Canon's Highlight Tone Priority

[John Chardine]

I am shooting with a new 1DIV and it seems to be behaving a little differently than my old one.

Yesterday I was shooting white subjects in bright sun and in the past I have had good success with the Highlight Tone Priority (HTP) setting in these conditions. With the new body this setting seemed to work only with the JPEG previews but not the RAW images. Exposing to the right in manual mode caused the JPEGS to be fine (I was checking the histograms as usual) but the RAW images had lots of blown highlight areas. I had never noticed this mismatch before with my previous body. I just did a test with HTP disabled and the histograms in camera (which are based on the JPEG preview) and the highlight warnings match more or less what I am seeing in Adobe Camera Raw. All this makes me think that HTP does not work on RAW images.

Any insight would be gratefully appreciated.

[Roger Clark]

Hi John,

As I'm sure you know, the joeg is derived from the raw data. The sensor is linear and in post sensor processing (in camera jpeg, or later raw conversion software), a tone curve is applied. Image data saturation can occur in the joeg generation or raw conversion, but in the linear raw data the raw data only saturates when the signal in a channel (red, green or blue) saturates the sensor or range of the A/D converter. So, if you are seeing raw conversion saturation in ACR but not in in camera jpegs, this must be due to the way ACR is converting the raw data and applying the default tone curve. HTP should not modify the raw data; it should only modify the tone curve applied to the raw data in raw conversion or in camera jpeg conversion. So I would look for software settings. You could verify all this by getting DCRAW and converting a raw file with no tone curve applied.

Roger

[Mike Milicia]

When you take an image with HTP enabled, the camera is simply underexposing by 1 stop at capture time. This is done by lowering the ISO 1 stop from what you have dialed in.
For example, if you take an image at ISO 200, the camera will actually capture the image at ISO 100. This underexposes the entire image by 1 stop and this is what gets recorded
in the RAW file. But when the camera goes to generate a jpeg (either for the CF card or for the jpeg preview), it will see the HTP flag, and apply a different tone curve which will
boost the shadows and mid-tones without affecting the highlights. Thus, HTP is potentially useful when shooting jpegs since there is no other way to achieve this effect.
Without HTP, you would run the risk of the RAW data being within range but the highlights would be blown when the standard tone curve is applied to the RAW data and there
would be no way to get them back.

But when shooting RAW, the only effect on the RAW file is that the image is underexposed by one stop and the HTP flag is set. The rest is up to the RAW conversion software.
When Adobe ACR sees the flag, all that it does is automatically boost the exposure slider zero point by one stop. If you needed the effect of HTP to roll off the highlights, you
can then just use the Recovery slider to do exactly that. But why not expose the entire image "correctly" at capture time and then use Recovery slider, etc. to roll off the
highlights as needed? There is no point in using HTP for RAW capture unless you think that underexposing all of your images by 1 stop at capture time and then selectively
boosting the shadows and mid-tones in post-processing is a good approach.

I do not use DPP but my understanding is that DPP handles HTP differently. When DPP sees the HTP flag set on the underexposed RAW file, it automatically applies a different
tone curve in the RAW conversion which again will selectively boost the shadows and mid-tones by 1 stop without affecting the highlights. I cannot begin to judge which
RAW converter is taking the best approach for dealing with the HTP flag but, in any case, I don't see how HTP is a useful feature when shooting RAW.

There are many discussions about HTP in various forums. Here is one post by Thomas Knoll of Adobe which talks about the ACR handling of HTP :

http://forums.dpreview.com/forums/re...ssage=25198619

Note : there's a typo in the first line of his post: "this" should be "that".

[John Chardine]

Thanks guys. This is all great information. I will have to look at why I seem to be getting different results with the two bodies. It will be difficult to tease apart the variables because the computer and Ps version, as well as the camera body, have recently changed.

One outcome of this is that now I understand more what HTP does, I don't think I'll use it.