Jump to content

Cinelab's new 4K scanner sample


Robert Houllahan

Recommended Posts

  • Site Sponsor

Here is a rescanned version of the 2010 Iron Guild performance at the Steel Yard in Providence, R.I.

 

 

Scanned on Cinelab's new DCS-Xena 4K scanner designed and built in LA more details about the scanner to come and pricing should be on the Cinelab web site this week.

 

-Rob-

Link to comment
Share on other sites

Hi Robert,

 

Do you have any stills that you can post? Either 4K or 2K would be acceptable. Little or no compression would be preferable.

 

I'm curious about your new scanner too since I've never heard of the model. Please let us know as soon as you've posted details on your website.

Link to comment
Share on other sites

  • Site Sponsor

This is S-16mm 7219 that I shot with an Aaton LTR-54 and a Canon zoom.

 

I think you will find the pricing structure atteactive, I am still working a few things out but it will be priced by the foot. We have a second machine that is being built right now that will be a 3.3K Color imager and a Servo drive running at about 15FPS. The current machine machine will eventually be a 6.6K Monochrome (Real 6.6k) and Oxberry pin registration runnig at about 2-3 FPS depending on resolution.

 

-Rob-

Link to comment
Share on other sites

This is S-16mm 7219 that I shot with an Aaton LTR-54 and a Canon zoom.

 

I think you will find the pricing structure atteactive, I am still working a few things out but it will be priced by the foot. We have a second machine that is being built right now that will be a 3.3K Color imager and a Servo drive running at about 15FPS. The current machine machine will eventually be a 6.6K Monochrome (Real 6.6k) and Oxberry pin registration runnig at about 2-3 FPS depending on resolution.

 

-Rob-

 

What stop was that shot at?

Link to comment
Share on other sites

  • Site Sponsor

This is a new modular scanner from a LA company run by a few guys who worked in Post at the studios, it is modular and allows for switching the imager and either Pin registration or realtime machine vision perf stabilization. It is a great design and I am sure DCS will have a web sit up for it in 2013 we are one of three users now but the machine is still in development and we have a NDA in place so i can't post pics of the machine yet. I am very happy with the scans so far and it is very reliable, I have been testing it with my own film since October.

 

-Rob-

Link to comment
Share on other sites

  • Site Sponsor

Stability is really a camera issue with the scanner as configured currently has an Oxberry dual pin registered movement, DCS will have a pinless gate with registration by perf recognition but we are not using it yet. The Iron Guild was shot with my Aaton LTR-54 which was tuned up a few years ago and has great stability. I put a scan of some Orwo-54 up (in the Orwo-54 problem thread under Film Stock and Processing) which I shot with my B&H Filmo 70DR which has some stability issues, entirely the camera/film combo as I was not able to successfully get the 54 stock to go through the Aaton.

Link to comment
Share on other sites

Not to be rude but the test you've provided is more or less useless...

 

The footage looks no different than tests I've shot with an old crotchety bolex on 7219 and ran through a Spirit. This is because as I'm sure you know... S16mm scanned at 4k or even 2k can be pretty overkill (especially if you're shooting a faster lower resolution stock with an old lens wide open at night...), to see any of the real benefits of this new scanner we need to see something that actually benefits from a 4k scan. Showing us some Super 35mm shot with a more modern lens and a comparison between a slow 50D stock and a fast 500T stock would be much, MUCH more helpful!

 

This is because (from my own personal experiences with 5219) I can hardly even tell the difference between a 2k/4k scan on the "Scanity" machine during A/B projection; on kodak 5219 at 2k you're already bordering on maximum "visible resolution"! I'm sorry if I sound condescending but i feel VERY strongly about the nature of good tests, because lets face it, film is dying, and if you've invested in this amazing machine, you need to be showing results that prove why film should stay alive and why your machine is worth it. For me, this test does neither of those and is therefore actually a bad thing. Some more extensive tests that actually utilize the machines strength would be very interesting for me and many others that still love film! Thanks Robert!

Link to comment
Share on other sites

  • Site Sponsor

Well...

 

You have to remember that even if you download the h264 from Vimeo it is still just a h264 and not a pro-Res444 or DPX sequence so it is not exactly a reference but about the best easily showable version for the web.

