg) are available, that's what gets scanned, graded, restored, and put on the disc. Smaller labels may only have what is already transferred, so that's what they use. Print is almost never a first choice though sometimes that's all that's left. Film prints were never meant to be scanned or viewed digitally, they were engineered to be seen in a darkened room, and designed to take advantage of the way the human vision system adjusts to that darkness. Digital sensors are not eyes, so scans of prints never look as good as scans of camera original material or in some cases, earlier generation intermediates.

That machine has been on ebay for years. FWIW, we sold our complete and fully functional Northlight 1 about a year and a half ago for $5000. Sadly the freight company dropped and destroyed it at the destination. It's a good scanner, not worth more than about $5k these days. Scan speed is measured in Seconds per Frame, not frames per second. Even a Northlight II, which is faster, is painfully slow. Like 12-18 hours to scan a single reel. They also require a 240V circuit, and they generate a ton of heat so you have to plan on ventilation to the outside to suck that all away from the machine, or it'll run even slower until it stops running entirely. The sensors are also prone to collecting dust, which shows up as streaks on the film so you have to be vigilant about cleaning all that with every reel change. And they don't like splices at all. We had many overnight scans fail partway because it was being fussy about a tape splice. My understanding is that Filmlight no longer sells these. The old Northlight product pages are on the site but they're no longer listed under Products. You have to google to find those pages.

This is a terrible idea. Alcohol will destroy the roller. And don't soak them, just wash them. Use lukewarm water and a gentle dish soap like Dawn. well if you clean them with alcohol they won't last. We still have the original pair that came on our ScanStation and they're in perfectly good condition. When the rollers on the scanner get dirty we rotate in clean ones and then wash the dirty ones. Once dry, they're good to go again. In the mean time you can use a little packing tape to remove dust. We rotate through a half dozen or so of these and millions of feet of film have passed through our scanner. PTRs that aren't kept in the right climate will eventually break down. We found some of the small 1.5" rollers in a box when we moved. They came in an auction lot and we had no use for them. When we were packing things we discovered that they had turned to liquid goop. Probably they were stored near too much heat before we got them, or were really, really old. Fully loaded, it has been around that price point since the 5k model came out, that's nothing new.

AEO Light is only as good as the resolution and sharpness of the scan because it's taking the sound from that image. If you're scanning on a low quality machine at low resolution you won't get nearly the same quality you'll get from a 4k or larger scan. The vast majority of Super 8 scans we do are at 4k. Nobody really scans Super 8 prints at all except some museums who have very specific work (usually art films). Super 8 prints were about the home entertainment market mostly: condensed versions of hollywood films, or maybe a reel you could buy at some major attraction to cut into your own home movies. But we scan 16mm prints at 4k all the time - It's rare these days that we do anything at lower resolutions, to be honest. Just doesn't make any sense to scan at 2k anymore. BMD uses an old-school photosensor and a light source, like in a projector. The quality is ...not good. It's has a very high noise floor, and the high end is cut with a low cut filter. Also, you have to manually adjust (with a knob) the position of the pickup, so if you don't have it perfectly aligned it's not going to sound good. AEO Light on a sharp, high res scan will give you much better results than their optical track reader. Steadiness is not a hallmark of the BMD Cintel. We've re-scanned a lot of stuff done on these and they bob and weave like crazy. The steadiness of the lasergraphics scanners is dead on, and so is the Xena, from what I've seen.

you need a new computer. Phoenix is anything but slow. Some of the effects are single-threaded, which are slow and are obviously clock speed-dependent. But it's not a slow application by any means. Supermicro M12SWA-TF Motherboard Threadripper Pro 5975WX 32-core 3.6GHz 128GB RAM GeForce RTX4070 Windows 11 we have a very big spinning-drive RAID 0 in the machine, made with old disks we had kicking around. It's just for caching so when a drive dies, which happens, no big deal. Source files are on our SAN. We cache to DPX files and the RAID can handle 30fps 4k playback no problem from DPX so it's not a bottleneck. Most 4k effects run slightly slower than realtime, which is faster than any restoration system I've ever used. I think some effects are still single threaded, so those go slower. Resolve's restoration tools are skeletal at best and aren't especially sophisticated about how they work.

Not sure what you're talking about here. Phoenix film frame stabilization is effectively instant. It may not use the GPU but it is very good. Not the old stabilization plugin, but Frame Lock. A 1 minute clip renders in about 10 seconds on our machine. They built it to compensate for the crappy stabilization in the GoldenEYE, I think. We have some film here right now from a client and I don't know what they scanned this on but it's a disaster in most respects. the frame floats all over the place but frame lock will completely fix that. The user interface is the same. The feature set is different. Nucoda is geared toward grading and uses the GPU more. Phoenix towards restoration and is primarily CPU based. They look the same because the underlying architecture is the same, it's just about which plugins you have, pretty much.

