Jump to content

Perry Paolantonio

Basic Member
  • Content Count

  • Joined

  • Last visited

  • Days Won


Perry Paolantonio last won the day on November 28 2018

Perry Paolantonio had the most liked content!

Community Reputation

90 Excellent

About Perry Paolantonio

  • Rank

Profile Information

  • Occupation
  • Location
    Boston, MA
  • My Gear
    Eclair ACL II, Pro8mm modded Max8 Beaulieu 4008
  • Specialties
    5k, 4k, UHD, 2k Film Scanning, Film Restoration, Blu-ray and DVD Authoring

Contact Methods

  • Website URL

Recent Profile Visitors

12771 profile views
  1. "4k" is not really a spec, per se. It's a rough pixel count of only the horizontal component of an image's dimensions, and it can vary. 3656x2664 is 4k as generated by our Northlight scanner for a 4k Academy Aperture 35mm scan. It does 4096 for full aperture, but if you don't include the soundtrack area the file is 3656. I'd respectfully disagree that UHD isn't 4k. I mean, technically, it's not 4000+ pixels across, but it's advertised widely as 4k and it's what people know as 4k. Arguing otherwise is kind of futile since that's what most of the world knows as 4k at this point. Of course, the vertical number is whatever you need it to be to match your aspect ratio. so UHD is 16x9 which is purely a resolution from the video world and doesn't exist in film or photography really. If you scan most non-widescreen motion picture film you're usually looking at 4096x3112 or thereabouts, but I've seen other resolutions as well.
  2. FPN is different on every sensor, so probably not. To a large degree it can be compensated for with proper sensor calibration. Our ScanStation exhibits very little noise, except with the densest of films (a situation where there's not enough light getting to the sensor - when you try to pull details out of that dense film in grading, you expose the underlying noise). HDR scans basically eliminate this by doing a special exposure for the dense areas, by getting more light to the sensor.
  3. 4k is 4 times the "resolution" (that is, pixel count) of 2k. Assuming you're comparing apples to apples in terms of aspect ratio, a 2048x1556 file is 4x fewer pixels than a 4096x3112 file. But as mentioned in my last comment, you're not going to squeeze picture information out of the film if it's not there. But if you scan at 4k you are going to get a better scan, because you're starting with more digital samples of the analog image.
  4. One should never upscale the film, as you have to create picture data where there wasn't any before, and that will always degrade the image. See the example about halfway down the page here: https://www.gammaraydigital.com/blog/busting-resolution-myth If the goal is to squeeze more picture information out of the film by increasing the resolution, that will depend on a bunch of factors: the film stock, the camera's focus, the lens quality, the steadiness of the film in the gate, the steadiness of the camera, the exposure, the lighting, and I'm sure more. The fact is, you can see more defined grain in a natively scanned 4k image than in a 2k image that was scaled up. You're using more pixels to make the initial image, thus you're taking more samples (think audio sampling - same idea). And the end result is that you get a better representation of the physical film, which is what holds the image. And that's the whole point.
  5. I guess what I'm getting at is that it's not an HDR scan if you're starting from the same, single scan. It's kind of mimicking what you'd get with HDR, but it's only able to work with what's in the initial scan. A true HDR scan, made from multiple images, works by making different exposures for each at the time the scan is made. that means for really dense areas of the film, more light gets to the scanner's sensor, and more detail, if it's there, is extracted. And for the thinner areas of film, a normal exposure is made, which ensures nothing gets blown out. But if you start from a single scan, you're only able to work with what's in that scan, it's not able to pull out detail in areas where the scanner didn't pick it up. What you're describing should be doable with any good photo editing software, but it's not an HDR scan. We're working on a web site redesign, which will include more HDR samples. Our current scanner does 2-flash HDR only, though we'll have a 3-flash scanner within the next few months. It's more expensive because the machine has to run twice (or three times, for 3-flash) as slow, in order to capture all the images. the scanner handles doing that, and it outputs a single SDR image. The basic idea is the same as what you're doing (at least algorithmically), except that instead of starting from one scan, it's starting from two or three.
  6. If I understand correctly what's happening here, a single scan is split into three versions, each with a "-1, 0 and +1" (stop?) exposure, then merged together in the same way a multi-shot HDR would be? If this is correct, while it may have some effect on the image, it's basically just a post production grading technique that could also be done with curves or multiple-layer or multiple node color correction. It's not really extending the dynamic range, or getting any of the benefits of scanning the film at two or three exposures and merging them. There is a significant difference. With a true HDR scan, you get three things: 1) Extended dynamic range by scanning for the dense, normal and thin areas of the film, then merging them into a single SDR image. The end result is detail in all three ranges. A scan done for the shadows would blow out the highlights, and vice versa. 2) Depending on how it's done, for color images you get more bit depth in the resulting file than you would with a single scan 3) Reduced noise (if there's noise in the scan in dense areas of the film, an HDR scan will pretty much eliminate the noise from the scanner's sensor, by pumping more light to it, and creating an image well above it's noise floor. -perry