 

Most quality scanners scan at a higher than output resolution and I have noted that this is scanned at about 3.2k for a 2K output. I feel the a 2K scan from about a 3K raster on the sensor is about what you can get from S-16mm but that is similar to what a Alexa is or an Arriscan or Scanity and just resolution is but only one aspect of a quality scan. I have used the Spirit and seen allot of film scanned on it and I personally always found it to be a bit too artificially sharpened and that the color fidelity from the original Spirit was lacking but that is just my opinion. I will be doing som more scan tests (with my own film) and posting them plus making Pro-Res444 files available. I am personally really happy with the quality of the scans I am getting from this machine and feel that they are far superior to a Spirit scan but I am also looking at the DPX 2K sequence on a calibrated monitor on the Resolve machine.

 

As far as a business decision Cinelab is a Film Lab and we need scanning so there was no question that we would be adding several high quality hi resolution scanning options to our list of services. These scanners (we are getting a Pin Reg Monochrome imager system, which is what these were scanned on, and a faster Servo based color imager system) are high quality, fully modualr and relatively low cost to own and maintain.

 

Film has been 'Dying" for thirty years yet it still shows up every day at the door and we still develop it and print it and scan it.... so i will believe that it is dead when it does not show up.

 

-Rob-

  • Upvote 1
Link to comment
Share on other sites

  • Premium Member

Film has been 'Dying" for thirty years yet it still shows up every day at the door and we still develop it and print it and scan it.... so i will believe that it is dead when it does not show up.

 

-Rob-

 

Rob, I really want to take this quote and put it on the side of a building! So many people who have no idea what they are talking about proclaim film dead but yet labs like yours and Paul's and even Spectra continue to buy new equipment which tells me that you obviously still get a lot of business. I just want to say I appreciate the work you guys do and hope you journey on. Dont pay attention to people like Evan...your scans look great and the colors are obviously rich and provide a look that Ive never seen on any of these "magic bullet looks" digital plugins of digital footage. I look forward to shooting some footage and sending it out to you guys to see what magic you can work with it.

Link to comment
Share on other sites

Especially when talking about 50D I have to say that I totally disagree with Evan. There is definitely a worthwhile advantage to 4k Super16. However, he does have some good points about this footage not fully testing the abilities of the new scanner. I'd live to see some 4K Vision3 50D Super 16 shot in proper lighting at f5.6 with high end lenses. That would truly show the full abilities of this scanner.

 

I do, however, like this current test too because it shows the shadow highlight abilities of the scanner. Compared to a projected print, how does this scan compare in dynamic range?

 

Lastly, if you think 2k is enough for Super 16 then you haven't seen a properly shot and 2k scan of Super 8 Vision (2 or 3) 50D. It's got a lot of grain but it looks nice and definitely has plenty of "resolution" to be pulled out of it.

Link to comment
Share on other sites

  • Site Sponsor

I scanned this because it is part of a film I am trying to do some work on right now and what I like about it is not only the shadow detail but the detail in the flames. Also this is a 3.2K scan which is not derived from a color mask but with a R,G,B and IR LED lamp system and a full frame single tap CCD so it has good nyquist properties for a 2K deliverable. I am still doing some tests but I am pleased with the lack of noise in the sensor and I have a film to scan this week that is all B&W Reversal. With Reversal I typically set transfer for the hilites and I am finding that I can do that with this machine and let the shadow fall into the floor of this sensor and recover that floor in the grade without seeing noise from the CCD so I think the dynamic range is pretty good.

 

I almost bought a Celestial Imaging CCD from a guy at MIT to try on this but the CCD was so big the housing would have had to be modified to get it far enough back from the film plane, but that would be Dynamic range... However there is a new 29Mp CCD in a 35mm form that we are looking at for an upgrade.

 

I like it...

 

-Rob-

Link to comment
Share on other sites

I wanted to add to the discusiion on film speed and resolution. Film speed has a very minimal effect on the resolving power of the film. If you look strictly at the MTF graphs Kodak 7203 50D has the lowest depth of modulation at 20lp/mm , however it has the lowest granularity of all the camera film stocks and also the least response variance between colors. 500T 7219 has the highest response at 20lp/mm. At this detail frequency 7219 actually "rings" it has a blue response of above 100%. At a 50% threshold of depth of modulation all the stocks have similar performance 50D has a higher red resolution than other stocks but the lowest green and blue. In the end these numbers are lowered substantially by the MTF of the lens and Kodak color film stocks have basically the same resolving power when looking at it in a digital resolution(pixel count) perspective. Graininess is where they differ substantially, it is not the spatial resolving power but the uniformity of response.

