fredmiranda.com
Login

Moderated by: Fred Miranda
  

FM Forums | Canon Forum | Join Upload & Sell

1              end
  

Archive 2010 · highlight tone priority and RAW test

  
 
wfr2
Offline
• • •
Upload & Sell: On
p.2 #1 · highlight tone priority and RAW test


I don't think that is correct. As I said originally, my camera did NOT change the exposure at all. Activating HTP simply added a processing tag to be applied in the RAW processor. There is no in camera change. I wonder if you think there is an in camera change because of the way the processor is working.

Edited on May 04, 2010 at 11:03 AM · View previous versions



May 04, 2010 at 10:59 AM
thedigitalbean
Offline
• • • • •
Upload & Sell: On
p.2 #2 · highlight tone priority and RAW test


wfr2 wrote:
I don't think that is correct. As I said originally, my camera did NOT change the exposure at all. Activating HTP simply added a processing tag to be applied in the RAW processor. I wonder if you think there is an in camera change because of the way the processor is working.


Your camera does change the exposure by dropping the ISO by a stop (and keeping the shutter speed and aperture constant) this is why ISO 200 is the lowest ISO you can set with HTP on. The camera does this behind the scenes without showing it on the UI.



May 04, 2010 at 11:02 AM
wfr2
Offline
• • •
Upload & Sell: On
p.2 #3 · highlight tone priority and RAW test


My camera did not change iso. RAW data is simply tagged. I think we may be saying the same thing but it is not correct that the camera does anything. A processing tag change is simply added and can be ignored in processing manually.


May 04, 2010 at 11:06 AM
Eyeball
Offline
• • • •
Upload & Sell: Off
p.2 #4 · highlight tone priority and RAW test


wfr2 wrote:
My camera did not change iso.


I think this is the key to your understanding. You say your camera did not change ISO (or exposure) based on what? The fact that the camera didn't indicate that ISO/Exposure was changed via the on-camera settings? It lied.

If you want to see a hint that the camera is "lying", set your camera to ISO 100 and then turn on HTP. Check your ISO now and it should be 200 instead of 100. Why? Because when HTP is on, the camera (unreported to you) uses an ISO one stop less. There is no natural ISO one-stop less than 100 so it has to boost the reported ISO to 200 so it can shoot at an (unreported) ISO of 100.

What is being said is that the only difference between shooting with HTP on and shooting with HTP off but with an exposure compensation of -1, is the flag that tells the raw processing software to re-adjust the exposure +1. The raw data in both cases has been altered in-camera, shifting the data one stop to the left on the histogram. What is different is whether the +1 exposure adjustment in the raw processor is automatic, activated by the flag, or done manually by you.



May 04, 2010 at 11:59 AM
wickerprints
Offline
• • • •
Upload & Sell: Off
p.2 #5 · highlight tone priority and RAW test


If it is indeed the case that HTP is just a tag in the RAW file, then why is it not an option in DPP, like all the other reversible settings that can be applied in post? (e.g., lens aberration correction, white balance, picture style.)

My experience with HTP is that it is not a post-RAW capture command, but rather it effectively changes the sensitivity of the camera at the time of the exposure. As such, it is irreversible; i.e., you cannot elect to undo the effect after the RAW is written to the card. The reverse is also true--you cannot turn it on after the exposure is made.

HTP is exactly what it says it is. It prioritizes the highlights. That increased headroom does not come for free--the cost is increased shadow noise--because the total dynamic range of the sensor is fixed.



May 04, 2010 at 12:27 PM
wfr2
Offline
• • •
Upload & Sell: On
p.2 #6 · highlight tone priority and RAW test


I tried it at 100 ISO and you are absolutely right. And when I disabled HTP, it stayed at 200. So the camera is secretly underexposing and hiding that fact by telling the processor to add +1? Now I am really perplexed. I try to shoot to the right side of the histogram and shoot +2/3 Exposure Compensation quite a bit to move the histogram to the right to reduce noise. If my camera is secretly underexposing, it would seem to be doing the opposite of what I am trying to do Your explanation explains why the histograms look alike (between HTP and nonHTP) but isn't HTP adding noise in the use of higher ISO and underexposure subsequently corrected in post? BTW, how do you know this is what is happening?


May 04, 2010 at 12:40 PM
garyvot
Offline
• • • • •
Upload & Sell: On
p.2 #7 · highlight tone priority and RAW test


wfr2 wrote:
My camera did not change iso. RAW data is simply tagged. I think we may be saying the same thing but it is not correct that the camera does anything. A processing tag change is simply added and can be ignored in processing manually.


The camera does indeed "do something"--it underexposes the image. (More precisely, it applies 1-stop less analog gain to the signal from the sensor and compensates by boosting via a custom curve.)

Your exposure settings do not change because all this happens under the covers. The resulting image has the same overall brightness, with a longer tonal scale in the highlights and increased shadow noise.

