This series of blog entries are a result of someone on Flickr pointing out that my in-camera processed BW wasn't as good as RAW work processed on a Big Computer.
In processing RAW images I've had a good feel for what makes a decent image. Raise the mid-tones and use the Lightness slider to control the overall distribution of tones. What I'm about here is something ever so slightly different. Greater understanding. Greater control.
This started off as an exercise to better understand the difference between in-camera and off-camera results. For this I needed a framework of understanding to work within. There are several valid ways to approach this and they involve measuring things.
I chose the Zone System because I can measure and see things in a format I am familiar with. Importantly, I can accurately measure luminance using Open Source Software image processing software that I regularly use (ie: daily) and am familiar with (RawTherapee, the Gimp).
To this point in trying to get my arms around what's going on I've gotten somewhat side-tracked into the details of how Sony RAW files are loaded into RawTherapee and the settings/functions which express the widest possible dynamic range. I'll try and wrap up this series soon by looking at in-camera jpg processing. I promise.
Before I move on, I'd like to share a few potentially useful details.
Perhaps it's already understood, but I'd like to stress that these efforts are simply a preamble to making beautiful images. There is nothing magic, though it might at first feel scary to take full control of the image processing sequence.
Process Details ~
- RawTherapee is the most flexible software for processing RAW images that I've thus far found. It allows us to take as much control of image processing as we can handle. Here are the first two details I used in making these studies.
- Demosaic algorithms: RAW -> Demosaicing ->
- AMaZe (for low ISO)
- LMMSE (for high ISO)
- Camera profile: Color -> Color Management -> Input Profile ->
- No Profile ... or...
- Auto-Matched Camera Profile with the following _not_ selected
- Tone Curve
- Base
- Look
- For one-click image processing setup after image import into RawTherapee here is the full recipe I use
- Exposure -> Tone Curve 1 -> Luminance Curves -> Input Correction Curve
- from exposure stepwedge Correction Curves choose one of the following...
- Linear Zone System
- Film Emulation
- Color -> Black and White -> Luminance Equalizer
- Color -> Color Management -> Input Profile -> No Profile (my current preference)
- Transform -> Profiled Lens Correction -> Automatically Selected
- Transform -> Correct ->
- Geometric Distortion
- Chromatic Aberration
- RAW -> Demosaicing -> select one of the following
- AMaZe (for low ISO)
- LMMSE (for high ISO)
- RAW -> Capture Sharpening (for low ISO)
- Save Current Profile with an appropriate name, for later use. I've built as many of these recipes as I feel I need. For instance, there is a Profile for low ISO film emulation, another for low ISO linear Zone System, and so on.
- Looking under Exposure -> Tone Curve I see there are actually two Curves I can use. For what I'm illustrating here I put the Input Correction Curve in Tone Curve 1. Then I use Tone Curve 2 to control the end points of the tonal range. Tone Curve 2 can also be used to move Zones from one EV to another and to manage image contrast.
- Concerning Camera Profiles, the two mentioned here, No Profile and Auto-Matched Camera Profile, are both useful starting points. There is no data conditioning when using No Profile. I see there is a little data conditioning with Auto-Matched Camera Profile. This gently changes the tonal values. What I want is as linear, as consistent a starting point for image processing as possible. That way when I make my measurements which ultimately are used to create the Input Correction Curve I can save the proper values in the saved Current Profile recipe.
- When I shot film I used a Pentax 1degree Spot Meter. I could accurately measure scene values and place them on the Zone System tone curve where I wanted them. The meter has long since been sold.
Many cameras these days come with a Spot Meter function built-in. The circle of metering is commonly larger than with the old stand-alone Pentax, but I don't find it all that important. If I were wedded to the old process, I'd likely look for another Spot Meter. But...
- In-camera digital metering systems are very sophisticated compared to light meters I used in the film days. The Sony metering system allows me to see when highlights are clipped and when shadows drop out. In general I don't use this feature. There is plenty of dynamic range for the situations I find myself photographing in.
The metering system does a good job distributing tonal values with information in each EV step. Further, dropping the Zone 5 EV to -1 does an excellent job of protecting the highlight regions.
This is why I've gotten lazy in the traditional sense. I would rather take a photo than stand there metering everything to perfection. This is why I set the EV and let the camera make the metering decisions and call it good enough. If I'm really in doubt, I bracket the exposure.
