Highlight Weighted Metering - ETTR Without Technical Complications (Page 2): No amount of waffle changes the fact that when using 'highlight weighted metering', (based on Z6 results)...

Grahame Loc: Fiji

selmslie wrote:

If you look at the raw histograms you will see the... (show quote)

No amount of waffle changes the fact that when using 'highlight weighted metering', (based on Z6 results), that results are varied (amount of clipping) due to the size of the bright area, e.g. sky. The only way to achieve accuracy with the example I have suggested you undertake is to vary EC.

Whilst it can be a useful metering method your over simplification and flawed examples only suggest a lack of practical experience with its use.

selmslie Loc: Fernandina Beach, FL, USA

Grahame wrote:

No amount of waffle changes the fact that when using 'highlight weighted metering', (based on Z6 results), that results are varied (amount of clipping) due to the size of the bright area, e.g. sky. ....

Here is an extreme case shooting directly at a light source. Only the light source is clipped.

That's about 60k pixels out of 44.7 MP or about 0.1% of the image. What more do you want?

I am using aperture priority and ISO 400.

Maybe you need to learn how to use it.

(Download)

Grahame Loc: Fiji

selmslie wrote:

Only the light source is clipped.

Well done, you are proving the point I have made.

We see here that although spouting your EC of +.7 in your 'simple' first examples you have found it necessary to use EC of O.O here and still clipped it.

Let's see if you undertake that simple test I advised you to, although I suspect you know what the results will be and continue posting 'simple' examples that confirm nothing of interest other than suit your waffle.

kymarto Loc: Portland OR and Milan Italy

a6k wrote:

FYIO, highlight weighted metering is available in at least some Sony cameras, too. I have not tested it extensively but if it errs at all it does so on the safe side.

Example: https://helpguide.sony.net/ilc/1710/v1/en/contents/TP0001629671.html

It is great. I use it 100% of the time in any mode but manual.

selmslie Loc: Fernandina Beach, FL, USA

kymarto wrote:

It is great. I use it 100% of the time in any mode but manual.

In the third example on page one (with the PVC fence) about 6k pixels were blown, about 0.01% of the image.

You really have to go to extremes, like shooting directly at a light source, to get it to fail.

Here is another extreme example - an almost full moon an hour before sunrise. About 30k pixels are blown, about 0.06%.

But if we measure a rectangle over the moon and look at its raw histogram we will see that most of it is not blown at all.

As with all other aspects of photography, we need to use a little common sense.

(Download)

R.G. Loc: Scotland

selmslie wrote:

Here is an extreme case shooting directly at a light source. Only the light source is clipped....

It seems to me the main concern is that Highlight Weighted metering is prone to the same sort of susceptibility as Matrix metering - in other words is it going to give 100% protection to the highlights or is the level of protection going to vary depending on the predominance of shadow (or perhaps more accurately, depending on the ratio of highlights to shadow). If it's the latter, perhaps that's why they call it Highlights Weighted rather than Highlights Protected.

One predictable problem is that if you design the metering to protect the brightest highlights, all it would take to cause serious underexposure would be the presence of small, bright specular highlights (e.g. bright sparkles on water, filaments in light bulbs and the like).

So what do you do - design it to ignore small, bright highlights? And if so, at what point do you stop ignoring the brightest highlights (as a percentage of the overall exposure)? Small bright highlights (which will produce small blown areas) won't be a problem, but larger blown areas could be very problematic, and that's exactly what we're trying to avoid.

Is your second last posted example (the lamp) showing that small, bright highlights are ignored? Or is the exposure, as Grahame suggests, being skewed by the ratio of highlights to shadows (and/or their intensity)?

selmslie Loc: Fernandina Beach, FL, USA

R.G. wrote:

It seems to me the main concern is that Highlight Weighted metering is prone to the same sort of susceptibility as Matrix metering - in other words is it going to give 100% protection to the highlights or is the level of protection going to vary depending on the predominance of shadow (or perhaps more accurately, depending on the ratio of highlights to shadow). If it's the latter, perhaps that's why they call it Highlights Weighted rather than Highlights Protected. ...

As my last two examples show, it's actually protecting the highlights with almost no weight being given to the rest of the image.

It appears to have recognized that the areas with extreme brightness may actually be light sources and that they probably should blow out a little. If we were looking at the sun (not a bright idea) it would probably do something similar to what it did looking at the moon.

Don't let the words get in the way For example, "white balance" actually means "neutral gray" balance where red=green=blue in the converted image. I think that the manufacturers did not want to go to great lengths with precise language. Experienced photographers could actually understand what they mean.

selmslie Loc: Fernandina Beach, FL, USA

R.G. wrote:

One predictable problem is that if you design the metering to protect the brightest highlights, all it would take to cause serious underexposure would be the presence of small, bright specular highlights (e.g. bright sparkles on water, filaments in light bulbs and the like).

