ClarkVision.com

To Clarkvision.com Galleries
Home Galleries Articles Reviews Best Gear Science New About Contact

Digital Camera Raw Converter Shadow Detail and Image Editor Limitations:
Factors in Getting Shadow Detail in Images

by Roger N. Clark

All images, text and data on this site are copyrighted.
They may not be used except by written permission from Roger N. Clark.
All rights reserved.


There are a lot of differences between digital camera raw converters, but no discussion I have seen discuss the digital workflow of these system along with the image editor in use and shadow detail. This page discusses effects of using different systems on information in darker parts of the image.


Figure 1. Shadow detail from two raw converters: an astounding difference, not only regarding noise, but color! The image at right is closer to the correct view based on visual observations at the time the image was obtained.

The test is a sunset scene, I originally did to test dynamic range. The image used on this page is from a Canon 1D Mark II camera, recorded as a raw and a jpeg image. The exposure was set so the brightest highlights were just under saturation. Raw converters used were: 1) Photoshop CS2 with ACR 2.4, 2) RawShooter Essentials 2005, and 3) ImagesPlus 2.5. The following shows what is observed.
Figure 2. Photoshop CS2 raw conversion using camera setting (cloudy) white balance. Color space: Adobe RGB (1998), converted to sRGB for web display.
Figure 3. RawShooter Essentials 2005 raw conversion. Color space: Adobe RGB (1998), converted to sRGB for web display.
Figure 4. ImagesPlus 2.5 raw conversion (IPs closest white balance is daylight). Color space: Adobe RGB, converted to sRGB for web display.
Figure 5. Photoshop CS2 raw conversion using sun white balance. Color space: Adobe RGB (1998), converted to sRGB for web display.
Figure 6. The shadow area studied is shown in the red box. The scene intensities are shown in linear data number (LDN) and in stops. The grassy area in the red box is -5.1 stops, the banks of the lake are -7.6 stops, and the sky around the sun is +2.0 stops, for a total range of 9.6 stops. Image saturation would be +2.45 stops (LDN=65535) for this exposure.

Figure 7. This is the cropped section at 100% scale, no additional processing. Cropping done in Photoshop CS2.
Figure 8. After conversion by Photoshop CS2 then processed in Photoshop CS2: the image, adjust, highlight/shadow tool was used to brighten the shadows using 50% in the shadow slider. 16-bit processing was done at all steps. Without other testing, one might conclude the noise was due to the camera.
Figure 9. After conversion by RawShooter Essentials 2005 then processed in Photoshop CS2: the image, adjust, highlight/shadow tool was used to brighten the shadows using 50% in the shadow slider. 16-bit processing was done at all steps. Without other testing, one might conclude the noise was due to the camera. The result is similar to the Photoshop result, but there is less color.
Figure 10. After conversion by ImagesPlus 2.5 then processed in Photoshop CS2: the image, adjust, highlight/shadow tool was used to brighten the shadows using 50%, then 25% in the shadow slider. 16-bit processing was done at all steps.
Figure 11. All processing in ImagesPlus 2.5. After conversion by ImagesPlus 2.5 then processed in ImagesPlus using levels: mid slider reduced to 3000. IP is doing 64-bit floating point processing.
Figure 12. After conversion by ImagesPlus 2.5 then processed in ImagesPlus using levels: mid slider reduced to 3000. Then the image was written to a 16-bit tif file, opened in Photoshop CS2 and the color was adjusted.

8-Bit Versus 16-Bit

Now that we've seen some issues with 16-bit raw converters, what happens if we do 8-bit processing? One might expect it to be worse. Figure 13 shows the Photoshop CS2 raw conversion as in Figure 2 then processed as in Figure 8. Figure 14 shows the in camera generated jpeg.
Figure 13. After conversion by Photoshop CS2 and written as an 8-bit tif file, the image was then processed in Photoshop CS2 using image, adjust, highlight/shadow tool which was used to brighten the shadows using 50% in the shadow slider. 8-bit processing was done at all steps. Without other testing, one might conclude the noise was due to the camera as this image looks almost identical to the same processing in 16-bits in Figure 8.
Figure 14. The in-camera generated jpeg had only 8-bit processing in Photoshop CS2. The highlight/shadow tool was used to brighten the shadows using 50% in the shadow slider. The image was still a little dark compared to that in Figure 13, so levels was used and the middle slider moved to 1.50. To me, the 8-bit jpeg looks better than either the 8-bit or 16-bit tif (Figures 13, 8) in this shadow area!

