Home of Professional Cinematography since 1996

class="style10">Film Resolution 550 MegaPixels On VistaVision

>Published : 11th Sept. 2006

>On 30 January 2006 I stated right here on CML (after another CML'er posted that I was being "arbitrary" and "silly" to think that 8K / 43 MP scanning was needed to capture all the information in a 35mm still or VistaVision negative):

>"8K is not arbitrary or silly and it may not be enough to fully capture the film negative, especially for the larger formats such as VistaVision.

>If the comparison picture is taken with MY 35mm camera (a Contax G2 - we're comparing still cameras right?) I think it will hold it's own against a 29 MegaPixel (MP) digital capture (ignoring the differences in DOF caused by the lens format size difference - which could be achieved by using an equivalent f-stop ratio)."

>I got somewhat flamed for the above statement. Some people thought I was full of it. I guess some folks think I'm breathing too much film fumes over here.

>Take a look at this page (Carl Zeiss Camera Lens News No. 24, February 2006)

>http://www.zeiss.de/C12567A8003B58B9/?Open

>Take a look particularly at the sub-page for the Biogon 2,8/25mm (for the Zeiss Ikon, the successor to my Contax G2). See how a tiny slice of the full image has so much detail. This is EXACTLY what I was describing having seen with my own photographs projected on a good slide projector.

>Carl Zeiss has been getting results up to 400 lp/mm using available filmstocks (Kodak Imagelink HQ B&W) and production cameras & lenses.

>The MegaPixels (MP) stated below are based on scanning the full 24x36mm area of a 35mm still photo negative.

class="style11">250 lp/mm (achieved with the 1,4/85mm Planar at f2,0) is equal to 215 MP
320 lp/mm (achieved with the 1,4/50mm Planar at f2,8) is equal to 350 MP
400 lp/mm (achieved with the 2,8/25mm Biogon at f4,0) is equal to 550 MP

>If we put that ImageLink film in a Super 16mm camera we would get 50 MP out of that little negative!

>AND we're not even accounting for the losses due to the optical low-pass filter that is needed in all CCD and CMOS digital imagers which results in the response at maximum resolution (equalling the pixel pitch) to be ZERO.

>Of course we are talking about Black & White film here (always higher resolution). But you can get close to 160 lp/mm with the best colour film.

>OK, putting on my nomex suit, I await the flames...

Jorge Diaz-Amador
Designer / Technician
CinemaTechnic, Inc.
Miami, FL USA
http://www.cinematechnic.com


>Why would anyone think that is was silly to scan 8-perf at 8K across when 4K to 6K is considered necessary to scan 4-perf 35mm at full rez without any loss of information? I mean, you should at least scan VistaVision at 6K if you normally scan 4-perf 35mm Full Aperture at 4000 x 3000 pixels, right? Since you're talking about turning the frame sideways and doubling it.

>The only problem I find with these arguments about the high resolving power of film (which I don't deny) is that in the real world, there are other factors that can give the impression of greater or lower resolution, so the numbers don't tell you everything. Otherwise, you fall into that trap of arguing that Super-8 film actually has more resolution than HD, which even if you can prove on paper, doesn't mean you can achieve practically. But the Super-8 advocates like to make this argument all the time.

>The lack of grain in digital images, and steadiness, can give them greater flexibility in enlarging the image without as many telltale visual clues about the original file size, hence why HD, though less than 2K, can be blown-up to IMAX with more, let's say, "passable" results than some Super-16 photography.

>I fully accept that that last point is debatable of course (I'm not talking about how good the image looks, just on resizability).

>David Mullen, ASC
Los Angeles


class="style12">>Why would anyone think that is was silly to scan 8-perf at 8K across >when 4K to 6K is considered necessary to scan 4-perf 35mm at full rez >without any loss of information?

>David,

>Agreed, but I did get called "silly" and my 8K number was "arbitrary" (in a post on 30 January of this year).

>You won't hear me arguing extreme numbers for Super 8. Don't get me wrong, I love the stuff and shot my first 2 short films on it. The problem is Super 8 was always intended as a CONSUMER format.

>First of all, the Super 8mm negative is TINY at 4.0 x 5.45mm, (slightly larger than the ubiquitous 1/3in 3-CCD camcorders) giving an area of 22mm sq. as compared to 93mm sq for Super 16 and 340mm sq for 35mm (3-perf 16:9).

>To get 480 x 720 resolution to match NTSC digital video you would need to get 65 lp/mm resolution. This would be doable IF you had a rigid high precision lens mount/aperture/pressure plate system, and a very finely adjustable ground glass (for focus calibration). But NONE of the Super 8mm cameras offer this.