Link to comment
Share on other sites

For my previous post when I wrote Kodak color film stocks have basically the same resolving power when looking at it in a digital resolution(pixel count) perspective, I should have stated film of the same size. The emulsion resolving power is the same regardless of format, needed digital resolution is based on the size of the image area used on the film.

Link to comment
Share on other sites

  • Premium Member

I think 29 megapixels is probably a bit overkill for 16mm, but hey, there's nothing wrong with overkill. Make sure you've got everything, then scale it down by half for another 6dB of noisefloor!

 

I do agree that the slightly raw look of the Iron Guild stuff doesn't show us much of what the format itself and postproduction equipment can do. It certainly doesn't challenge the idea that 16 isn't a good idea for HD origination. Personally I think that 16mm can, under extremely careful circumstances with the best of everything, be OK for producing HD material, but you'd have to be very careful to make it look adequately sharp. I'd like to see someone try it.

 

In terms of sensors, well, I have a monochrome version of the Red Epic not three feet from me as I type, which would probably be a pretty good bet!

Link to comment
Share on other sites

  • Site Sponsor

I have to take umbridge with your statement here Phil "It certainly doesn't challenge the idea that 16 isn't a good idea for HD origination" There have been allot of people who have shot this event with both still and HD digital cameras and all of their footage looks like pure crap. Noisy blacks and completely clipped hilites do not a pretty picture make IMO. Maybe next year I will get an Alexa which may be able to capture this kind of dynamic range. I also do not see a lack of sharpness or detail on either a 42" Plasma or a 110" DLP projector.

 

AS for higher MP sensors, I would not want to use a rolling shutter sensor for film scanning as the field correction would probably be problematic, also the new 29MP sensor I am looking at increases in speed when you window it but still has a small number of taps and a global full frame shutter plus far better noise performance than that sensor you mention. Also this machine will scan 35mm and there will be a version for 8K+ scans of 65/70mm as well. I think a 6K scan of 35mm especially anamorphic is a great fit for the format, and it allows for lots of oversampling and with better oversampling the grain of the film is better resolved.

 

As Thomas said the resolving power of the film is really about the area of the film scanned/used for picture and one thing that I have noticed with scanning is that resolving the grain structure of the film is an issue especially when the target resolution is not super sampled. A 2K scan from a 2K sensor raster produces blocky grain as far as I have seen and then adding NR usually makes the picture overly clean to my eyes. With a 3K or 4K imager raster the 2K scan has much better looking grain. I think the old Domino 3K format might be a great fit for S-16mm if scanned at 4K or higher and I will be trying it out this week.

 

-Rob-

Link to comment
Share on other sites

I wanted to add to the discusiion on film speed and resolution. Film speed has a very minimal effect on the resolving power of the film. If you look strictly at the MTF graphs Kodak 7203 50D has the lowest depth of modulation at 20lp/mm , however it has the lowest granularity of all the camera film stocks and also the least response variance between colors. 500T 7219 has the highest response at 20lp/mm. At this detail frequency 7219 actually "rings" it has a blue response of above 100%. At a 50% threshold of depth of modulation all the stocks have similar performance 50D has a higher red resolution than other stocks but the lowest green and blue. In the end these numbers are lowered substantially by the MTF of the lens and Kodak color film stocks have basically the same resolving power when looking at it in a digital resolution(pixel count) perspective. Graininess is where they differ substantially, it is not the spatial resolving power but the uniformity of response.

 

Kudos for the info :), but I know what my eye sees, and on a big screen, 35mm scanned at 4k, 50D will always look sharper than 500T. Just my opinion

Link to comment
Share on other sites

Well...

 

You have to remember that even if you download the h264 from Vimeo it is still just a h264 and not a pro-Res444 or DPX sequence so it is not exactly a reference but about the best easily showable version for the web.

 

Most quality scanners scan at a higher than output resolution and I have noted that this is scanned at about 3.2k for a 2K output. I feel the a 2K scan from about a 3K raster on the sensor is about what you can get from S-16mm but that is similar to what a Alexa is or an Arriscan or Scanity and just resolution is but only one aspect of a quality scan. I have used the Spirit and seen allot of film scanned on it and I personally always found it to be a bit too artificially sharpened and that the color fidelity from the original Spirit was lacking but that is just my opinion. I will be doing som more scan tests (with my own film) and posting them plus making Pro-Res444 files available. I am personally really happy with the quality of the scans I am getting from this machine and feel that they are far superior to a Spirit scan but I am also looking at the DPX 2K sequence on a calibrated monitor on the Resolve machine.