You will get better results with the software version, probably, for the reasons I outlined above. It's outside the scope of the article, because you shouldn't be working at those speeds, but sure. I can do that just to see. The picture quality at 60fps is not as good as the picture quality at lower speeds and if you're scanning film at that speed expecting good picture quality then you're using the machine wrong. And the sound will not be great at 60 fps from the hardware reader. Both of these caveats, Lasergraphics makes very clear up front. That being said, 60fps only works with lower resolution scans (2.5k or lower), and that means you're starting with way fewer pixels in the scanned image of the soundtrack. Half as many, in fact, compared to a 4k scan of the same film. That should have a negative effect on the sound quality. FWIW, High speed capture should only be done in specific, limited circumstances: We only do 60fps capture when we're making low resolution access copies (and only that) of a large collection of film. Or if we're making a reference scan of a print that we're conforming high res A/B roll scans to - in that case the scan of the print is for our internal visual reference only, not a client deliverable. We don't capture customer work at those speeds because it doesn't look as good.

Of course they're all different. Each company is going to process the sound in a different way depending on how they implement it. My point in specifying that it's a line array camera was to differentiate from the main sensor in the imaging camera (used for software track extraction as well), and from the more traditional optical photosensor track readers like the one in the BMD Cintel. As for curling, that's a red herring - that's going to be an issue no matter what kind of soundtrack reader you're using. It is possible to enable the software reader on our system. It requires physically disconnecting the hardware optical track reader. The scanner's software only allows one type of soundtrack reader to be active at any given time. You cannot soft-switch between them. We don't usually muck around with the insides of the scanner unless necessary, but I will be doing this when I do some test scans for the blog post I mentioned.

I can't find the post now, but I did post something here a year or two ago about this, and I hope to get a version of it onto our site soon to try to put to bed this nonsense you keep posting. But the gist of it is this: The hardware optical audio module for the ScanStation is a line array camera (not the same camera that takes the image of the picture). It images the optical track as the film passes its gate. Each line (row of pixels) of that image corresponds to an audio sample. The camera takes about 80k lines per second. The slower you run the film past it, the better the sound quality will be, because the audio sampling rate is higher. Any slower than 24fps and the differences are largely academic. Lasergraphics themselves will tell you that scanning the film too fast will result in lower quality sound and they recommend scanning at 24 or slower for best results. And this is easy to test - scan the same film twice, once at 60fps and once at 12fps, then load the files into an audio application that shows a spectrograph. You'll see a lot of noise in the faster scan but not in the slower one. But let's say you ran the film through at slower than 24fps and it's still hissy. That's because you don't have the (unfortunately named) noise reduction feature turned on. I say unfortunate, because in most of the audio world, noise reduction is a step done *after* capturing the sound and is inherently destructive. In the case of the hardware optical reader in the ScanStation, it's done before the audio sample is even an audio sample. It works like this: The optical track camera takes an image of the track. It then looks at that image and eliminates the random noise of the film grain as well as transient gunk like dirt. The waveforms of the audio are smoothed, and the end result is a virtually noise-free track. But then you might say "but that's altering the sound!" -- it's not. If you capture the same audio with and without the noise reduction, look at the spectrographs of the two and overlay them, you'll see that the *only* difference between the two is that the noise reduced version has no noise. The actual soundtrack is unaffected by this step. You remember incorrectly - I never said that. The ScanStation has two options: the far superior hardware track reader, or the ability to read the track from the scanned image. It does not offer both at the same time, it's one or the other. The SSP was the first to offer the software track decoding feature because they didn't have a hardware optical reader option for that (one or two of the very first SSPs did, I believe, but it was removed from the feature set of that model). The Archivist has the software option as well. The difference between the two is that the resolution you scan the picture at, as well as the file format you scan to, can affect the optical track reading. If you're capturing to low bit depth files, or at lower resolutions, like 2k or HD, you're going to get fewer audio samples, because your starting point has fewer lines than the hardware reader does. If you're capturing to a compressed file format, you're potentially introducing compression artifacts to the image of the track that could affect the sound quality. Probably not an issue with ProRes 4444, definitely an issue with ProRes 422 (non-HQ), or ProRes LT/Proxy, etc, where there's more compression. The purpose of SoundView is to handle certain edge cases where the track is misaligned on the film. Sometimes when the track was printed to the film, it was slightly off. The hardware track reader does a good job of detecting and automatically centering the track but there are some cases where the track is misaligned enough that you need to manually set some boundaries. This is so the track reader is only picking up the track itself, and not noise outside the track area. This feature is for the hardware track reader. I don't think it's available for the software reader (might be, but I can't test that as you can't have both running at the same time on the same machine). Perhaps @Robert Houllahan can test it on one of his machines with the latest version of the software and let us know. Well that's certainly a throwback. I knew they had some hardware ports but didn't realize they made it work. I guess it makes sense since the Cintel engineers still treat that machine like a telecine in many respects. Nobody really works like that anymore though.

The standard gate looks just like the photo above, minus the hinged pressure plate cover. The left and right sides of the gate assembly are curved but the film sits flat in the gate itself.

