LFI: A different kind of process

[Alberti]

Advertisement (gone after registration)

In LFI 2/10, Michael Hussmann takes us through some observations about the linear nature of digital sensors vs the logarithmic nature (or compressed nature) of silver halide films.

 

On page 44 a graph is shown with a solution: underexpose the digital image (expose on the highlights), and develop for the shadows.

 

It would be nice if Michael would

1. share the corresponding curve used in C1 (and other developers)
   
2. give an indication of the underexposure: for instance 2/3 stop underexp
   

 

Of course using an incident light-meter (such as Lunasix) is imho the best solution, but I almost never take it along, just lazily taking a point and shoot attitude (aim at a mix of highlights and hold down the release knob).

The discussed method looks much more promising.

 

Who supports this idea? I expect we can import such a curve

alberti

[Marquinius]

Euhmmmmm ... I haven't read it, but I've got some questions regarding exposing to the left (shadows) for digital cameras. As far as I understand, half of all the information goes on the first available track (1 of 8) of the chip, half of the next batch (so one quarter) goes on the next, etc etc. That would point toward exposing to the of the chip RIGHT (highlights). I mean: why throw away half or more of your available information carrier? In other words: one should expose so the max of your chip (SD card) is used.

 

But I'm sure I missed the pont of the story in LFI.

[jaapv]

Yes, the point is different in pointing out that in reality digital has a larger dynamic range than film, but digital photographers handicap themselves by exposing a digital image identically to a film capture.

 

A similar argument can be found here:

 

 

Clarkvision: Dynamic Range of an Image

[thompsonkirk]

Like Marco, I don't have a copy of the article. But it makes sense to place highlights in what we used to call Zone VII – which means 'expose to the right, but not all the way.' In other words,

– get as much data as you can, in the shadow zones; but

– be even more careful to keep your highlights from going 'over the top.'

 

'Developing for the shadows' would mean, when converting your DNG files,

– using the Blacks slider to come up & meet the histogram in LR/ACR (or whatever you use);

– or (more often) using the Fill slider to open up the shadows. But you have to be careful not to overdo this, because as you differentiate the tones in the darker areas (where there's less data), you make noise more visible.

 

There's no 'one size fits all' Curve that would do this properly for all images. Their histograms are as different as human body types, & you need to tailor individually to each one.

 

Using an incident meter would be, I believe, no help at all. What you want do do is check the camera's Histogram. An incident meter is a perfect Averaging meter; it can never tell you where your shadows & highlights fall.

 

If you want a rule-of-thumb for preserving good highlights, with M8 expose -2/3 in harsher light, -1/3 otherwise; with M9, -1/3. Then check the Histogram early in your shooting session & season to taste.

 

Kirk

[Alberti]

Jaap,

I do agree I shortened (and probably mispresented) the story line.

The concept as I see it is quite different though from what Kirk says. I have done that in the past as an experiment and found that this still does not guarantee a good match.

 

By taking a -1/3 and then afterwards sometime increasing the exposure again (depending in the situation) is no standard way for me. For instance, a shiny skin in a face (there is some cream on it) has a complex micro structure that easily has small blown highlights, or the more traditional floodlights, that make post-processing of the image difficult.

 

The issue: we are in both cases looking at a linear 'curve'.

 

What Micheal says is that by underexposure and a non-linear treatment of the result, we could get a exposure line (is that a good word?) that looks more like film.

 

In fact, probably the M8 16bit/8bit/16bit conversion adds a little bit of this effect already.

 

The diagram that Micheal has, does not have any scale. It could refer as well to the small differences like in the compression/decompression cycle as to a more sophisticated one, intended to mimic film even more. I just don't have the time, stamina, nor know-how to deal with the issue. Like C1 has a button to increase the shadows, decrease the mid-tones without in general degrading the colour texture, I am looking for a similar profile that does what Micheal talks about.

Of course I tried it myself but was afraid that I (on screen) had made something that had influence the tonality or on the light density of the image.

 

Talking about exposing/developing isn't too strange isn't it?

 

Alberti

[roguewave]

My two cents is based on Kirk's reading of the zone system. It's always easier to get back details when under exposed that having no detail when you over expose. The real issue is that most digital users don't think like film photographers. The initial exposure is critical, but the real magic is in the "darkroom", whether it's in PS, C1 or chemicals & light. The never ending search for an algorithm that achieves perfection out of the camera is an illusion. Great image making encompasses at lot more than getting the correct exposure,