- Which leads to a comment about why I feel comfortable using -1EV as the Zone 5 middle gray, and not 0EV. Remembering that using 0EV on my Sony cameras crowds the light tones, and remembering that -1EV more evenly spreads the tones between light and dark across the linear-ized Zone System, there was a concern about shadow noise.
In the early days of digital I owned a number of Canon DSLRs. They _all_ suffered from noisy shadows, even when Zone 5 was shot at EV0. Sony, on the other hand, showed me what is possible with their now 10 year old A6000 APS-C. Sony is of course known to build "quiet" sensors. That is to say the dark regions contain less noise than sensors from other manufacturers. As manufacturers improve their design and manufacturing processes this might change.
- In any event, if shadow noise is too great, RawTherapee has a local Selective Editing local adjustment where dark tone noise can be managed.
- Once an image is selected and opened using one of the film emulation or linear Zone System curves, the true fun begins: interpreting the image.
I commonly use standard image processing tools ->
- Exposure -> Lightness, which respects the curve end-points
- Exposure -> Contrast
- Exposure -> Tone Equalizer with sliders controlling...
- Blacks
- Shadows
- Mid-Tones
- Highlights
- White
- Exposure -> Vignette Filter -> Strength: -0.30 to -0.90 (to taste)
Observations ~
- In this exercise I've built two different styles of Input Correction Curves, one for Film Emulation and another for correct to the Zone System definition linear Zone System curve.
The Film Emulation Input Correction Curve was built to illustrate how RAW files could be processed to match film. Looking at the old film luminance curves I know what the Output Curve from digital RAW processing should look like and make it so. Such is the flexibility of digital image processing.
As I've noted elsewhere,
the original Zone System definition specified 1EV steps from Zone 0 to
Zone 10. The problem with film were the toe and highlight regions. The
toe rolled on and the highlights rolled off. Some film photographers
did a lot of image processing manipulations to ensure their images put
valued information on the linear portion of the curve. This commonly narrowed dynamic range and the Zone tones were spread over a shorter exposure range.
In digital work I've found the 1EV per Zone challenge much easier to manage. Using a synthesized step wedge where tonal values exactly match the 1EV per Zone definition, I'm able to build fully correct Input Correction Curves and retain 11EV to 12EV dynamic range.
- The reason I select Color -> Black and White -> Luminance Equalizer in black and white conversion is to get as much tonal separation as possible. Since the Zone System strives for as much tonal separation as possible, I feel that luminance adds just a bit more of a good thing. This is an area of improvement digital provides over what silver halide film delivers (which is mainly simple de-saturation).
- If Zone 5 has been exposed at 0EV, as the vast majority of
my work has been, truth be known, there ways to still take advantage of good tonal separation. That is, accepting Zone 9 as pure white in
place of Zone 10 can still yield excellent results. As I mentioned
earlier the lights will be compressed above Zone 5 by one EV. However,
if I don't like the steep slope from Zone 8 to Zone 9, I can always
violate the Zone System definition of Zones 5, 6, 7, and 8 by gently raising their
values, and then use the Exposure -> Lightness slider to adjust the overall tonality of the image into something rather pleasing.
- I realize that I have to be discerning when looking for useful information. YouTube videos on the digital application of the Zone System don't take into account that +4EV, Zone 9 in the traditional sense, is the digital top end of the scale, not Zone 10 as they're describing it. I've found the EV shift to be important, particularly in normal to high contrast situations, and I've learned to take what I've seen on YouTube with a grain of salt.
- I also have to be discerning and intellectually critical when strolling around the web looking at "testing" assessments. One popular tester's assessment of digital adaptation of the Zone System, in the end, has been less than helpful. Their conclusion that there's only 7EV of usable information in digital Zone System application does not match my experience. I'm drawing a very different set of conclusions, and understand I could've been more cautious when I thought this particular tester was providing a properly researched conclusion. For several years I felt that a digital Zone System was not possible based on his "testing."
- It sometimes pays to be skeptical, to have an attitude of curiosity toward details, to talk with others, and to not take everything/anything on the 'net at face value. This is why, if I've done a decent enough job here, anyone who's interested in such things can do the very things I have for themselves to see what is true and what is not.
Testing is not in and of itself the end. Testing can lead to understanding, which might lead to better control. In the end, making good photographs is the really only important goal.
No comments:
Post a Comment