Small specular highlights are usually too small to be considered. There may be many of them within an area being measured like in the fountain in the second example. They just get averaged in with the rest.

The exception would be the reflection of the sun on a relatively flat surface like glass or a car bumper. But at what point is that a specular highlight and not simply the reflection of a light source?

R.G. Loc: Scotland

selmslie wrote:

....at what point is that a specular highlight and not simply the reflection of a light source?

That's a good question. Do you have an idea as to what the camera "decides"? In your picture of the lamp, is it just the filament area that's blown, or is the blown area a larger area that's just very bright?

morkie Loc: Simi Valley CA

I spent a few hours testing my Sony a6500 and found (for what I typically shot) I can usually overexpose 2 stops and recover highlights - using matrix metering. Simple.

a6k Loc: Detroit & Sanibel

morkie wrote:

I spent a few hours testing my Sony a6500 and found (for what I typically shot) I can usually overexpose 2 stops and recover highlights - using matrix metering. Simple.

I also have a Sony a6500. Using the zebra indicator allows me to prevent blown highlights easily. The technique which you describe may work for you most of the time, but in my humble opinion it's crude and wastes the light. It's also less accurate most of the time.

I don't know what your typical subject matter is. But normal outdoor scenes cannot be over exposed by two stops safely. That said, if your typical scene has low dynamic range over exposing is, in fact, exposing to the right. In that case it's a good idea.

It is worth noting that this sort of choice implies shooting raw, not JPEG. Usually.

jlg1000 Loc: Uruguay / South America

a6k wrote:

FYIO, highlight weighted metering is available in at least some Sony cameras, too. I have not tested it extensively but if it errs at all it does so on the safe side.

Example: https://helpguide.sony.net/ilc/1710/v1/en/contents/TP0001629671.html

Yes, my A7C has it and I use it extensively. It is sometimes a bit conservative, but it can be compensated by using the zebras

mikegreenwald Loc: Illinois

In situations with unusually wide DR, I choose to shoot HDR, 3 or more shots with 2 stop intervals. That leaves three basic shots to choose from, plus the HDR. Shooting only ETTR risks losing shadow detail.
HDR is a simply, unsophisticated solution, and rarely fails!

selmslie Loc: Fernandina Beach, FL, USA

R.G. wrote:

That's a good question. Do you have an idea as to what the camera "decides"? In your picture of the lamp, is it just the filament area that's blown, or is the blown area a larger area that's just very bright?

Probably very few engineers at Nikon could actually answer that and they are not going to tell us except in vague language. But I can make an educated guess.

Suppose that the image area is divided into a matrix equal sized squares 80 squares high and 120 squares wide for a total of 9600 squares. Each square would contain 8288x5520 pixels ÷9600 squares for about 4,765 pixels in each square. Each square will be represented in an 80x120 matrix as an average of the luminance value (mostly from the green channel) of the pixels in each square. That in itself is quite an achievement.

If there were only 960 squares there would be ten times as many pixels in each square. You would still have to calculate the averages from the same total number of pixels but the matrix would be 1/10th as big and there would be 10x as many pixels for each square in the matrix.

We don't know how many squares Nikon uses. We can only speculate about the next step.

Now consider the number of pixels it takes to record a specular highlight. Here is a 20x18 snip of some of the fountain highlights in the second image.

That's only 380 pixels but it contains about a dozen specular highlights from water droplets.

So it depends on whether the matrix contains 9600 or 960 cells or any other number. It's always going to be a collection of averages.

We might now examine all of the elements in the matrix and base the exposure on the brightest cell, But that would not guarantee that none of the pixels in that cell is blown, only that the average is not blown.

Don't take all of this data processing for granted. The mainframe computers I started with in the 1960s would have struggled with this much data and this many calculations. Now your camera gets it done in the blink of an eye.

selmslie Loc: Fernandina Beach, FL, USA

a6k wrote:

I also have a Sony a6500. Using the zebra indicator allows me to prevent blown highlights easily. The technique which you describe may work for you most of the time, but in my humble opinion it's crude and wastes the light. It's also less accurate most of the time.
---
It is worth noting that this sort of choice implies shooting raw, not JPEG. Usually.

Neither method is crude and neither will waste light once you learn how to use it.

I also have a Sony with zebras and I wish than the Z7 had them for photographs. Bit it only has them for video.

Zebras have the advantage that you will see them before tripping the shutter.

On the other hand, with Highlight Weighted Metering you can use auto exposure without having to touch the EC dial.

Of course, it's not fool proof. Your ISO can't go below base, your aperture can't get smaller than the minimum and your shutter speed can't go faster than the camera's maximum. Any of these limits can result in blown highlights. But we already knew that, didn't we?

And yes, since we are talking about the raw file, it's safe to assume that we will be developing the raw image on our computer. Highlight Weighted Metering is not for JPEG shooters. For JPEGs Zebras tell us when we might want to skip the image or learn how to process raw.