Highlight Tone Priority

Can anyone tell me how Canon's new Highlight Tone Priority mode works?

Thanks.

Doug, it's apparently "proprietary", but here's the description from RG:

A Highlight Tone Priority image is processed differently in the camera, regardless of whether the camera is set to CR2 or JPEG. Specifically, the amount of gain applied during the analog-to-digital conversion step is less. For instance, if the camera is set to ISO 200, the amount of gain applied is similar to or the same as ISO 100, which means more of the highlight detail captured by the sensor is preserved during this early in-camera processing step. This is the main reason the camera can't be set lower than ISO 200 when Highlight Tone Priority is enabled; it requires the extra highlight headroom it gets by applying a level of analog signal boost to the sensor data that is commensurate with a lower ISO setting.

Click to expand...

This sounds like it needs cooperation (awareness) from the RC by applying a different film curve.

- DL

Hi, Don,

Thanks.

So basically we cut the sensitivity of the imaging system to half that implied by the "ISO" setting we have made (and thus to half that which is advised to the metering system) in order to get 1 stop more headroom.

This sounds like the same thing we would get if we set the ISO setting to one stop lower than we would otherwise do and set -1 Ev of EC.

I wonder if there is something more to it than that.

Ironically, the EOS digital cameras normally have about 1/2 stop less headroom than that used as the premise of the standard "metering/exposure strategy" that is implemented by the interaction of the ISO standard for AE system "calibration" and the ISO standard for determining the ISO speed rating of a digital camera. They do this by "rating" the ISO speed lower than what it would actually be if determined per the ISO standard, and feeding that value into an AE system that obeys the ISO standard.*

Why does Canon do that? I believe it is to "make the images look brighter" when viewed directly out of the camera. Is it good to burn headroom for "punch"? Not always. So now we have a feature to retreat from that (and then some)!

* Of course, in the EOS 1D Mark II, the EOS 40D, and the EOS 1Ds Mark II, they no longer state an ISO speed. Instead they state an ISO recommended exposure index (REI). This is a number that the manufacturer is free to state as they see fit to put into effect the "metering/exposure strategy" they think is the most attractive - without having to "misrepresent" the ISO speed of the camera.

Thanks again.

Best regards,

Doug

> This sounds like the same thing we would get if we set the ISO setting to one stop lower than we would otherwise do and set -1 Ev of EC.

Basically, but I suspect that a different tone curve with a broader shoulder may also be used. I think the primary advantage might be for jpeg shooters.

- DL

Don Lashier said:

> This sounds like the same thing we would get if we set the ISO setting to one stop lower than we would otherwise do and set -1 Ev of EC.

Basically, but I suspect that a different tone curve with a broader shoulder may also be used. I think the primary advantage might be for jpeg shooters.

Click to expand...

Possibly, but I've learned from the ISO 'L' confusion which, when properly evaluated on my 1DsM2, appeared to be an ISO 70-80 analog gain setting, while the metering system (under-)exposed for ISO 50. This also did improve general highlight tones, but at the expense of maximum Dynamic Range. With (proper) exposure 'to-the-right' the DR increases (slightly) vs ISO 100 and non-clipping exposure is optimal.

I expect this "Highlight Tone Priority mode" to work in a similar fashion (perhaps with an additional in-camera tone-adjustment for JPEGs), unless proven otherwise.

Bart

Hi guys

I must admit that this facility/functionality really appealed to me for when I shoot weddings.

I have now used it for a few weddings and find it makes the images much denser than without the option set (as you might expect I suppose) although it does seem to do what it says on the tin - the highlights ARE less prone to being blown and for a subject like a bride in a conventional outfit, this does seem like good news.

However, I do find that the shadow areas are much noisier in consequence and this seems to be exaggerated in interior shots with flash is being used.

Consequently, I have reverted back to NOT having this option set and instead relying on my judgement and experience when determing exposure. Not always accurate but I feel happier working this way.

Barrie

I don't shoot weddings but I've been on various forums for 10 years (DPR, RG, here) and have seen many many "problem" wedding photos posted and frankly very few if any of them actually had blown highlights, or for that matter hopelessly plugged shadows. The problem was just than when adjusted for midtones the highlights (dress etc) rode so high up the tone curve that they landed in the shoulder roll-off and hence pretty much lost local contrast and hence detail.

For someone who shoots raw there are other ways to deal with this, the simplest being to pick a more linear film curve as is available in C1. As I noted previously, I believe the primary value of this feature is for jpeg shooters, and Barrie you just more or less confirmed this.

