Canon 1D MkIV images too dark

Question

Hi there
Having a problem here. The RAW images from my Canon 1D Mark IV come in a stop too dark!
When I open them, they lookfine; a second later they switch to a too-dark version. I am using Camera Standard, and no other adjustments.
It's not just visual: the histogram of the opened file is shifter a stop to the left compared to the same phoito's histogram on the camera.
Any ideas? This is a major issue for me since I rely on both the 1D MkIV and Lightroom to make a living...
Many thanks
Michael
www.michaelwillems.ca

Answer 1

Hi M, the question is simple, but the answer is complex. I'll provide an example to try to illustrate. The attached image is a downsampled version of a JPEG from the 1D Mark IV. It's a test picture of a color chart. I've drawn a red circle around the 2nd gray patch in the 4th row. When I take a picture with the 1D Mark IV zoomed into this patch at this exposure, I get a histogram all the way to the right. And indeed, when I open the file in any image editor (e.g., Photoshop, or DPP), and use a color picker / eyedropper to sample the values, the average value of that patch is (255,255,255). So this is like in your example when you say you look at the histogram (e.g., in DPP or on the camera LCD) and it's all the way to the right. In other words, the histogram effectively says this patch is clipped to the max value.
However, if we look at the average raw values in this example for that patch, they are actually (on an 8-bit scale) about (72, 209, 187). In other words, the actual camera-recorded raw values are still pretty far away from clipping to 255. (I should also mention that the camera-recorded raw values aren't white-balanced, which is why the green value is much higher than the red & blue values, above.)
So you can see there is a discrepancy between the camera-produced JPEG and the actual raw values. In this example, the camera histogram and JPEG values indicate a clipped patch at (255,255,255). If you are going by that histogram, then you would conclude reasonably that this patch is overexposed and clipped -- when actually it is not.
More generally, there are always discrepancies / differences between camera-recorded raw values and the histogram shown on the back of the camera (as well as the JPEGs rendered by the camera). This isn't error. This is preference. Cameras and software apply "preset" exposure and tone curve processing (and possibly sharpening, noise reduction, etc.), and the histograms you see on the back of the camera are based on those presets (not the actual raw data). This is why your histogram on the camera will change depending on your in-camera settings of white balance, tone curve, contrast, etc.
There is nothing wrong with the 1D Mark IV.  Many Canon DSLRs currently leave a lot of highlight headroom, i.e., when metering they will be conservative on the highlight side, and expose a little less. As I recall (don't have my full notes in front of me), the 1D IV spot meters about 3.5 stops below the actual clipping point. In other words, if I had spot-metered that gray patch in the attached example, and started with the meter needle in the middle, I could then increase the exposure by about 3.5 stops before I actually clip that patch. That's quite a lot.
To summarize: you have to be careful when interpreting camera histograms. In many cases, it is easy to interpret that something is clipped when in fact it is not. If you see a histogram that appears bunched up to the right, the tendency is to decrease the exposure, and therefore make the histogram move to the left (so you aren't overexposing). In reality it is possible the original exposure was fine (without overexposure), and all that's happened is that now the image is even more underexposed.