True Resolution of film

[larkis]

I ran across this

video on youtube and was fascinated by it for a couple of reasons. It seems to throw around a bunch of numbers that seem factual, but arrives at a crack pot conclusion, easily debunked by anyone with eyes. The video seems to follow the same playbook as a lot of medical quackery videos/articles online.

I would be curious if one of the more engineering focused members could point out where the big mistake in interpreting the data is happening. As someone who has shot and developed film for years, I understand the hipster and gen-z draw towards something "magical" of the past, but for me the reason to shoot film has nothing to do with its resolution characteristics, especially not in the 35mm format.

 

[MGrayson]

To me, the biggest howler is the persistent use of the 1000:1 contrast ratio lp/mm number. Can you imagine an image where EVERY pair of adjacent pixels has a 10-stop brightness difference? I'd bet the usual number is 1.01 (just a second - I'll do a measurement... Huh. First image I tried gave 1.007 - lucky guess).
Now 1000 and 1.007 are wildly different numbers (17 stops different). The other quoted number was, I believe for 1.6 contrast ratio of adjacent line pairs. That number was about 1/3 to 1/4 of the 1000:1 number, and that would cut his pixels calculations by 4 to 9.

So 8x10 film being a few hundred MP? Absolutely. IF every pair of adjacent pixels has a 1.6 contrast ratio. See the 1.007 number above.

That's my take. I'm sure those more knowledgable will have better answers.

Matt

Ok, a more honest calculation, as only a few places need to be sharp to look sharp - it's not the average contrast that really matters. So in my random test image, only one pixel (out of 65,000,000) had a contrast ratio > 1000. Ten thousand of them (1 in 6,500) had a contrast ratio > 1.6. No, I don't know what this all means.

 

[Duff photographer]

The video seems generally legit. He over-emphasises a few things like contrast and that 800 lp/mm large format film would outperform everything - it would if it existed and had the lens that could convey that resolution to the film (that too, doesn't exist), but does balance out the theoretical resolution with practical resolution, by listing all the issues that reduce resolution; something which also applies to digital.

What he says about Adox is fine too. I have seen a 6 foot print from, I think Gigabit 35mm, which was effectively grainless as I recall. I can't remember how "sharp" it was as it was just before digital came in, but it looked fine to me with regard to resolution, i.e., there was a lot of detail on show and the film was out-performing the lens with regard to resolution.

There are aspects where digital is better than film and film is better than digital. Neither one is "better" than the other, as all these aspects meld into one to produce the final image. They're just different in many cases and these differences will be favoured by different photographers.

However, it was a pointless video as are all videos trying to determine which system has the better wotsit or the highest thingy. Film and digital can all produce resolution beyond that required by most people. What matters is the photograph itself, and the ability of the photographer to not just see the shot, but to take it too. Quite often the latest craze, such as high pixel count, high resolution, or high contrast, count for nothing. (Okay, personally I do like high resolution images generally, but you get my drift).


Cheers,
Duff.

 

[JeffK]

I've printed 13x19 from an 8mp camera (Rebel XT) and the print was fine. Ultimately, your print size determines your resolution requirements.

 

[CurtisJNeeleyJr]

Have any comparisons been done for making an image using medium format film and then outputting with ink using high resolution scans of the MF film? This; compared to printing on a Bessler 4x5 enlarger. The scans at hi-res can be too big for slower computers to manage. I have scanned 6x9 film and printed the image at 3’x4’ and doubt an enlarger could compete.

 

[Shashin]

He is basically skewing the conversation to a theoretical maximum if you were photographing high contrast lines. Resolution dramatically falls of with lower contrast subjects, like most things are, and when you are not photographing lines--a point target, like a line pair target but a grid of points, give far lower resolution. In a book I have from the 1980s estimates that a 35mm ektachrome slide would be a digital equivalent of 11MP, which means they were not using the resolving power for lines as the criteria.

But here is the thing, the subject you are photographing and its contrast will change the resolution of the imaging system. So when we say something like my camera makes a 50MP image, it is not actually the resolving power of the image. That is just the number of pixels on the sensor.

Another fun fact, the line pair test target was developed, not because it was the ideal way to describe the resolving power of an imaging system, it was developed because it was the easiest way for an human observer to estimate resolving power. That is why MTF is used because you can take the human observer out of the measurement. Unfortunately, the line pair method was kept because it give the highest numbers, not because it reflects the ability of a system in the real world--unless that word is populated with high-contrast vertical lines.

And one last thought about "detail" and "sharpness." Those are subjective qualities in an image and nothing to do with resolving power (his description of resolving power equaling sharpness is not correct). You can have a lower resolution image perceived as sharper and more detailed because of contrast. In fact, contrast and resolution are mutually exclusive--optimizing for resolution will impact contrast and vice versa.

Here is the best test for any image--if you look at it and it looks detailed and sharp, then it is, regardless for how much it resolves.