Perry Paolantonio

So here's something I haven't seen before: A client sent us some Super 8 Sound film shot in the 1980s, but the audio is consistently about 1 second out of sync throughout the reel. How is this possible? The sound record head was a fixed distance (18 frames) from the gate in all sound cameras, and it's clear that this is sync sound with appropriate ambient noise (not post-dubbed on a projector or editor). The subjects are holding the typical cheap condenser microphone that came with most of these cameras. You can hear them bump the mic against a table about a second before you see them do it. I could see a sloppy manufacturer having the head be maybe a frame or two out of place, but given the way S8 Sound carts were constructed, there's not enough of a window for the record head to be a full 18 frames *more* away from the gate than it already is. Any ideas? I'm stumped. (and just to verify it's not our scanner, we've scanned several other films known to be in perfect sync and they're absolutely fine). -perry

We are doing 5k 16m and Super 16 on our ScanStation and we're 1-day shipping via UPS Ground (cheap) from Cinelab.

Here's an example of a 2k scan made on our scanner in 5k mode. That is, the full 5k sensor is used, to output a 2k file. If you look at the Super2k example, you can see that the grain is much more finely resolved than a straight 2k scan done on the same scanner. I realize you're asking about 4k, but the question is whether the detail is there, and the answer is clearly yes, whether the scanner is scaling down to 2k or to 4k. http://www.gammaraydigital.com/blog/case-super2k Assuming you're using a scanner that has extremely high quality optics in it (like our ScanStation), the main factor in terms of sharpness is probably going to be the lens on your camera. I'd second the suggestion to use a pin-registered camera like an Arri SR, though you could also look at Eclair ACLs or Aatons, both of which produce extremely stable images and are available for reasonable prices these days. Lens options are better on all of these cameras since they have more "professional" mounts available, where the Bolex is just a C-Mount. Nothing wrong with C-mount, but if you were to rent lenses, you'll probably have a wider range to choose from if you've got a variety of mount choices at your disposal. My ACL, for instance, has a C-mount built in, but also has an Eclair mount ring that allows for Eclair's bayonet mount. There were similar adapters for this camera for Nikon and PL mount, I believe. Your choice of stock and the lighting will affect graininess. Generally, overexpose a bit with neg. The scanner is really resolving grain, so as long as you're working with a high quality scanner, the rest is up to you in terms of the lens and lighting... -perry

Just curious - why go to digital betacam these days?

I've got one of the Beaulieu 4008 Pro8mm Max8 modified cameras, but I'm noticing that the battery life is pretty bad - It's supposed to last "50 rolls" but I'm barely getting through one. I think these are NiCads, which sucks, because of memory issues. I'm wondering if anyone has the schematics for these battery packs that I might look at, because I think I want to build my own. The way it's constructed, to take it apart is to basically ruin it (looks like it's basically a cluster of sub-c cells stuffed inside a big heat shrink tube. So there's no easy way to open it without destroying the case. Happy to do that if I need to, but I'd like to look at some pictures or a schematic first, if one is available. My thought is that I might just build my own using better, more modern batteries. Or if a standard belt exists that I could pick up cheaply, I'd go for that too. Thanks! -perry

Actually, with Super 8 there's also the issue of Kodak's sloppy perforations, which cause lateral movement on pin registered scanners or digital stabilization done using the perfs as reference points, if the camera used is either pin-registered or exceptionally stable. That's totally independent of both camera and scanner, and is squarely (heh, get it? square -- like perfs?) on Kodak's shoulders. It's annoying, to say the least.

http://www.revostock.com/SearchResult.html?text=super+8&themediatype=video http://www.shutterstock.com/video/search/?searchterm=super+8&lang=en&language=en&search_source=footage&safesearch=1&version=llv1 http://www.istockphoto.com/videos/super+8#17502aa2 If you search "Super 8" on most stock footage sites, there's stuff there.

