Undoing Adobe's Raw Conversion hijinx - understanding your camera's baseline exposure compensation

https://photographylif​e.com …ine-exposure-compensation

This article is really important, even if you like the look Adobe's raw conversion gives you right out of the box. It is important because it demonstrates how to understand your raw conversion and what you can do to control it and get the most out of your data. There is a lot of manipulation of your data going on behind the scenes that might, unfortunately, make you change exposure settings on your camera to suit a particular raw converter software versus acquiring your raw data in an optimized way that is converter-agnostic.

This applies to other converters as well, but Adobe Camera Raw (and Lightroom) is probably the most ubiquitous and is known to do a whole lot of automagic things to your raw data, and the article uses ACR as the example case study.

The article highlights the use of Raw Digger, which is a really powerful tool for exploring, analyzing and understanding your raw data.

enjoy!

kirk

That article left me confused Kirk as does the whole exposure compensation issue.

My workaround is to shoot in manual mode, adjusting exposure in live view using the exposed histogram (ETTR)
after neutralizing all the picture style settings to the left to compensate for the jpeg image on the LCD.

It is about exposure "compensation" that is being applied by the raw converter, without your control, to make your images "look better" during raw conversion. It may give people the impression that, based on the appearance of their images in the raw converter, they are overexposing in-camera when in fact they are underexposing in-camera - these potential erroneous exposure choices are based on how their images appear in ACR or LR with the "Exposure" slider set at zero, versus understanding the relationship between the camera's meter and the actual exposure based on the raw data and the tolerance to highlight clipping of raw pixel values.

For Canon users who have supported cameras, the Magic Lantern raw histogram in Live View is a great way to visualize the raw data in the field. If you cannot run Magic Lantern, then it is worth going through the exercise of shooting a controlled scene and working out how the histogram on your camera reflects the actual raw data that gets captured. Once you understand how to expose for the raw file, you can explore how various raw converters mess with your data to make your images "look nice" - if you prefer not to have this control taken from you, then you can develop strategies to undo those behind-the-scenes manipulations in raw conversion and start with a working image the way you want it.

kirk

Interesting article thanks. I have to say that I rather like the Adobe PV2012 "Secret Sauce", but there is one thing better than just eating secret sauce, and that's knowing the recipe.

Alan

In reference to Kirk's second post, I have already used RAW Digger to work out my required metering offsets for my camera to get a good ETTR based RAW exposure, so I am not relying on the LR histogram for calculating my exposures. I have ML on my 50D, and have the RAW histogram switched on, but on the 50D the way it is displayed is so small that is is actually just about useless. I do know what a good exposures histogram should look like on the camera LCD with my usual settings, again thanks to doing tests with RAW Digger. I really like RAW Digger as it allows you to effectively check your camera's metering system against the actual data recorded by the sensor. Then you can calculate some suitable offsets to apply to your images to get the optimum exposure.

Alan

BigAl007 wrote in post #17893216
.... I do know what a good exposures histogram should look like on the camera LCD with my usual settings, again thanks to doing tests with RAW Digger. I really like RAW Digger as it allows you to effectively check your camera's metering system against the actual data recorded by the sensor. Then you can calculate some suitable offsets to apply to your images to get the optimum exposure.

Alan

Exactly!

kirk

On the issue of ETTR, or knowing how far to expose in-camera, I have set a 'uni white balance' preset (which affects the multipliers used to generate the color balance in the JPG), and have turned down sharpening and contrast in Picture Style. I need to explore the 'Picture Style' settings, too. All of these affect the JPG processing, which the histogram is generated from.

It is helpful to eliminate the camera's JPEG/Picture Style processing and white balancing from the equation all together (i.e., do not even use the camera histogram at all, except for a loose guide at best). I concentrate on trying to understand the relationship between what my spot meter is reading and what that means in my raw file. See this exercise from Raw Digger:

http://www.rawdigger.c​om …-to-improve-dynamic-range

Basically, when your meter tells you "X" for a particular spot area, you know (because you have tested and analyzed your raw files) that the raw file can actually accommodate "X+something" exposure before clipping occurs in that spot area. I used to do the UniWB custom WB, low contrast picture style thing. It does not work reliably and results in the typical green UniWB JPEGs that are useless for assessing things in the field like focus and white balance. Now with the Magic Lantern raw histogram you can simultaneously do the above exercise with the raw histogram activated and check that versus the full Raw Digger analysis and see how accurate the ML histogram is as well.

kirk

Heh, I've been saying for quite a while that 'the histogram' is useless. Not sold on ETTR either. Hand-held light meter and/or its 1° spot reflective meter aimed at a mid-tone (mid grey, bright red, &c): perfect exposure 99% of the time. Less post processing time and no techno mumbo jumbo that makes one head spin while still leaving one completely clueless and poorly exposing.

