Dynamic range of the color red in RAW files.

[KarlKrueger]

I have seen this issue with my 70D and my SX-50 so it seems to be an issue with Canon cameras in general (I don't know if other camera makes have this too). Whenever I photograph something that has the color red (for me I often photograph birds) the jpegs look fine but I always start with post processing of RAW files. Anywhere the red color is found I find the color saturates out very easily. I use Adobe Photoshop Essentials (PSE) for post processing and cannot get the RAW adjusted/processed to look anywhere as good as the jpeg as detail is lost when red saturates out. This is not a problem with any other color.

For now I end up opening the jpeg as RAW in PSE and start my adjustments there which give me a better picture with finer detail than starting with the original RAW file. I will post this same question in the PSE forum to see who else noticed this and what their work around is.

[KarlKrueger]

I have tried to calibrate my monitor at home and things seemed set up quite well. It is a HDMI monitor and the differences in red saturation are quite visible on that monitor. That is where I do my photo editing/post processing.

At work I have a lousy monitor that seems to show all pictures a bit dark. Here it is harder to see the differences in the pictures I link above.

With my HDMI monitor I could make out feather detail in the red in the jpegs the camera produced of the red-winged blackbirds but not so well in the raw files unless I knock down vibrance which loses richness of the color. Lowering the red channel alone makes you lose red worse than reducing vibrance. Those were full resolution pics while the ones I post are 700x500 so that level of detail cannot be zoomed into. The saturation problem is much more evident with the Summer Tanager. Fortunately the camera jpeg and ability of PSE to open it as a camera raw saved that picture to acceptable standards.

[TCampbell]

RAW images have to be debayered -- the sensor is covered in "photo sites" and this is overlaid with a "bayer mask" which looks a bit like a checkerboard only it's a pattern of red, green, and blue squares. This is how a monochrome sensor is adapted to measure color.

Of course, when you zoom in on an image after importing it into the computer, you don't see those red, green, and blue color component squares... you just see a color image (each individual pixel is a blended color hue). This is because a software algorithm measures the the individual photo site as well as it's adjacent sites to interpret the color value for each "pixel" and the pixel is assigned a real color value (not just a component color value).

But this means that whatever software is used to open the image (and there are MANY algorithms for doing this) will interpet the data in it's own way. You can get a different look opening the software with Canon's software (Digital Photo Professional) vs. Adobe's software vs. Apple's software -- all for the same image.

[KarlKrueger]

Tcampbell,

Thank you for pointing this out. Now it makes sense to me why the camera derived jpeg is actually better than what PSE gives me.

Maybe I need to play with the DPP that came with my 70D as another alternative to feed something into PSE. As you see the open jpegs as a raw file in PSE is actually pretty handy. It saved me one critical day when somehow raw got turned off on the camera.

[KarlKrueger]

Just updating one more time here. I finally installed Canon's DPP and tried post processing the raw files I higlighted earlier. It seems DPP mimics what the camera does. As TCampbell pointed out this is probably because DPP and the 70D use a similar algorithm to debayerize the raw data. In other words the quality of the red I saw starting off in DPP looked very similar to the jpeg that the camera gives me. In fact I could not really improve the picture much between using DPP first or just taking the jpeg in PSE12 and opening in camera raw.

This probably explains what ronnienyc commented on saying he was able to do better post-processing in DPP versus an Adobe program when red was involved. This was probably due to the differential way Canon's software debayerizes the data from the adjacent photosites when compared to Adobe's algorithm.

In conclusion the red seems to saturate out more quickly on Canon cameras (at least on the 70D and SX-50 in my experience) but the quality of how that is processed depends on the software interpreting the raw file. When red is critical in the photo I will be opening the camera's jpeg in PSE as camera raw for post-processing. That seems to be the best work around for now.

[ronnienyc]

I'm glad DPP solved your issue. It's not the best converter, but it's good for some things 🙂