  7. HDR is useful for all kinds of film, not just intermediates. To be clear here, HDR in the context of film scanning and HDR in the context of viewing platforms (screens) are different things. Unfortunately they use the same name, but they're basically unrelated. HDR in scanning works by taking two or three exposures of a frame and combining them into a single image that includes shadow and highlight detail, rather than favoring one over the other. It is very much useful for just about any type of film (camera original, intermediates, etc). Print sees the least benefit, but it can still be useful there. There are three advantages to HDR scans: 1) Increased dynamic range. If the sensor in the scanner has a limited dynamic range, a 2 or 3 flash scan will increase this range, sometimes dramatically. We've seen shadow details come out of dense Kodachrome that clients didn't even know was there. 2) Improved signal to noise ratio: If the sensor has issues with dense film (similar to a low-light situation with a digital cinema camera) where you'd see the sensor noise, an HDR scan eliminates this problem by doing a special exposure for the dense film, with either more light or more exposure time to overcome the sensor's noise floor. 3) Increased bit depth: This will depend on the scanner and the image processing, but I can tell you definitively that in the ScanStation the standard bit depth is 10bit. With 2-pass HDR it's 14bit. In the Director, with 3-pass HDR it's 16bit. In the end, the file that comes out of the scanner is the same file format in the same color space as a standard dynamic range scan. The difference is how much more you can push and pull the image in grading, because the extremes now become more workable. BTW, with print, the increased bit depth can be really useful, especially if the print has some color fading. It gives you a bit more flexibility when grading and we've found that we've been able to pull color out of some prints with HDR that we couldn't with an SDR scan. When the color is that marginal, every bit counts. It's less useful with a good print, since prints have their dynamic range limits baked in from when they're timed at the lab.
  8. I don't have direct experience with the Cintel scanner other than demos at NAB. But it's got an incredibly noisy sensor with dense film - so noisy you could see it on the demo film they were using at the show, for at least the first two years they had it there. Assuming the end result of their HDR is similar to what you get on other scanners that do the two or more flashes at the same time, like the Lasergraphics or Arriscans, (rather than as a second pass with post-processing), you should see a significant improvement in noise, if not dynamic range. That said, Tyler is right - you're not getting 4k UHD from 16mm with that machine. Roughly HD is the max res for it . If you're being sold a 4k 16mm scan, whoever is selling you that is cheating and upscaling from HD to UHD (which is a substantial upscale and will definitely result in degraded image quality and significantly increased softness).
  9. Why? It's either going to work or it's not. We've bought plenty of them - DLT4000 (at least three), DLT7000, DLT8000 (2 of these), DLT80, LTO2, LTO4, LTO5, LTO6, LTO7. All used. All have made hundreds of tapes with no issues. All tapes verify correctly, and all of the tapes made work on client systems. These are robust machines, designed to run tens of thousands of tapes in their lifetimes. Most of the used ones were upgraded by IT departments when they moved to a newer format, so they've depreciated somewhat and they're just looking to recoup costs. Just look on ebay and you'll find them used and refurbished. You might need to buy an internal drive and a separate enclosure for it to get the best deal, but putting them together is as simple as installing a hard drive. Here are recent sales on ebay: https://www.ebay.com/sch/i.html?_nkw=lto5+external+drive&_sop=15&rt=nc&LH_Sold=1&LH_Complete=1
  10. As Rob said, stick with LTO a generation or two behind the current one and you'll be fine. An LTO 5 tape holds 1.5TB uncompressed and costs about $20 - cheaper in bulk. A used drive will run you about $300 and a PCIe SAS HBA card about $50. If you're on Windows or Linux, just download the LTFS drivers and you drag and drop your files to it. Mac is a bit more of a pain (you need a drive with a thunderbolt enclosure, and software like YoYotta or similar, or you have to jump through a lot of hoops). LTFS tapes will open on any platform - they just show up as if they're a hard drive. It's not an expensive format, and when you upgrade that LTO5 to an LTO7, it'll still read your LTO 5 tapes so you don't need to migrate as frequently. As far as your experience with those DVDs, I mean no offense but that's just anecdata. It's your experience only, and while you're stress testing them, that's not telling you anything about how they'll hold up over time or how likely it'll be you'll find a machine to read them on in the future. Basically, you know you can keep them in the sun for a while - at least, that batch of discs you tested.
