highlight tone priority starts at 200 ISO. Why?

[vantomas]

when i switch on the highlight tone priority on my 1Ds MK3, the ISO will move up to 200? why?
does anybody know?

thomas

[mahleu]

I presume it's because it underexposes everything to preserve the highlights?

[Panopeeper]

The direct effect of HTP On is, that the shot will be made by one stop lower ISO than selected.

The indirect effect is, that the intensity will be increased by 1 EV in post processing, but the highlights get pulled back a bit by DPP, Breeze Browser and in-camera. When working with ACR, it adds 1 EV to the "exposure", but the slider does not show that. You have to use for example Recovery to achieve the intended effect.

[dwdallam]

The direct effect of HTP On is, that the shot will be made by one stop lower ISO than selected.

The indirect effect is, that the intensity will be increased by 1 EV in post processing, but the highlights get pulled back a bit by DPP, Breeze Browser and in-camera. When working with ACR, it adds 1 EV to the "exposure", but the slider does not show that. You have to use for example Recovery to achieve the intended effect.
[a href=\"index.php?act=findpost&pid=201221\"][{POST_SNAPBACK}][/a]

Is this a mechanical methodology instead of an analytical, meaning does the camera say, Ah, 1/250th, let's move that back to 1/500th?" Or does it calculate the blown areas, if any,  and then back the exposure off -1EV?

If it's mechanical, I really don't like that.

[vantomas]

the lowest iso value just moves up to 200. which is a mechanical method. and it really is quite useless, when you are using flash in bright dayligth even 100 iso sometimes is too high.

Is this a mechanical methodology instead of an analytical, meaning does the camera say, Ah, 1/250th, let's move that back to 1/500th?" Or does it calculate the blown areas, if any,  and then back the exposure off -1EV?

If it's mechanical, I really don't like that.
[a href=\"index.php?act=findpost&pid=201309\"][{POST_SNAPBACK}][/a]

[Guillermo Luijk]

Because HTP is a fake, it's a postprocessing applied to the image actually exposed with 1 f-stop less than set by the user.

When shooting ISO200+HTP, you obtain _exactly_ the same RAW file as shooting with ISO100 without HTP and the same aperture and shutter values.

Find here a comparision between two RAW files obtained in that way (HTP vs non-HTP at ISO 1 f-stop lower, taken at the same shutter and aperture). Identical RAW files were produced.

[vantomas]

thanks! that makes perfect sense. so let's forgett about the HTP than.

Because HTP is a fake, it's a postprocessing applied to the image actually exposed with 1 f-stop less than set by the user.

When shooting ISO200+HTP, you obtain _exactly_ the same RAW file as shooting with ISO100 without HTP and the same aperture and shutter values.

Find here a comparision between two RAW files obtained in that way (HTP vs non-HTP at ISO 1 f-stop lower, taken at the same shutter and aperture). Identical RAW files were produced.
[a href=\"index.php?act=findpost&pid=201327\"][{POST_SNAPBACK}][/a]

[203]

Here is a shoot I worked on recently, where HTP came in very handy indeed.

HTP off:
http://www.pbase.com/r_p/image/98482207/original

HTP on:
http://www.pbase.com/r_p/image/98482212/original
http://www.pbase.com/r_p/image/98482217/original

In my opinion, the HTP really saved me a lot of effort. In two seconds I had the white tie and shirt properly exposed, instead of having to drop down the whole exposure, then screwing around after the fact with curves to get the rest of the image from going too dark. I used HTP on and off throughout this catalog shoot, and in my experience it's great when needed. And the minor added noise did not end up being an issue at all. I will post a larger HTP file from this day when I have a second...
Of course there are people who are adamant that it is a waste of time, a farce, etc., but when it comes to saving me time when under pressure and shooting thousands of files per day, this is a very useful feature, with very little real-world down side.

All I.M.O.

[Guillermo Luijk]

so let's forgett about the HTP than.

Yes and no.

If you are a JPEG shooter, HTP is actually useful to compress the DR and preserve highlights with a good exposure in the shadows, for instance in high dynamic range situations where the subject of interest is some f-stops lower than the highlights of the scene.

If you are a RAW shooter using Canon's DPP to develop your RAW files, which implements the HTP algorithm (if I am not wrong), HTP can also be useful to save postprocessing time. The images and experience shown by 203 are good samples.

In any other case HTP is not only unuseful, but can be confusing to obtain proper results in terms of exposure (basically you can leave empty an important portion of the histogram on the right side like happened here, increasing overall noise and reducing captured dynamic range with no added advantage).