Thanks, David. That makes more sense. I hadn't considered the spacing between photosites might be wider, with the sensor physically being the same size as if it had 6k horizontal photosites. That's pretty clever, actually. -perry

Assume the image is projected onto a plane that the sensor is focused on. If the sensor is 3x2k, and the image fills the sensor area, and there's a miro-movement in the sensor position between images that it takes of the film frame (all of which is what I'm getting from this thread), then how is a 6k image a direct sample? if the sensor is 3000x2000 and the output is 6000x4000, then it must be using those multiple exposures from slightly different positions to interpolate up a new image. Again, this would be more accurate than a single 3000x2000 upscale, but it's still scaling, no? Or is the sensor taking an image, moving 3000 to the left and taking another, then 2000 down for a third, then 2000 to the right for the last, then stitching? That would produce a 6000x4000 image, directly sampled. If that's the case, then I'm not understanding the 'half-pixel offset' that David Mullen describes, thus my confusion. -perry

What I'm not really understanding is how it's making a 6k image from a 3k x 2k sensor. From their web site, on the technical specs: The way I interpret this is that the sensor is taking 4 shots of the frame, to result in a 6k x 4k image. There are two ways one could do this, I'd think: 1) Move the sensor down 2k or across 3k, depending on the image being taken, stitch them together. However, based on what you said earlier in the thread about it moving a half pixel, it sounds like: 2) Take multiple images with a micro-shift in the sensor position, and use those to interpolate a 6k image. Now, that's not the same as blowing up a 3k x 2k image to 6k, which would involve making up a lot of image data. A lot of accurate information can be derived by subpixel changes in an image, so you'd probably get a significantly better interpolation to 6k this way than with a single shot at 3k x 2k. But I'm not understanding how this is truly a 6k image if it's doing that. It's still interpolated. Or am I completely misunderstanding what's going on inside the machine? -perry

Ahh - I misunderstood. I was under the impression the Arriscan used a full 6k sensor and downsampled to resolutions like 2k or 4k. Instead it's using a 3k sensor, taking two images to make a 6k composite, then downsampling that? Is there a technical advantage to making a 6k scan this way, or is it a result of the lack of 6k sensors at the time the scanner was designed? on its face, it sounds kind of kludgy. -perry

Now I'm curious about this - what's the reasoning behind the offset?

I agree that the CIRO splicers are easiest for editing. We deal with a lot of archival film scanning, and need to attach head and tail leader as well as performing the occasional broken splice repair. I personally prefer tape splices for these repairs, because you can back out of them by removing the tape (vs a cement splice, which permanently alters the film). In these cases, the Rivas splicers are better because they're a bit more flexible. But again, when editing new film, you can't beat the speed of a CIRO splicer.

Their credit card system is definitely a bit weird. I recently ordered some 7203 and some 7266 Super 8. The charge for the total amount appeared on my card, but because the Tri-X was backordered, the charge disappeared and they charged me again just for the color neg. Then when the Tri-X was ready to ship, they charged for the balance. The total added up to the same amount, but it was confusing to see the charges appearing and disappearing from my account. Kodak, sadly, is totally old school on customer service and ordering, preferring to do everything over the phone - why they haven't set up proper online ordering is beyond me. It would save them money and it would be a hell of a lot more convenient for most people. I would recommend getting the name of a CSR who is helpful, and make sure to ask for them whenever you call so that you're dealing with the same person every time - it'll make resolving issues like this easier. Get their email address, too.