 

As far as a business decision Cinelab is a Film Lab and we need scanning so there was no question that we would be adding several high quality hi resolution scanning options to our list of services. These scanners (we are getting a Pin Reg Monochrome imager system, which is what these were scanned on, and a faster Servo based color imager system) are high quality, fully modualr and relatively low cost to own and maintain.

 

Film has been 'Dying" for thirty years yet it still shows up every day at the door and we still develop it and print it and scan it.... so i will believe that it is dead when it does not show up.

 

-Rob-

If you read...

 

I'm talking much less about how the film is presented (vimeo, compressed) and far more about the actual subject matter and technique behind the tests. I was unaware that you were simply displaying some film you had shot. I thought that this was legitimate "test footage" designed to show the capability of your scanner, which, as I'm sure you would agree with me, is not adequate!!! Does the footage look pretty? Absolutely! But it does nothing to show me the capability of your scanner, and if you think it looks any different than a scan of bolex s16mm footage I've gotten on an old spirit, I would be happy to show you footage that says otherwise.... sorry for being difficult :(

 

Definitely don't want to hijack the thread and begin a "is film dying or not" argument. But lets not kid ourselves, fuji doesn't even produce motion picture film anymore! And it is INCREDIBLY rare that labs invest in new scanners. Last time I spoke with Paul at Cinelicious I believe he said they were one of two in the states that owned the Scanity, most are still running spirits etc. Admittedly I am far from an expert on the matter but I think it is COMMON SENSE that if you care about film you would present tests that stagger the audience and leave their jaw dropping! 35mm 6k anamorphic scan?? Sounds great!!! If you have film whizzing in and out of cinelab like you say, then can you not ask a talented DP to utilize some of his footage to show off the scanner???

 

Again, I am sorry to be difficult! But this IS important! Film IS dying and we NEED every example of film to absolutely shine if we want it to stick around. Personally I am absolutely in love with film and I am sticking around for the long run, I don't need your tests to prove to me film is excellent. There are millions of established DP's who know film is excellent. They will stick around too. But we need the newer, younger, generation of DP's who are learning on 5d/RED/ALEXA to see what film can actually do to motivate them to try out a new format!! Being a young DP myself, I can confidently say that about 10% of my peers shoot film. This is not because they don't like film, but because they've been told its either expensive, difficult, or antiquated. They couldn't be happier with their RED images and don't see how film could be any better. If a lab like yours could politely offer them incredible test material and fair pricing....! It would be incredibly helpful :)

 

Thanks for reading my rant, all the best, and congrats on your scanner! Kudos for fighting the good fight!

Link to comment
Share on other sites

Evan,

You are correct that younger dps prefer digital for convenience. They almost cannot screw up while using digital, because of instant playback. To me, I use my light meter, and pretty know what to get in the end. It's funny, everytime they see a fottage shot on s16, they say "wow, that's retro" film is whatever you want it to be, retro or modern..

Link to comment
Share on other sites

.....If you look strictly at the MTF graphs Kodak 7203 50D has the lowest depth of modulation at 20lp/mm , however it has the lowest granularity of all the camera film stocks and also the least response variance between colors. 500T 7219 has the highest response at 20lp/mm. At this detail frequency 7219 actually "rings" it has a blue response of above 100%. At a 50% threshold of depth of modulation all the stocks have similar performance 50D has a higher red resolution than other stocks but the lowest green and blue. In the end these numbers are lowered substantially by the MTF of the lens........ Graininess is where they differ substantially, it is not the spatial resolving power but the uniformity of response.

 

Prefix this with me saying that I'm not trained or experienced in interpreting MTF data relative to grain to get a satisfactory feeling (intuitive explanation) about percieved sharpness.

 

I overlaid those Kodak MTF graphs for 5203 (50D) and 5219 (500T) and the B, G response of 5219 is noticably higher from 20cycles/mm. At 50cycles/mm, quite a big difference, (10% and 20% for Green, Blue. Is there a simple way to understand this? Does larger grain make it easier to maintain those higher MTF values?

 

I often wonder why Kodak MTF graphs always seem to stop at 80cycles/mm. For 16mm, lenses that can resolve 200 lp/mm have been in common use since the 80s. I followed the technician looking at a set of MKII Zeiss recently and all but one looked extremly crisp at 200 lp/mm. I don't know what % value I could put on it, very close to black and white.

Edited by Gregg MacPherson
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...