It’ll be very basic quality. You’d get about the same quality out of a TVT 16 Tobin or a ELMO TRV 16. If you do get one though let us know - the main thing you’d want to do to improve it is to replace the light with a better one - or just diffuse the supplied light properly if it has a decent LED COB. You could tell them to see if they can get the light even so it isn’t showing a pattern on the scan by positioning a piece of Opal Diffusing Glass over the light - they may be willing to do that in the factory before shipping it out. You can see the pattern I’m talking about here:
It is pretty unlikely I could ever afford the Filmic scanner, but my point was that I could potentially build a mass scanner derivative of the Kinograph, that scans very fast with minimal set up time for each reel. Then once all my films have gone through a first pass I would tear down that first iteration (cannibalizing everything I could) and build a "proper " kinograph v2, then take my time to make some primo scans from carefully selected portions of the collection. With the volume of films I have something has got to give…
For me, the first pass (let’s call them “pre-scans”) doesn’t have to be pristine - I need good enough scans to make strategic decisions. I need to catalogue the dang thing for starters! Which films have decent colour? which films are silent? is there damage? etc. I am willing to sacrifice quality to get the whole thing captured so I can analyze it all efficiently as a whole, ie. as files on my hard drive. I imagine there are a few people in the forum who can identify with my use case (or maybe not…?)
Filmkeeper, I’m a bit confused about the origin of that scan you provided in the last post. Is that a scan from a Filmic machine - or is it what you suppose one would look like?
It’s a cropped screenshot from their Vimeo video - so yes it’s representative of the quality.
What gauges are your films?
The limitation for doing mass scans will be the capture software. To do it with the Kinograph v2 you’d want capture software that can go straight to MP4 for you, and you’d want a high-speed 2K camera so you can run it at 60fps+ (like this one).
There are portions of the promotional video that use the machine output, and other portions that use shots from an computer screen. The later (and the screenshot) have very visible screen artifacts, which would not be representative of the machine scan.
The light looks plenty diffused to me? I don’t see any cracks showing up.
No form of collimated light would give that pattern in the example photo - that’s just showing the pixels on a computer monitor.
Here’s a comparison using the bed poop picture from the Johnny Depp trial. It was shot through a monitor and you will notice that it exhibits the same patterns.
…Sorry I could not resist “flushing” misinformation one post at a time.
Why would they film off a computer monitor for a promotional video? A computer monitor would show a Moiré pattern as well as flickering which it doesn’t have.
There’s deformity seen in every single example of the scans in the Vimeo video including the one above which doesn’t have the visible pattern in the bare light, but the same checkerplate pattern is present in the section above.
If I investigate that example (which is off their promotional video near the start) it has a resolution of 570p in the scan. That may have been downscaled of course from the native sensor resolution, but 570p is our starting point. The standard way to check how much detail the video contains, and this is the same thing that literally everyone does, is downscale it and upscale it back to the original size. Anyway I can resize this by 50% and upscale back to the original video size and it reveals that the picture detail appears the same, meaning that the true resolution for 16mm is 285p at the very most. Even for applications where the picture quality isn’t essential, that’s just far too low to be able to answer the important questions that @Jitterfactor pointed out. The dynamic range also appears to be so limited that you’re not going to be answering any questions about colour density quality either.
They would have something if they can get both 16mm and 8mm to at least 1K resolution with good colour, but if not its application even for “high speed” evaluation scanning is severely limited.
There are visible lines in some of the scans. The issue is when you photograph film at low res like that with poor focus it hides most of the lines anyway because they’re too fine to show up.
Of course it may be something entirely different, but that what it appears to be. Take one apart sometime and look at how it works - it has a LED chip in the centrer, then reflective/white material in a cone directing the light towards the front, and finally a plastic piece of diffusing material which has a pattern to it. If you were to photograph film directly in front of a light like that you would see a similar pattern. The diffusion is designed to spread the light in a room, not for scanning.
You wouldn’t see that pattern even if it were a downlight or other non diffused source. Where have you seen a pattern like that that could be said it was directly related to the light source?
I just don’t want people reading that and getting confused about their lighting setup.
And don’t worry about what they put in the video, some people do whacky things.