These are good splicers for new film, but I wouldn't use them on older or shrunken material. The tape used by these has no pre-made perfs, so the splicer creates them when you push the lever down - this is perfectly fine on new stock, but on older film it can be problematic. For that reason, we use Rivas splicers for 16mm and 35mm. In these, the film is held down on either side of the splice by only two pins, and one of those pins is spring loaded, so you can put shrunken film in it without fear of damaging the film. The splicing tape is pre-perforated, and if the perfs don't line up perfectly, that's ok for our scanner, but it's easy enough to cut out the excess tape with an Xacto knife if need be. -perry

Hi Will - that's correct. In the ScanStation, the sensor/lens assembly pulls back from the film and refocuses (optically) on the film to fill the frame, if it's in 5k mode. If it's in 2k mode, the assembly is closer to the film and a 2.5k window inside the overall 5k sensor is used to create the image, allowing the scanner to run at a faster speed. Past a certain point (and there's no single answer for where that point is, since it depends on things like the film stock, the lens used, and the exposure), you're not gaining much by going with that high a resolution from such a small frame. But if you need to integrate Super 8 in to a 4k production, for example, it's a better way to do it than to scan at 2k and blow it up in post. Also, you can scan to a 2k file in the ScanStation, with the scanner in 5k mode. It's a lot slower this way, but the results are better: http://www.gammaraydigital.com/blog/case-super2k -perry

Thanks. I think the command line ffmpeg with the default settings out lined here seems to do the right thing, at least with the prores file I tested with this afternoon - reasonably quick, too- about 11fps reading and writing from the same disk. I tried AnotherGUI, but it seems overly complicated for what we need to do - the command line is so simple, that's just easier. Ultimately, I'll probably do what we do for most of our in-house tools and build a simple GUI for it for our own use, but for now, this seems to do the trick. Thanks! -perry

It doesn't come up often, but occasionally we get a ProRes file that we need to convert to DPX. In the case of the current situation, it's a ProRes 422HQ 1080p file, but we need DPX for our restoration system. Normally I'd do this in AfterEffects, but I'm wondering if there's a better/more efficient way to do it, say with ffmpeg? Since we don't have AE installed on the restoration system, doing it on one machine and then moving the files means a lot of waiting for file copies to complete. So I'd like to be able to do it right on the restoration PC to save some time. Googling mostly brings up conversions in the other direction: DPX->ProRes, but that's not what we want here. My main concern is avoiding any color-related issues when using ffmpeg - gamma shifts, color space changes, etc. Any suggestions or recipes? Thanks! -perry

Interesting. I'm wondering if Diamant has improved in terms of the quality of the fixes it does - we found the interface to be a bit clunky when we tested it several years ago, but more importantly, the quality of the fixes it was doing in dustbusting mode didn't hold a candle to the MTI. PFClean is a great application, and is a veritable Swiss Army knife as far as film mastering tools go, but it suffered from the same issues - fixes that looked like fixes, instead of being completely seamless. We spent more time futzing with the cleanups to make them invisible in both applications than we ever have in the MTI software. Algosoft is a really interesting tool, and I think it's promising, but so far our 10 years of digital restoration experience has shown us that *no* automated film restoration tool us up to our standards -- too many false positives and too much work to QC the "fixes" they make. The problem with automated restoration tools is that if you run an auto-pass and then go back through it, you're much more likely to miss a subtle artifact than if you just do all the work by hand. That way every fix requires a conscious effort and it's much harder to end up with artifacts. For us, the best strategy is to approach film restoration 100% manually. It's more expensive but the results are superior and we can be totally confident that it's artifact-free. We just upgraded our MTI to DRS Nova (the newest version), in fact, in anticipation of our next large project - a major 35mm feature from the 70's that's riddled with negative dust from the 4k scan that was done in Italy last year. By the way - good call on the ScanStation. You'll love it - It's an amazing machine. I absolutely adore ours! -perry

Yeah, I've checked out those Kodak mono CCDs - really nice. I'm thinking something like 8k makes sense. I have 4-perf and 8-perf gates for the scanner, so it would be fun to be able to use that vistavision gate. Not that there's a lot of call for 8-perf scanning these days, but still... As for the sensor, I figure I'll cross that bridge when I get to it though - I'm using a library for the camera control software that's made by the manufacturer of the test camera and its proprietary PCIe card - the nice thing is that it's compatible with their generic Cameralink cards as well, so swapping the camera out should be more or less seamless - just an inexpensive CameraLink card and the camera itself. This has been a fun project, but it'd be nice to have some more free time to work on it!