  11. Burned media and replicated media are completely different things, and you can't compare them. A disc pressed in a factory is physically molded by a press and is significantly more robust than a disc you burn on your desktop. Of course CDs from the 80s can work - they were designed to last. Burned discs are throwaways and were never meant to be a long term storage solution. Any disc you burn in a burner is going to fail sooner or later, marketing claims aside. Our business was DVD and then Blu-ray authoring, for almost 20 years. Believe me, we've tried hundreds of different types of discs, used quality burners, etc. They can't and shouldn't be trusted to last long term. And even if they were trustworthy, they're too small to be useful as a backup medium, especially for video files. Trying to find a single format that will last for a long time is a fools errand - it's a waste of time at best, and at worst, it's putting your backup in jeopardy. Aside from the fact that nobody can definitively prove that discs will still work in 50-100 years like some claim, there are the much more basic issues of: 1) Can you find an optical disc reader to read them with? 2) If you're setting one aside now for future use, will it still work? 3) And if you did, will computers 10-15 years from now even have USB? The only sane strategy for digital backup is to put the files on multiple formats, and keep migrating it to new formats so you don't fall behind. Thinking back 20 years... try working with a ZIP drive today: if you can find a drive, good luck getting the cart to mount. I suggested LTO because it's cheap per gigabyte, reasonably fast and easy to work with and is widely supported by the enterprise IT world, with a published development roadmap and schedule. It's not tied to a single company, it's a consortium, with agreed upon standards. It's the best long-term physical media backup solution out there right now. The best overall solution is to have multiple copies in multiple locations and to periodically copy them to new media. Cloud storage makes the most sense and is where things are headed, for obvious reasons. There's too much IT infrastructure needed to make a personal cloud viable - for one thing, if you have a fire or flood in your basement server room, you lose everything. You need to piggyback on what large corporations are doing, and take advantage of the underlying technology (which is making sure there's redundancy and plenty of backups so you don't lose anything).
  12. I wouldn't trust optical media for backups. Ever. and I've personally burned about 250,000 discs over the past 19 years (on high quality media), and they simply don't last. I also wouldn't believe claims about "archival" optical discs lasting for 100 years as some claim. For one thing, the odds of finding a player for the media in the next 10 years are questionable, let alone 20, 50, 100 years from now. Any digital backup strategy that doesn't involve migration of data from one format to another is doomed to fail, either through mechanical/media issues or technological obsolescence. There are some cloud storage services that might make sense depending on the volume of material you have, but right now the best bang for your buck is LTO tape. yes, tape. It's robust, designed to be bulletproof, and is widely adopted by the enterprise IT world. Hell, Amazon Glacier is on LTO... But even that needs to be migrated to a new format every few years.
  13. I'm willing to bet they're selling a lot more scanstations. That said, there are other advantages to the Director (triple flash HDR, IR scanning, higher resolution, etc)
  14. In some cases maybe. We recommend Super 8 is overscanned. Because the frame is so bouncy in most cameras this gets you a frame line to grab onto if you want to stabilize. Regular 8 can be interesting if overscanned, because there's often picture between the perfs. For 16, S16, 35mm it's usually less useful. This is because those cameras tended to be so much more stable (at least the good ones), and the scanner is optically pin registered. So it's already as steady, if not more steady than a mechanical registration pin, on the scanner side. That said, I've seen some pretty bouncy images from different scanners so it depends on the machine you're scanning on too. Archives tend to like overscan so they're sure they get the whole frame. Mostly, though, it's a trendy look. Even when we crop inside the frame lines, you're getting more picture than you likely would have in most 16mm projectors...
  15. While this is technically true, we did some tests with our ScanStation in HDR mode vs our Northlight, and found that the ScanStation's picture quality is pretty much on par. The ScanStation is up to 14bit color in HDR mode, and the Northlight was actually noisier in the denser areas. Again, I'm talking HDR here, in SDR everything you say about the bayer limitations and the noise are true. However, it's also worth pointing out that if you do a Super2k SDR scan (an oversampled scan), you're basically overcoming the theoretical bayer limitations, and that, in the real world, they're not really limitations unless you're doing some extreme pushing and pulling on the color.
  • Create New...