The examples in moviestuff mkii promo video look terrible, very “amateur HDR”and noise all over it. But the stuff from the scanner actually looks much better than the promo makes out.
When NSW Police in australia have a person of interest, they photograph the security monitor with their phones to post on social media. I expect because they want to get it done quickly, but yeah… people do it!
I respectfully disagree with drawing conclusions from promotional video materials, and with the conclusions drawn by @filmkeeperpromotional video investigation.
While I said from the onset that there is room for improvement for the light source. If any commercial product is seeking my feedback, I will be happy to provide it (and bill them for it).
The purpose of the forum, as I understand it, it is to assist each other in making our scanners better. With that perspective, in my opinion, making unsubstantiated claims about commercial equipment does not contribute to the above purpose.
The new light source technology is much higher power than the light source used on the previous generation of Cintel film scanners. It consists of a square array of 576 high power LEDs arranged into a grid pattern, and then focused onto the film using a new light cylinder. This new design provides more than twice the silicon area for generating light and the result is much more light is directed at the film.
they boosted the light but it appears it’s the same camera as before. So the fixed pattern noise might be masked a bit, but it’s still there. That’s one of the major problems with that scanner. I saw FPN in a demo at NAB several years ago, on the footage they were using at the show – which wasn’t especially dense. it was plainly obvious but the people demoing the film had never noticed it before.
Looks like now they can do HDR in a single pass, like the ScanStation does, which seems to have been the primary goal of this change.
I don’t think it does that it would need greater overscan like the Scanstation has (the Cintel’s overscan can be seen in their resolution chart), and the capture software would need to be totally rewritten. They’d also have to fix the misalignment problem that happens with HDR scanning as you almost always have misaligned frames (that’s why no one with Blackmagic Cintels offers HDR scanning as a standard service).
They’ve said repeatedly on their forum that it can’t ever do HDR in a single pass, examples:
“It is a continuous motion scanner, thus you can’t take two images fast enough. The old scanners used to have stop and go motion which was very harsh on the film and required all manner of other nasty processes to accomodate that (i.e. clamp gates etc…).” (James Little Jan 11 2022).
"No, the scanner is a continuous motion scanner and exposure/readout times etc… and stop and go motion of old scanners is a thing of the past.
"The only viable option is running the high exposure in ‘reverse’ to avoid excess film running but we explored that thoroughly and ran into a few roadblocks that don’t make it reliable enough.
"In short, it works how it’s always going to work.
“Cheers” (James Little Apr 28 2021).
Some of their users have made the mistake in the past of assuming that if a competitor’s CFA scanner can do something in a certain way that Blackmagic should be able to as well, and I know for a fact that users have requested support for single-pass HDR in the past at any speed even 4fps or 2fps if necessary, and they won’t support it. The reason why they want it is because the rewind between the HDR passes often/usually fails and then you have to start the entire HDR scan from scratch. If the rewind worked perfectly then all they’d have to worry about is the alignment problem which plagues almost every HDR scan, and you need to open up the .CRI’s and manually align misaligned frames. It will depend on the individual machine but HDR can fail an average of 3 or 4 times per reel.
On top of all that there are other issues with the scanner as well, it really isn’t just the imager.
Blackmagic has done a nice job of explaining why they’re using a new light for the new upcoming model (coming 2023? 2024?). It’s certainly relevant to the discussion here. As mentioned above I would be sceptical that it will fix the problems with HDR scanning, however it could make significant improvements for its regular SDR scanning and especially if they pair it with a better camera.
The really interesting thing is that Lasergraphics did the same thing years ago with the Scanstation. They started out with pretty much the same light as is in the Director which was R/G/B LED with an integrating sphere. A sphere scatters the light 180 degrees, so they came up with the column light/“integrating cube” as they call it on the website to direct all the light upwards and use diffusing material instead to get the diffusion, therefore they loose less light overall compared to a sphere. The director 10k still has a sphere but that’s because the film is held steady in the gate unlike the Scanstation where the film is in constant motion through the gate, and also because the sphere provides better diffusion for concealing base damage. So to get the same amount of light for the exposure just means a longer exposure when the light is flashed, but for the constant-motion scanner design you want the shortest possible exposure instead.