>The S8mm pressure plate is in the plastic film magazine, which is not going to give consistent back pressure on the film, or even a perfectly flat pressure plate. Also most S8 cameras do not have a frame bar across the top and bottom of the gate (they had the original "hair free gate"). If you don't have a flat surface to work with you can't get the film flat.

>Also, with a fixed zoom lens and all-in-one construction on most S8mm cameras, it is very difficult to adjust the lens focus and ground glass focus to tight tolerances - absolutely necessary for the best results (and the main reason most people don't know the real potential of Super 16).

>In contrast, a 16SR has both a very flat gate and a very flat spring loaded pressure plate with a spring loaded floating pressure pad. This keeps the film flat to 0.01mm tolerance and allows high resolution to be achieved.

>Granted, 16mm was also originally intended as a consumer format, but when it became the standard for television news, there was a reason for manufacturers to develop truly professional 16mm cameras (starting with the Arriflex 16St in 1952).

> Jorge Diaz-Amador
Designer / Technician
CinemaTechnic, Inc.
Miami, FL USA
http://www.cinematechnic.com


class="style12">>>hence why HD, though less than 2K, can be blown-up to IMAX with >more, let's say, "passable" results than some Super-16 photography

>As someone who makes a living partly by blowing stuff up to IMAX, I don't think it's debatable at all - HD definitely works better than S16. What's debatable is whether it works better than 35mm 4-perf. 8-perf is definitely better than HD, no contest, but there's a lot of camera craziness.

>In my experience, you get more mileage from interframe analysis of film, because so much of the resolution is intermittent, whereas HD is about the same still or in motion, but HD is so much cleaner overall that it cuts into 15-perf 65mm better, though it never looks as sharp, unsurprisingly. That's if it's shot well.

>And PS I don't know why anyone would mess around with Super 8. It's crap. Regular 8mm rules!

>Tim Sassoon
SFD vfx & creative post
Santa Monica, CA


class="style12">>hence why HD, though less than 2K, can be blown-up to IMAX with >more, let's say, "passable" results than some Super-16 photography

>Tim,

>I'm not disputing your results (which I haven't seen) but notice I'm always talking about _image resolution_ not _perceived image sharpness_.

>Its very possible for a lower resolution image with a big bump in the MTF response at the middle frequencies to look "sharper" than high resolution image with a smooth linear MTF curve. In fact the HD cameras are designed to achieve this result (see the Sony CineAlta "Issues of Image Resolution" white paper).

>For example, everyone thinks the Zeiss 10-100mm T2 zoom is very sharp, but it can barely resolve 160 lp/mm. However it has good response in the 10 – 50 lp/mm range (due to high contrast) and that's what people notice.

>The Angenieux 11.5-138mm HR zoom has higher resolution (easily 200 lp/mm), but on an NTSC transfer you won't notice the difference. All the high spatial frequencies above 30 lp/mm are cut off by the SD transfer.

>However, I think that it would be possible to boost the MTF response of a Super 16mm film to digital/HD transfer in the same way that the HD camera's natural response is boosted.

>However if the transfer is to 1080p the format's cut-off is 80 lp/mm (from S16mm) so if you are getting any resolution above that on the film it won't be seen on the HD transfer.

>As I've mentioned before I plan to do a test to answer this question. When I catch up with my back-log of work I will have some time to dedicate to this project. I've already lined up the cooperation of some key players.

> Jorge Diaz-Amador
Designer / Technician
CinemaTechnic, Inc.
Miami, FL USA
http://www.cinematechnic.com


class="style12">>>However if the transfer is to 1080p the format's cut-off is 80 lp/mm >(from S16mm) so .if you are getting any resolution above that on the >film it won't be seen on the HD .transfer.

>The main objection to S16 blowup to large format is the noise level, rather than a specific measure of sharpness, which is why HD punches above it's weight. BTW, Vistavision scanned at 6K is pretty noisy, about the same as 4-perf at 4K. HD can be tricky to sharpen for blow-up.

>Tim Sassoon
SFD vfx & creative post
Santa Monica, CA


class="style12">>The main objection to S16 blowup to large format is the noise level, >rather than a specific measure of sharpness,

>Noise level - grain, in other words?

>It's entirely possible to have high image resolution and grain because the eye can perceive details even when they are just barely above the "noise level" of the grain. This is due to the analog nature of the grain and its random structure.

>I do see where intercepting S16 with IMAX is going to show a BIG difference in grain.

>BTW, do you know how the images for the Disney "Soaring" ride were captured and posted? There are some CGI elements in it, but I could not see obvious pixel issues.

>I think it was OmniMax projection. I saw the "horizontal dirt effect" like IMAX.

> Jorge Diaz-Amador
Designer / Technician
CinemaTechnic, Inc.
Miami, FL USA
http://www.cinematechnic.com