The results above imply that 8-bit versus 16-bit editing in photoshop do not show much difference. There has been quite a controversy going on for a while, which I was not previously aware of. See http://mike.russell-home.net/tmp/DanMargulis16Bit/. Dan Margulis has challenged anyone to show an image where 16-bit versus 8-bit editing made a large difference. Perhaps the results above illustrate why differences between 8 and 16-bit editing have yet to win Dan Margulis' challenge.

Photoshop does 15-bit processing and I believe that is causing some issues with the "16-bit" processed data. I've been experimenting doing more and more processing in ImagesPlus as it does scientifically correct math (64-bit processing, but I don't believe that 64 or even 32 bits are really warranted-it's more the math).

I think this experiment shows that the main reason people have been having trouble with Margulis' challenge is not that 16-bits is not better, but the unusual way photoshop processes data.

I do the major processing steps in ImagesPlus then switch to photoshop for minor tweaks and color management and ICC profiles for output. I did that on this image: Image: The Kiss, Great Blue Herons at Sunset #8149. and was amazed how simple my work flow became and that I had so little work to do in noise suppression of the background (none, when I normally have had to do a lot). I need to go back and work the image using only photoshop and show the resulting difference, which I believe will be significant. I believe that many images do not need 16-bits, and you can produce good images with 8-bits. I have. In fact most of my images are 8-bit, even 8-bit jpegs! I have placed in Natures Best international competitions with 8-bit jpegs. Most of my bear images are 8-bit jpegs, and I make 16x24 inch prints from them: http://www.clarkvision.com/galleries/gallery.bear.

But now that I've found the differences in photoshop's 15-bit math versus real image processing, I find I can produce better images from true 16+ bit processing. However, few of my images require high precision over a large dynamic range. So the vast majority only need 8-bit processing, or photoshop 15-bit processing, or even 8-bit jpegs. For example, the Kissing Herons image, above, produces a similarly great print when processed as an 8-bit tif with all processing in Photoshop (from raw conversion to final output).

True dynamic range of modern DSLR cameras is shown on these two pages (and greater than 11 stops is common on better DSLRs; the 1D Mark II gets 11.6 stops):

Procedures for Evaluating Digital Camera Noise and Full Well Capacities; Canon 1D Mark II Analysis.

The Signal-to-Noise of Digital Camera images and Comparison to Film.

What the data one the above pages show is that digital cameras have a huge dynamic range. But what we see on this page is that some raw converters lose shadow detail, increasing noise and losing color. That might make some conclude dynamic range is lower than reality, noise is higher, and draw erroneous conclusions about color fidelity if those raw converters are used in their evaluations. Whether you need such large dynamic range and high-bit processing, depends on the image and the result you want to achieve.




Conclusions

There are large differences between raw converters and the noise and shadow detail they get in the conversion of digital camera files. Photoshop does 15-bit math to speed operations, and it appears it is limiting the noise and dynamic range of current DSLR cameras. Promoting the raw converted file to 32 bits in Photoshop CS2 does not change the noise, and most editing functions do not work in Photoshop CS2 in 32 bit mode. ImagesPlus 2.5 (and later), produces the best shadow detail, but lacks flexibility with color balance. The raw converter also lacks sharpness requiring the use of sharpening tools like Richardson-Lucy image reconstruction. See my Digital Workflow for more details. But, in the end, few images, at least in the style of my photography, need high dynamic range plus high bit-precision. Sunrise/sunsets and astrophoto images are typical exceptions. The sunset image on this page would only benefit from the high bit precision if you wanted to bring out shadow detail. Personally, I like the image better with the area used as a test as a silhouette, so in my opinion for this particular image, the high bits are not needed! But I do like the ability to make further corrections to images at a later stage, so 16-bit files maintain the precision to allow that work without having to start from scratch.


Figure 15. Summary comparison of the extremes of shadow detail. On the left is the Photoshop CS2 converter and Photoshop 16-bit (really 15-bit) processing. On the right is the ImagesPlus 2.5 converter and 64-bit floating point processing.




Notes

I have no affiliation to ImagesPlus, Photoshop, or any other raw converter manufacturer, nor any image processing commercial vendor. I do scientific imaging and do write my own scientific software, which is publicly available at no charge (but is not designed for photography; see my bio on my home page).


Home Galleries Articles Reviews Best Gear Science New About Contact

http://www.clarkvision.com/articles/raw.converter.shadow.detail

First Published January, 15, 2006.
Last updated April 16, 2006.