It saddens me that this forum software doesn’t allow you to block people, because my quality of life has improved greatly since I’ve blocked Daniel (@filmkeeper) elsewhere. I’m spending so much less of my time correcting his mistakes, misinformation and bad assumptions…
All you need to see to do frame alignment is some of the perfs - not all of them, and not all of the film. In fact, on the ScanStation, you only see the whole film on 8mm/Super 8, all other gauges are partially cut off. The BMD overscan is very similar to the ScanStation. Not as much overscan, but certainly enough to do software frame alignment.
Blackmagic Design is as much a software company as a hardware company these days, so I don’t see why this would keep them from making an improvement to one of their (really bad) design decisions on the original HDR version: HDR in two separate passes.
Again, so what? That’s part of their job - to make the software and hardware work. If they came up with a solution to alignment issues (which shouldn’t be especially difficult as it’s a well understood problem), then why would they not do this?
Of course they’re going to say publicly that it can’t be done. Have you never met a software engineer? The first thing a developer will tell you is that it’s impossible. If they’re good at their job it’ll get under their skin and they’ll figure out a way to do it - if there’s a way to do it. FWIW, Lasergraphics told me the same thing when I brought up HDR scanning on the ScanStation at NAB about 8 or 9 years ago. “Impossible!,” they said. It wasn’t more than a few months after that, that they figured out how to do it and they added the feature to the scanner. The ScanStation works in basically the same way as the Cintel (continuous motion, flashed light, high speed exposure, bayer sensor camera).
By the way, there’s even a T-shirt for this phenomenon:
Here’s an article that came out right around NAB 2022, where BMD announced the new light source. It clearly states that the main reason for the light output increase is to facilitate faster HDR: Blackmagic Design Cintel Scanner C-Drive HDR - Newsshooter
It doesn’t specifically say that it’s single pass. But the only real reason to increase the light output is so that you can decrease the exposure time. And the main reason to decrease the exposure time would be because you need to take two images of the frame before it exits the gate.
But just to clear this up once and for all (he said, knowing Daniel will argue that the sky is not, in fact Blue), BMD does say it’s single pass, unless they’ve figured out a way to alter the space-time continuum, to do two pass scanning in real time:
It features redesigned digital servos, a high intensity diffuse light source and advanced imaging system so you get real time HDR scanning of 35mm and 16mm film up to 30fps in Ultra HD resolution!
(Cintel | Blackmagic Design – first paragraph)
I thought a well designed IS should emit almost the entire intensity of the input light from an exit hole? Just from the idea that they use IS to test the LUX outputs from various bulbs accurately.
I wonder why they ditched the IS. I guess it’s larger, and maybe the reflective coatings longevity isn’t great?
Note - These are only my uneducated speculations with no real substantial evidence
I would not be so quick to assume they ditched the sphere, even though they’re calling it an integrating cube now. I suspect that’s a bit of marketing to make the Director look better.
We have the first model scanstation, which has been upgraded many times and is now a 6.5k model. The lamp house in this has remained the same through all the iterations. (The Archivist is different and may very well use a completely different lamp configuration, I can’t speak to that as I haven’t seen one in person).
The lamp house is the black box at the bottom in this picture - the top of it it sits about 5" below the top of the gate, which is the part with the “16mm” label:
The gate is an aluminum box with a skid plate on the top for the film to run over. Inside this box are four mirrored walls, which are tapered to reflect the light that doesn’t escape at the gate aperture, back down towards the light source:
From my own experience building an IS for our 70mm scanner, the size of the sphere needs to be significantly larger than the exit port, in order for it to cover the full film area. Our IS uses a sphere that’s almost 6" in diameter, with an exit port big enough to cover an IMAX frame. As such, most of the top of the sphere is not rounded, but it seems to work just fine.
I would imagine that the inside of the ScanStation’s lamphouse is a bit like ours - kind of a hybrid sphere/cube arrangement that’s designed to take advantage of the spherical properties as much as possible, while allowing for an exit port big enough to cover a full 35mm frame.
“flushing” misinformation one post at a time.