This is objectively false. Seriously, why do you keep posting this? What scanner does this? Telecines did this, but there hasn't been a need for it for years. Any late model dubber uses stepper motor control and runs at a fixed speed. xtal clocks govern the speed, so if you scan to a 24fps file and capture your mag at 24fps, they will sync perfectly. There will be no drift unless there's a problem with the machines, and no interlock is required.

It is a hinged, spring-loaded pressure plate that closed down around the film. Additionally there are some rollers you replace on the machine that don't have the V-tracks like the standard rollers. These flatten the film a bit as it goes through the gate.

Hi Andreas, Sent you a DM. But to answer some of your questions: I don't think you can overscan enough to get the soundtrack without modifications to the gate and the pressure plate units. They'd need to be enlarged in order to do that. If you do this, I would also lengthen the gate openings further, so that you can run VistaVision on it - this is just a configuration file change to support how far the projector moves. I did some tests with it using the 35mm 4p gate and it seemed like it would work, but the gate itself was too small, so it picked up black on the sides. We never did the gate mods to support the wider 8-perf frame size The main thing is that with the physical gate doesn't allow the sensor to see beyond the current frame, and for AEO-Light to work, you need some overlap with the surrounding frames otherwise you have gaps. Regarding the framing question, look at the config files in the packet I sent you. specifically the northlight.cfg file, which is in the Software Installers folder: Software Installers - v1.1.2711 Linux/fl/northlight-1.1.1297/etc/northlight.cfg -- those are how they were when we got the scanner, and they're a little bit of a mess. Definitely save a copy of your existing config file first, just in case, but it might help to figure out settings that might work. Down near the bottom are the settings for the different gauges. If I recall correctly, the "format" section defines the area scanned (how much the projector moves). And the crop settings define the output. It may be that your crop size or positioning is not correct, and that's cutting off part of the image. my guess is that things like the offsets are specific to each machine, so it's possible you have a config file that came from another machine, or maybe is a default value, and that you'd need to play with to get things lined up correctly.

Please don't bake your full coat mag stock. Most mag that would require a treatment for sticky shed syndrome has an acetate base. (Sticky shed is the problem that baking addresses with audio and video tape) . Baking acetate based stock will destroy it. Acetate shrinks with heat. Please don't bake mag. That said, baking might work for newer polyester based stock. ...maybe. But poly based mag was introduced much later and is usually in much better shape than acetate to begin with, so it's less likely to need much treatment at this point. There is some very old poly stock (i think it was introduced in the early 70's, but I'm not positive about that). We've worked on films from the mid-80s that still had acetate mag stock. And the widespread use of fullcoat tailed off not that long after, in the 90s, when digital audio became the new standard.

You're a real piece of work, man. At the time, confirmed with both Lasergraphics and Galileo digital, the archivist did not offer HDR as an option. The very first ones did, as I said. Then they didn't. Now it is apparently an option again. What is there to concede? I stand by what I said in your quote of my post. BTW, Galileo Digital is the sole worldwide reseller, not just the US. Resellers in other parts of the world work through Galileo. Putting you back on Ignore now. I don't know why the cinematography.com decided it needed to email about this post in the first place.

I think the problem is likely that Yankees hat. But seriously - the overall image (at least on my iMac, not looking at it with scopes) feels a bit too green. I'm not talking about the grass, which does feel a bit oversaturated, but there is a very slight green cast, which can really make skin tones look funky. The highlights look ok, so in a lift/gamma/gain window in Resolve, I'd pull the gamma a bit away from the green to see if that helps.

Any modern scanner can create a 16fps file from your film, so just scan it at 16fps. Most edit systems will allow you to work at that frame rate as well and export to 16fps. Then when you're all done, if you need to deliver something at a more standard frame rate, you can pull up to that frame rate. If you're only going to upload the file. just upload the 16fps, because all the standard web-based video sharing sites can handle that.

DPX is almost never necessary and it's a huge pain to work with - both in terms of the storage capacity required and the speed of that storage just to play the files. Most VFX work, if using image sequences, is done with EXR, which you could easily make from the ProRes files. ProRes 4444 is a good format that is essentially uncompressed in terms of visual quality. The benefits far outweigh the theoretical downsides. See: https://gammaraydigital.com/blog/prores-or-how-we-learned-stop-worrying-and-love-compression As for the scanners, they're different beasts, and they work differently so it's hard to make an apples to apples comparison. The ScanStation uses a bayer sensor and runs faster as a result. The Arriscan is a sequential RGB scan, so no bayer loss, but it works at a much slower speed. That being said, in HDR mode, the ScanStation 6.5k's imager comes very close to giving you the same kind of color as you get from a sequential RGB scan. We've had some clients who did side by side tests with a Lasergraphics Director (very similar to the Arriscan in terms of how it makes images), and a ScanStation 6.5 in HDR mode, with an oversampled image (scanning the frame at 5k but outputting to 4k). The results were *really* similar. Close enough that they opted for the less expensive ScanStation scans and were very happy with them.