Does underexposure or overexposure lead to more details in the highlights, when using film?

[silvan schnelli]

From most of the sources I have read, it was always written than overexposing film (thicker negative) would lead to increased latitude in the highlights and underexposing film (thinner negative) would lead to increased latitude in the shadows (the opposite is true for digital which makes sense). However, I am currently reading Glen Kennels excellent book on "color and mastering for digital cinema" and am now a bit confused if what I have read regarding overexposing = more highlight detail and vice versa is true. Perhaps I must be misunderstanding something.

1) Now in this book he mentions that to obtain more details in the shadows, you would want to overexpose to obtain a heavy negative and then print down to extend the range into the shadows. This would lead to deeper blacks as he shows in a diagram, but as he also mentions, is that can uncover "details that might otherwise have been lost". As well as to underexpose the negative to assure that all the highlight information is obtained in the negative.

Perhaps I am confusing underexposing to ensure that all the highlight details are captured and overexposing to obtain a larger range of details in the highlights as the same thing, when in reality they are different. However, it seems like it would make much more sense to underexpose and then print up to obtain more details in the highlights. Furthermore, I don't understand how underexposing would lead to more details in the shadows  compared to as is mentioned in the book, overexposing and then printing down. 

I hope someone could maybe help me clear up some of my confusions regarding this.

2) Another point I would like to mention is that overexposing the negative is said to lead to more contrast; however, when I'm looking at the curve of the color print film, the gamma of the curve seems to be the same, which is theoretically the definition of contrast in projected print. Although I thought that perhaps the reasoning is that we now have a larger difference between the minimum and maximum amount of density on the color film print. Is this correct?

 

[Karim D. Ghantous]

Others can speak with more authority on this. But, in sum:

- You do not want to underexpose a negative. You will, theoretically, protect your highlights, but at a huge cost to lost detail in the deeper shadows. Negatives like a bit overexposure if you don't need all the highlight headroom.

- Exposure determines the shadows, development determines the highlights. So, underexposing and push processing increases contrast. And, overexposing and pull processing decreases contrast.

- Negatives have plenty of highlight headroom, in general, so some overexposure is arguably ideal.

This is from 2011:

[David Mullen ASC]

Overexposing doesn't increase latitude or information in the highlights. Basically film has more latitude in the highlights compared to the shadows, which is why Ansel Adams used to say "expose for the shadows and print for the highlights."  In other words, the gains in improved shadow detail and tighter grain patterns is worth some minor loss of bright highlight detail because film can handle a lot of overexposure.  Unlike digital, film responds to light on a log curve, so as exposure increases, density stops increasing equally, it falls off -- in other words, the contrast starts to flatten out.

[silvan schnelli]

@David Mullen ASCThank you a lot for your response. However, if we ignore for a second that it would be mostly be disadvantageous to underexpose film, is it reasonable to say that film works in many ways similar to digital then? You underexpose for more highlight latitude and overexpose for more shadow latitude. I feel like a changing a digital sensors EI and then exposing accordingly shares many parallels to rating a film stock at a different EI and then printing up or down. It seems to me that the EI on a camera basically encompasses that mechanism, as to my knowledge EI for digital sensors is similar to an “exposure LUT”. I was wondering what your thoughts were on this.

I also am confused about the tighter grain patterns by overexposing, doesn’t that lead to more grain, because we expose more silver halide crystals or is it cause we print it down? Or does overexposing lead to less shot noise like with digital sensors?
 

And lastly I thought that digital cameras responded similar to film as the log curves are based on the original Cineon scans of negative color film. Although it confuses me why, for example LogC4 didn’t  roll off at the top.

Edited October 24, 2023 by silvan schnelli

[silvan schnelli]

@Karim D. Ghantous  Thank you for the response and the diagram. It’s interesting to see the differences in their over- and under- exposure latitudes, between film and digital sensors.

[David Mullen ASC]

The reason why with color negative film (which has dye clouds technically, not grains, once processed, unless you skip the bleach step and leave silver grains in) looks less grainy when mildly overexposed is that in normal exposure, the smaller (slower) grains in darkest areas don't get any exposure so those grains are washed away, and the darker detail is captured on the larger (faster) grains.  So overexposing allows more of the slow, smaller grains to stay in the image, filling in the gaps between the larger, faster grains. The image isn't less grainy but it feels less grainy because the grain structure is tighter. Technically this is true for b&w film as well but since silver grains are sharper-edged than color dye clouds, some people feel that overexposure, creating more silver, makes the image feel grainier.

Now when you through digital scanning into the mix, there can be an issue that overexposed (denser) negative can create noise on the scanner from having to push the signal through the densist areas (the highlights, like bright skies).

Yes, film captures a range from shadow to highlight, so overexposure shifts that range to favor shadows and underexposure shifts that range to favor the highlights. But since the film's response to light is non-linear, to have to factor in the way things look playing on the shoulder and toe of the characteristic curve.

[silvan schnelli]

@David Mullen ASC Ah I see very interesting. Thank you a lot for the reply. I also went back to reread the Wikipedia on photographic processing and Kodaks essential reference guide for filmmakers and the whole process and way film stock is structured is much clearer now.