- DL

Barrie Harwood said:

I have now used it for a few weddings and find it makes the images much denser than without the option set (as you might expect I suppose) although it does seem to do what it says on the tin - the highlights ARE less prone to being blown and for a subject like a bride in a conventional outfit, this does seem like good news.

However, I do find that the shadow areas are much noisier in consequence and this seems to be exaggerated in interior shots with flash is being used.

Click to expand...

Hi Barrie,

Could you share some of these pictures in which the highlight detail was well rendered at expense of noisier shadow areas? I'd love to see how bad it might be with Canon's Highlight Priority Feature in Wedding Photography.

With film, I've even seen very happy brides showing off their book of prints that have blown veils but they don't see it! I wonder whether they would ever realize there's anything wrong with the shadows? The groom's face will be always well lit and the black suit is not going to be examined that much. So will the shadows matter much after all?

Not that I don't seek perfection. I'm just interested in your practical findings.

Asher

Hi Asher

I'll dig out a few and post.

As a slight aside and picking up on your comment about bride's veils, it is easy with photography of this type to forget that on the whole, our clients are not photographers and whilst obviously we need to do as technically a good job as possible, the aesthetics and story telling component is far FAR more important to the couple than whether the 'rule of thirds' has been obeyed, whether the shadows are noiseless, veils are blown etc.

It would be no use explaining to the bride why you haven't photographed her mum blubbing into her hanky by saying ".... yes, but have you noticed that by using f/4 on my CaNIkax 12-750 f/1.8 lens I have created such beautiful bokeh don't you think....."

Barrie

Don Lashier said:

Doug, it's apparently "proprietary", but here's the description from RG:

Quote:
A Highlight Tone Priority image is processed differently in the camera, regardless of whether the camera is set to CR2 or JPEG. Specifically, the amount of gain applied during the analog-to-digital conversion step is less. For instance, if the camera is set to ISO 200, the amount of gain applied is similar to or the same as ISO 100, which means more of the highlight detail captured by the sensor is preserved during this early in-camera processing step. This is the main reason the camera can't be set lower than ISO 200 when Highlight Tone Priority is enabled; it requires the extra highlight headroom it gets by applying a level of analog signal boost to the sensor data that is commensurate with a lower ISO setting.

This sounds like it needs cooperation (awareness) from the RC by applying a different film curve.

- DL

Click to expand...

This is exactly how it works. I don't have a 1D3 but someone who does was kind enough to supply me with two RAW files taken with HTP on and off. Subtracting the black point of 1024 and doublling the raw values of the "HTP on" file gave those of the (blackpoint subtracted) "HTP off" file to within a few percent. So basically HTP keeps the metering indication in the viewfinder the same while behind the scenes halving the ISO gain applied. One could do the same by halving the ISO and applying -1EV exposure compensation when shooting RAW. I suspect the main application is a convenience for jpeg shooters who don't have the latitude to push exposure in post.

Hi Emil,

It might still be interesting to do a little more refined experiment. Shoot the same shot, same exposure, one with HTP on, other with it off but ISO halved, then process them both thru DPP and compare. This would reveal whether or not DPP was paying attention to the tag and providing different processing. Looking at the Canon samples I'd guess there is a different tone curve.

- DL

Hi, Emil,

Emil Martinec said:

This is exactly how it works. I don't have a 1D3 but someone who does was kind enough to supply me with two RAW files taken with HTP on and off. Subtracting the black point of 1024 and doublling the raw values of the "HTP on" file gave those of the (blackpoint subtracted) "HTP off" file to within a few percent. So basically HTP keeps the metering indication in the viewfinder the same while behind the scenes halving the ISO gain applied. One could do the same by halving the ISO and applying -1EV exposure compensation when shooting RAW.

Click to expand...

Thanks for doing the reverse engineering on this.

What you report is what I had suspected.

Others have mentioned some additional subtleties that they haven't really be able to describe in any cogent technical way (like "compressing the rest of the gray scale" etc. etc.).

I suspect the main application is a convenience for jpeg shooters who don't have the latitude to push exposure in post.

Click to expand...

As you point out, one could of course do the same thing by setting EC to-1 and using 1/2 the ISO setting. But this is done with a single move (albeit harder to access!) and has a sexier sound to it.

Thanks again.

Best regards,

Doug

Never mind!

Doug