Unlike other in-camera settings (ALO, PIC, Picture Styles, etc.) HTP *does* affect the RAW data because the exposure has been adjusted during capture.

Again, you could achieve this same result manually by reducing the exposure and applying a different curve in post, but I like that I get this result without having to mess around with exposure too much, so for me it's a useful feature.



May 04, 2010 at 12:46 PM
wfr2
Offline
• • •
Upload & Sell: On
p.2 #8 · highlight tone priority and RAW test


OK, you guys are convincing me but I am still thinking HTP will be a noise causer at the higher ISOs I shoot at--400-1000 ISO for big lens use. Right? I don't like the fact that HTP is creating noise by underexposing. Or does the decrease in ISO and underexposure balance out the noise issue?


May 04, 2010 at 12:56 PM
Eyeball
Offline
• • • •
Upload & Sell: Off
p.2 #9 · highlight tone priority and RAW test


wfr2,

HTP does have a noise visibility penalty since the camera is underexposing and then the exposure is being moved up in post. Whether that penalty is worth it or not is your judgement call.

You state that you use "shoot-to-the-right" technique. HTP is the opposite. When switched on, it is "shoot-to-the-left" (by one stop). If you are using STTR at +2/3 and have HTP on, HTP is cancelling your +2/3 and adding another -1/3 of underexposure.

Ultimately, you have to decide what brightness levels are important to you if the scene has brightness levels which extend beyond the dynamic range of the camera. If you want to reduce shadow noise, you shoot to the right. If you want to avoid blowing out highlights or bright tones like a wedding dress, you shoot to the left.



May 04, 2010 at 01:15 PM
wickerprints
Offline
• • • •
Upload & Sell: Off
p.2 #10 · highlight tone priority and RAW test


Please read the following post about ETTR that I wrote some time ago:

https://www.fredmiranda.com/forum/topic/890495/0#8379614

ETTR and HTP are opposite strategies. The basic principle of ETTR is that you want to make sure the brightest part of your subject is represented at the rightmost part of the histogram, even if that means your camera's meter thinks you are overexposing.

Remember, a light meter exposes for the middle gray. The goal of the camera meter is to determine what exposure gives the proper average illumination. If, however, the distribution of tonal values is narrow (i.e., low contrast), then while the meter's calculation generally gives you a good exposure out-of-camera, it doesn't take into account the S/R ratio or the fact that you can reduce it by deliberately overexposing (push the histogram to the right as far as it can go without blowing highlights), then pulling the exposure back down in post.



May 04, 2010 at 01:24 PM
wfr2
Offline
• • •
Upload & Sell: On
p.2 #11 · highlight tone priority and RAW test


Interesting post by wickerpoint. Bottom line seems to be that we should be shooting to the histogram's right and, if necessary, increasing ISO to do that is better than getting a middle biased histogram. As Eyeball says, HTP seems to be the exact wrong thing to do so far as noise is concerned. I just experimented shooting a contrasty scene at ISO 6400 (real ISO of 3200 per wickerpoint) Enabling HTP saved the bright spots better than I could do in post on the nonHTP shot. Nothing was blown out on the HTP shot but I could not pull back the histogram completely in the nonHTP shot although the images looked the same on my screen so the blown pixels must have been very small in number. Interestingly, noise was slightly worse in the nonHTP shot. Cannot decide what to do now given that result.


May 04, 2010 at 02:07 PM
wickerprints
Offline
• • • •
Upload & Sell: Off
p.2 #12 · highlight tone priority and RAW test


When shooting a high contrast scene, ETTR is not generally an employable strategy, unless you choose to clip your highlights to a greater extent than what the camera's meter indicates.

For simplicity's sake, let's consider a monochromatic scene. The histogram represents the distribution of tones in the scene--a bin count of areas on the image of dark shades to light shades. Suppose the scene has a very wide distribution of such tones--i.e., there are parts that are very nearly completely dark (reflect almost no light), and there are parts that reflect a great deal of light.

Now, the sensor, being an electronic instrument, is only capable of measuring a fixed, continuous subinterval of the histogram. Anything to the left of the chosen interval is clipped to pure black (reflectance is so low the sensor registers no light); and anything to the right of the interval is clipped to pure white (reflectance is so high the sensor cannot distinguish whether it is just "bright" or "super nuclear blast bright").

You generally pick the interval to record by adjusting shutter speed and aperture. You can also adjust the ISO, but this falls under a different category which we'll talk about later.

The question of which shutter speed and aperture to choose for a given scene is answered differently by three methods: ETTR, HTP, and the camera's meter, which we'll abbreviate by CM. The answers are different because they have different intents:

ETTR's intent is to maximize signal-to-noise ratio. This is done by picking the right-hand endpoint (the white point) of the histogram so that you clip only "non-essential" highlights. In other words, pick the exposure for which everything that exposes as pure white is something you don't care about seeing any highlight detail.

