Maximum Scan Resolution for 35mm Films?

[Tom Lowe]

Obviously, 4K is the gold standard now for 35mm scans, transfers and restorations of motion pictures shot on 35mm.

 

I think I read a recent tidbit in the latest AC saying that most 35mm scans max out around 4K, and 6K can be fruitless and futile in terms of more resolution gain for many 35mm pictures.

 

What do you guys think is the max scan resolution for 35mm negatives in "good" condition?

 

Is this a matter of 35mm having a ceiling, or is it a matter of getting better scan technology? What is the theoretical max resolution of 35mm negs? I assume more recent, better-kept negs can scan higher res.

[Paul Bruening]

This has been a frequent topic and rightly so. David Mullen and Dominic Case, some of the most authoritative guys here have ballparked 35mm, low ASA negatives at around and under 5K. (I hope they don't mind me speaking for them) So, even though scanning technology can handle 6K, there's little or no gain in staying that high. The scan salesmen will say, "We scan at 6K but downres to 4K". If you ask them if they could keep it at 6K, you can get a rather long and disturbing silence.

[Chris Keth]

I don't know the number, I haven't done the tests. I can tell you that there is a practical limit. ABove that, you're really just scanning each grain of film with more pixels. Who needs to see each grain of film clearer when it doesn't make a visible difference to picture quality at normal viewing sizes?

[David Mullen ASC]

I think 4K is a good figure, there are some 35mm movies that have less resolution than that and a rare few that have more, particularly something shot in 35mm anamorphic on fine-grained stock with good lenses at a decent f-stop, scene contrast, etc.

[Stephen Williams]

Hi,

 

The Arriscanner scans @6k then down samples to 4k, that seems to be around the practical limit.

 

Stephen

[K Borowski]

Hi,

 

The Arriscanner scans @6k then down samples to 4k, that seems to be around the practical limit.

 

Stephen

 

I'd agree that this is the practical limit for current and any future 35mm 4-perf stocks. The experts argue that with 35mm still film that 12-22 MP is the practical limit. 12 is far too low, and 22 is too high except with maybe 50-speed films. These are 8-perf frames. So do the simple math and divide by two, and you're getting at most 11MP, about 3.3K with 4-perf. I'd say that, for 4-perf., 3.5K is a safe number in terms of actual resolution stored (with 500T being more like 3K and the other stocks falling somewhere in-between), and, to make sure all of the information makes it from film to digital, 6K is the maximum needed scan resolution with a 4K output being all that is needed.

[Dominic Case]

I don't know the number, I haven't done the tests. I can tell you that there is a practical limit. ABove that, you're really just scanning each grain of film with more pixels. Who needs to see each grain of film clearer when it doesn't make a visible difference to picture quality at normal viewing sizes?

A film emulsion contains grains - or dye clusters - of a whole range of sizes, some very much smaller than others. This is what gives film its dynamic range, as the larger grans are more sensitive, and the smaller grains provide the detail. But this range of sizes means that while it is generally agreed that 4K does a fair job of capturing most of the grain structure of the image, you can go on with more and more, smaller and smaller pixels, and continue to capture more information with much finer resoluton than 4K.

 

Even if the smallest grains, at normal viewing sizes, are below the threshhold of visual resolution, they still contribute to the desnsity and colour of their immediate area of the image.

 

However, this is a theoretical argument - increasing pixel resolution doesn't so much have a practical limit as a low of diminishing returns - of course if you go on decreasing pixel size, you reduce the sensitivity of the photosite, so apart from having to process more pixels, it takes longer to capture each one, with ever less addition to the actual information, such that it eventually sinks beneath the noise level.