Alveric wrote in post #17895631
Heh, I've been saying for quite a while that 'the histogram' is useless. Not sold on ETTR either. Hand-held light meter and/or its 1° spot reflective meter aimed at a mid-tone (mid grey, bright red, &c): perfect exposure 99% of the time. Less post processing time and no techno mumbo jumbo that makes one head spin while still leaving one completely clueless and poorly exposing.

Oh it's great to hear someone else who like to use a separate lightmeter, personally I prefer to use an incident meter, as it's really hard to spot meter an aircraft that is moving at a couple of hundred mph and is maybe a total of only 1 degree wide anyway. As long as I am in the same light the incident meter is just so much easier. I also like the analogue type meters, where you can instantly see every shutter/aperture combination available for a given ISO. I will be getting a Sekonic L308 before the next air show season starts, a traditional selenium cell incident lightmeter. Although the L208 model is a lot smaller and only about £90, I'm going for the larger unit as I won't need to use my reading glasses to use it, well not yet anyway. There are just so many times when an incident light reading so much better than reflected light readings, although a good tight spot meter is also really useful too, as you can measure subject DR with one.

I would still use RAW Digger to check the calibration of the particular meter to the particular camera's results, something that is so much easier to do that it would be with film.

Alan

I tried to follow the article, but admit that it is difficult and would require more time to understand.

But I found it curious that it does not discuss camera profiles. Only in one place do I find mention, when step 3.3 (1) says to set the profile to Adobe Standard.

Camera profiles contain a "base tone curve" that does a lot of "automatic things to your raw data." The base tone curve in every Adobe Standard profile I've inspected is close to (but not an exact match) to a basic gamma 2.2 curve. The base tone curve in other Adobe Camera profiles varies. On most, it's the same as Adobe Standard. But many have a different base tone curve.

In my limited experience I've found differences between camera models of the same manufacture. Example, Camera Faithful for a Canon 1Ds is not the same as a Canon 5D1, 2 or 3. And the base tone curves also vary across manufacturers (Canon vs. Nikon).

Raw data is in linear form, and a tone curve is needed to make the image match human perception. But to see what's really in your raw data exposures before Adobe or anyone else applies "secret sauce", you would need to have a profile with a linear tone curve. That's fairly easy to create with the DNG profile editor. DNG PE is also the only tool I know of that will show the base tone curve of a profile.

I keep a couple "linear" profiles loaded for temporary checks on exposure, to see if any values are really clipped in raw. RawDigger is a better way, of course, but a linear profile provides a handy quick check inside ACR/LR.

> a tone curve is needed to make the image match human perception.

The scene we see in Nature is linear. Why would we need a tone curve for a photo to match the perception?

Illiah,

Much better authors than I can explain the gamma encoding needed to adjust digital images to match human vision.
Here is a good one: http://www.cambridgein​colour.com …ials/gamma-correction.htm

The question is - where the gamma is applied to a natural scene when we are looking at that natural scene. The answer is - in our perception. That means, any gamma in a file needs to be inverted and the resulting gamma for presentation needs to be close to linear, if, of cause, we are after faithful reproduction of the original scene - because our eyes/brain are expecting the linear image, both in the original scene and in its reproduction - same pair of eyes in both cases, same brain. In other words, we need to eliminate the gamma curve, as the gamma curve that matches the Nature is linear

Can you please explain how linearizing the profile would indicate possible highlight clipping in the original Raw (prior to the application of the profile) if:
A. It is the intermediate tones that you are refraining from gamma altering but the white point remains the same.
B. No matter what the base TRC in the profile, Camera Raw is applying secondary "content aware/adaptive" curves in P.V. 2012 unless you are cancelling them (so far as it is possible) by setting Exposure to -[1 + the camera model exposure bias].
C. White balancing is still being done, thus inflating primarily red and blue channel values and occasionally the green as well.

The histogram/numerical readouts in ACR/LR and other main-stream Raw converters (C1, DxO) are predictive of the results of whatever processing the software will do by default, plus any alterations or additions you may care to make, but they are no more representative of the capture data than the camera lcd.