I'm fairly certain that JAI is what's in our 5k ScanStation. The picture quality is outstanding - I'm very impressed with it. For small gauge film, the downsampled 2k output looks really good: http://www.gammaraydigital.com/blog/case-super2k I know that Lasergraphics has a lot of proprietary stuff going on between sensor and disk though, so I'm not sure if you'd get the exact same result in a DIY setup. They're doing some of their own processing of the image to deal with sensor noise, plus each scanner's settings are individually calibrated since there are differences between the camera units. The 35mm Imagica I'm rebuilding has an inexpensive rolling shutter camera in it now for testing (4.5k, though it looks best if you output 2k from it), but that works because the scanner is pin-registered, so it's an intermittent motion. Most likely I'll end up putting a mono sensor in this though, as I'm thinking of designing and building my own RGB+IR LED array. It won't be fast, but it should look great! -perry

I would imagine the best way to do this would be to hand file the bulk of it, always filing away from the film path (toward the camera), so that any burrs are directed away from the film. Then take a Dremel tool to it, with a good set of abrasive brushes and wheels, and polishing tools. That should be able to make it nice and clean and burr-free, though it will require testing on scrap film. Using one of the inexpensive drill-press style mounts they make for these will probably make the job easier.

In most cases, the ProRes 4444 will be sufficient - it's really a pretty great codec. We've found it to be as good as uncompressed, but with smaller file sizes. It's easy to work with and doesn't require super high-end computers to wrangle. If you want a no-compromises format, I'd go with DPX, but be prepared for massive files and a clunkier workflow (you'll need to make proxies you can edit with, then conform the DPX files to those when you're done). This is common enough, but is probably too unwieldy to try to do on your own unless you have a big, fast RAID to deal with the files - especially at 4k. A couple weeks ago I did a blog post on our web site about file sizes for resolutions above 2k (not all, but the ones we deal with most often). It includes ProRes 422, 4444 and DPX: http://www.gammaraydigital.com/blog/just-how-big-are-those-files-anyway So it becomes a trade-off: For most use cases, DPX is probably more than is necessary. If you're doing a bunch of CGI compositing, or if you need to really eke out every last bit of color in your grade, then it probably makes sense. But ProRes 4444 will get you 95% of the way there, in a file that's just a hell of a lot easier to work with. All depends on your workflow and requirements, really.

That's a pretty sleazy way to do business! We usually try to get a feel for what the client is looking to do, and then make some recommendations based on that. For example, if someone has a bunch of Regular 8 home movies that they want to edit together in Final Cut and give to their family, we usually recommend sticking with ProRes 422 for ease of use (but probably at 2k, for flexibility and because the aspect ratio matches their originals, vs HD, which means pillarboxing). Also, you're not gaining much by using ProRes 4444 with reversal film, and the files are bigger, more processor-intensive to decode, and generally harder for a consumer to deal with. If it's an archive or a library, and they want digital preservation copies and something to edit, we usually recommend 10 or (sometimes) 16bit DPX for the preservation copies and some flavor of ProRes that works for them for editing. Most students want ProRes because it's easy to work with. DPX can be a real bear if you're not set up for it, and if you're on a budget and working on a laptop or a low end desktop machine, you're not really set up for it! -perry

Not *about* filmmaking, per se, but: Woody Allen's Crimes & Misdemeanors has him acting as a struggling documentary filmmaker - shooting in 16mm, and some scenes of him using a flatbed. It's a minor sub-plot but memorable nonetheless. There's also the bit in Gimme Shelter, where Jagger is watching the footage of the murder of a fan in the crowd, on a steenbeck, while the film was being edited.