Dynamic Range and Colour Depth on Canons - any tips for improving?

OK, I finally did a test of Canon's highlight tone protection (HTP). The bottom line is that there is virtually no disadvantage in using HTP. Properly used, it does not cause any increase in noise. Except for decreasing the exposure to preserve highlight detail, it has no effect on the raw files, but it does increase the dynamic range of the jpeg files.

To test HTP on my SL1 I took two photos of a newspaper with a desk lamp aimed at one end, so there is definitely an extreme range of brightness. Both photos were taken with the same f stop and shutter speed, so the exposures are identical. One was set at ISO 400 with HTP; one was set at ISO 200 without HTP. (Remember that when the camera is set at ISO 400 with HTP it's actually shooting at ISO 200. That's why you can't select ISO 100 with HTP, because the camera can't shoot at ISO 50.)

The raw files are virtually identical, and the noise in the deep shadows is identical. Thus, HTP had no effect on raw files.

The jpeg files show very similar highlights, but the mid tones and shadows are considerably brightened (by one stop) in the HTP photo. The deep shadows show higher noise in the non-HTP photo. Visually, it is possible to distinguish two different types of noise. The photo noise (so-called "shot noise") in the shadows appears visually very similar in the two photos, as you would expect because the exposures were identical. But the non-HTP photo shows a lot of extra noise in the deep shadows, which I believe is quantization noise. The reason is simple: because the numbers are small, it is not feasible to represent small differences in brightness. For example, a value of 5 is 25% higher than 4. Using HTP causes shadow values to be mapped to higher numbers, and thus reduces the quantization noise. Consequently, the HTP photo shows a wider range in the jpeg files -- which, after all, is the purpose of HTP. As far as I can tell, the results are exactly as expected from the graph presented by DPR ( http://www.dpreview.com/reviews/canon-eos-700d-rebel-t5i/18 ).

The DPR reviews of the T3i and T5i cited increased noise as a penalty for using HTP. This needs to be considered in context, and in my it's rather misleading. Misunderstanding of these reviews may be the source of a common misperception. If HTP causes you to underexpose, then I suppose you could say that in that sense it does increase noise. But in the same sense, reducing exposure to prevent overexposing highlights is exactly equivalent, and also increases noise.  HTP is a another way of adjusting exposure. I've never heard anyone complain that correct exposure increases noise.

I can think of one possible, slight disadvantage to using HTP, in that it disables the Auto Lighting Optimizer (ALO). I have never used that. I suspect that it just remaps the image values in jpegs, to increase the contrast.

I have the images to show, complete with lots of detailed clips showing noise, etc.., but I'm gone and won't have access to those files for a few weeks. You can try it yourself, but don't make the mistake of using the same ISO settings. If the raw files are not virtually identical, then you will have made a false comparison.