CM's intent is to match average luminance to middle gray. That is, it picks the interval so that the midpoint is middle gray. Of course, evaluative or spot algorithms are not quite so simple but in all cases, the basis of comparison is the middle gray tone, not the highlight or shadow. So if your scene contrast is very high, the exposure selected by the meter will still cause clipping. This is why Canon offers HTP.

HTP's intent is to overcome the limitation of using a digital sensor to capture a high-contrast scene by trying to preserve as much highlight detail as possible. The idea is to use the CM to obtain the exposure, but then deliberately set the sensor's sensitivity one stop slower. The result is underexposed, but then the shadow and midtone data is pushed one stop back. (The highlights aren't pushed because then they'd clip to white.) Back in the days of using negative film, it was difficult if not impossible to overexpose to the point of completely losing all highlight detail, because the film just keeps getting more and more density the more light hits it. Underexposure was the thing to avoid, because an underexposed negative has no shadow information--it is simply not there.

HTP is simply an extra safeguard, one stop more highlight latitude, to be used in high-contrast situations when the camera's metering might select an exposure for which the highlights are clipped more than acceptable. But because the exposure is still chosen by the meter, it is really just a modified CM method, and ultimately relies on the assumption you want things to appear middle gray. The ETTR method doesn't care about middle gray--all you care about is pushing the histogram as far to the right as is reasonable. The result could be overexposed, or it could be underexposed. (If underexposed, you are stuck because you cannot clip more highlights by definition, yet your scene is too dark--this is where you use exposure bracketing and HDR.)

Changing ISO does not change the amount of light reaching the sensor. Changing shutter speed and aperture does affect the amount of light reaching the sensor. Thus, it is always preferable to use the latter controls to pick your exposure, but when you cannot, the ISO is the only parameter you can pick.



May 04, 2010 at 03:03 PM
eric_m
Offline
• •
Upload & Sell: Off
p.2 #13 · highlight tone priority and RAW test


Thanks! This has been most informative. Since I rarely use jpegs, I guess there's no reason for me to shoot with highlight tone priority. I thought it was actually changing the exposure "S" curve.


Eric



May 04, 2010 at 05:23 PM
wfr2
Offline
• • •
Upload & Sell: On
p.2 #14 · highlight tone priority and RAW test


Thanks wicker. I should have mentioned I was not doing ETTR, really just testing for noise caused by HTP. Since I did not notice more noise (actually appeared as less), what disadvantage is there to using HTP? Also, I assume HTP ISO reduction is constant so when shooting 400 ISO I assume I am really shooting 200ISO and can jump it to 800 ISO to really get 400 ISO? (WHEW, what a sentence!)


May 04, 2010 at 05:30 PM
garyvot
Offline
• • • • •
Upload & Sell: On
p.2 #15 · highlight tone priority and RAW test


ETTR was first proposed by Thomas Knoll several years ago as a technique to minimize shadow noise on digital sensors, and it has been accepted more or less as "gospel" by many. Its core assumption is that the most desirable aspect of image quality is extended dynamic range and reduced noise in the shadows.

However in many images the most effective *creative* treatment demands a long tonal scale in the highlights, not the shadows. Digital cameras that can render nuanced highlight detail in high-key lighting situations are often said to produce less "digital" and more film-like results.

In addition, the human brain seems perfectly willing to accept blocked up shadows in images, while blown out highlights are highly objectionable. This is what our eyes and brain do naturally when adjusting for changing light--we optimize our pupil diamter so that we see the brightest things clearly.

I submit then, that in some cases ETTR techniques (not meaning to point fingures at anyone here) can be counterproductive and lead to less convincing exposures. I know I'd rather have every drop of highlight detail at the expense of some shadow noise (which is often easily removed).

So--I think ETTR is a good technique in some situations but should not be slavishly followed. HTTP is a valid alternative exposure model when a long highlight tonal scale is desired.




May 04, 2010 at 07:22 PM
wfr2
Offline
• • •
Upload & Sell: On
p.2 #16 · highlight tone priority and RAW test


Hi Gary. I think you make a good point in a well presented argument. Also, as cameras and the new ACR get dramatically better at handling noise, ETTL may be fading in need. Of course, ETTL should not be used to blow out the highlights but it does appear that some detail is sometimes lost even when brights are not blown out. I try to keep my brights no higher than 245 but even then I feel some detail may be getting lost. I think I will continue to use HTP. I shoot white egrets and need the added protection. But my question remains: is shooting 800 ISO with HTP enabled really shooting 400 ISO?


May 05, 2010 at 09:23 AM
1              end




FM Forums | Canon Forum | Join Upload & Sell

1              end
    
 

Welcome back
Log in to your account