The most important thing is to be conscious of how HTP internally works and decide whether it can be of use to you or not. It is also true that anything than can be achieved with HTP can be achieved without it shooting RAW plus adequate postprocessing.

I personally wouldn't use it since I am not a JPEG shooter and I don't use DPP.

BR

[Panopeeper]

Is this a mechanical methodology instead of an analytical, meaning does the camera say, Ah, 1/250th, let's move that back to 1/500th?"

It is a thoughtless process; no considerations of what so ever, simply reducing the ISO (NOT the exposure!).

Or does it calculate the blown areas, if any,  and then back the exposure off -1EV?

How could the blown areas be calculated at all without actually exposing the sensor?

The idea is not fully absurd, the sensor is exposed with live view. Now it would need a very (but very) fast sensor chip and processor (DIGIC XXV) to take samples in short intervals and evaluate the expectable outcome of the exposure based on the current time, aperture and ISO setting. But this would have to occur so fast, that when panning the camera, one would get almost simultanously the suggested exposure adjustment.

[203]

Panopeeper, are you a photographer? I am just wondering since you disparage this feature, when the actual photographers I know find it useful, as I do. If you are a photographer, care to show some of your work? I am just curious what kind of work you do where HTP is useless. - when I have shown above where it has come in handy for me - In Actual Use. (it saved my highlights, with no real penalty in terms of image quality, *and no added work!*)

[Panopeeper]

I am just wondering since you disparage this feature, when the actual photographers I know find it useful, as I do

I wonder what made you think I am disparaging HTP. HTP is good for in-camera JPEGs; it is counterproductive when recording the raw data. That's it.

I am just curious what kind of work you do where HTP is useless. - when I have shown above where it has come in handy for me - In Actual Use. (it saved my highlights, with no real penalty in terms of image quality, *and no added work!*)

I am using a neutral setting with a neutral WB (called sometimes uni-WB); thus I see a very close approximation of the raw histogram. That way I can adjust the exposure to maximize the dynamic range, instead of sacrifying a stop only "to be safe".

This is particularly important when shooting panoramas, because the dynamic range of the entire scenery may be much more than that of a single frame. If I choose to shoot with fixed exposure, then that has to be the best possible exposure(panoramas with variable exposure incur more labour in PP, I do that only in extreme cases).

[203]

I guess I said disparaging since you called the HTP implementation "thoughtless," among other things. But no worries. :-)

Care to show your work?

[Guillermo Luijk]

I am using a neutral setting with a neutral WB (called sometimes uni-WB)

Gabor, how is the UniWB working on the 40D? on my 350D is a bit optimistic since the camera unfortunately does not blink blown highlights if only partial saturation was there so I have to be careful. Do you prefer to look at the blinking highlights or the histograms in the 40D?

[Panopeeper]

I guess I said disparaging since you called the HTP implementation "thoughtless," among other things

The question was, if the adjustment is "mechanic" or "analytic". It is "mechanic", meaning that you select it and it will be done that way, no matter of the actual circumstances. That's what I meant with "thoughtless", not the design of HTP.

"Thoughtful" would be what dwdallam envisioned as calculating the potentially blown areas.

Care to show your work?

I am trying to avoid exhibiting the attitude of taking advantage of every semi-opportunity to show off my hobby; therefor I did not attach my site address to my signature. However, your insistance broke my anyway week resistance.

My photo site

[Panopeeper]

Gabor, how is the UniWB working on the 40D?

I am satisfied to about 90% (why exactly 90?). It occurs very seldom, that something is blown without me knowing it. The clipping indication (blinking) is totally reliable.

On the other hand, judging the "underexposure" (i.e. the distance from the right edge) is more of a problem, for

1. there is no visible edge of the histogram (it is as black as the surronding area),

2. the non-linear transformation can not be turned off; that compresses the highlights more than the area below. The consequence is, that if there is no clipping, one does not see, how much it is from the right edge in terms of EV. I see it very close, increase the exposure by 1/3 EV, it is still not there. Once more 1/3 EV, it is still not there.

Anyway, I don't spare the CF cards, I shoot bracketed very often, and if necessary, I shoot bracketed repeatedly. For a few days ago I made a trip to the Rocky Mountains (not far away from here, measured to Canadian relations). I have 10Gig on four cards; the maximum was 7Gig a day, with lots and lots of bracketed shots (I save them on HDs on a laptop at the evening). At home, I started out with checking the raw exposures and keeping the best ones. Rawnalyze makes it simple to pick based on the raw histogram.

[203]

Cool man, thanks ;-)