What do you meter off of when the scene is not evenly lit in landscape?

[New Daddy]

What do you meter off of when the scene is not evenly lit in landscape? (I know it's such a basic question, but I couldn't find good answer on the internet.) Is this the time for matrix (evaluation) metering without any weight on center or active focus point?

Just to clarify the question: when you have a subject in an unevenly lit scene - an environmental portrait for example - you'd just meter off of your subject. (What I do is spot meter off of the subject's face and make EV compensation depending on the tone of the person's face.) But if you are after the whole scene - e.g., you're taking a shot of your livingroom for open house brochure or a landscape scene at a national park -, what do you meter off of? Is there a good rule of thumb, such as spot meter off the brightest point in the scene and apply +3 EV? If nothing else, I'll have to use matrix metering, but I'm hoping there is a method where I have more control.

[Glenn NK]

There are no rules, but for a starter I'd offer this:

If the DR of the scene exceeds the DR of the camera, then compromises are in order:

1) Don't blow the highlights if they are important,

2) Don't blow the "darklights" (for want of a better word), if they are important,

3) Use bracketing and combine with software to preserve the lights and the dark areas if both are important,

It's quite a complex decision as to what to do, consequently relying on the thought processes of a camera will be futile as cameras don't think too well (at least mine doesn't).

Further to this:

(a) specular highlights (sunlight reflections off water for example) cannot be captured - even our highly adjustable eyes can't see the details in specular highlights, so for the camera it's really hopeless.

(b) very often when we look carefully at a scene, there are dark areas that our eyes cannot see well - my personal take is that we accept this as being a natural feature of the scene. For example, looking across a lake into a sunset, can you see the details in the trees across the lake that are in full shadow? I can't, and if the exposure in the dark area is bumped up revealing detail, I find this unnatural. Maybe this is why so much HDR looks fake and over done - we intuitively know something is unnatural.

(c) if there is movement in a scene, multiple exposures may not t work for some subjects (leaves on a tree for example).

What I do is take a shot, check the histogram, and adjust if required. Although I've taken multiple (three) shots in the field, I've never yet resorted to combing them as (perhaps fortunately), careful PP has been adequate.

Last year I shot a distant mountain with snow on the peak - exposing to capture detail in the snow would have been pointless as the mountain was a few miles away and there would be no detail in with any exposure level. So I let the highlights in the snow blow out.

Glenn

[DanK]

Zone metering or exposure fusion.

If the dynamic range is manageable and don't need multiple images, you can find the brightest area you don't want to blow out (as Glenn said, forget about specular highlights), meter off that, and then try for a start opening up 2 stops. You can do the same with the dark end instead, as Glenn suggested, but obviously closing down rather than opening up. Depending on the illumination of the scene, 2 stops may not be exactly right, but you can quickly adjust after a test shot or two.

For blending multiple exposures, I much prefer exposure fusion to HDR. Google it if you are not familiar with it. For that purpose, I use a Lightroom plugin called Lightroom Enfuse.

[ajohnw]

Generally I find it's best to make sure that high lights aren't clipped and then if needed bring up the dark end in post processing. Taking shots is the only way to find the best way to use a particular camera in that respect and any other way you use it.

People say cameras meter by averaging the whole shot and exposing for mid grey. I am not convinced that is the case in all modes eg matrix metering.

Spot metering in my view is a good way of causing a lot of problems. What do you do - choose something mid tone and meter of that but that doesn't really give any control of other tones. It also leaves the question has the mid tone been judged correctly. Or maybe take some readings all over the place and wonder what to do with them.

Best control in situations that need it is the in camera histogram. Histograms are explained here

https://www.cambridgeincolour.com/tu...istograms1.htm

People sometimes mention shooting to the right - setting an exposure so that the histogram just touches the right side end of it or falls a tiny bit short. In principle if a purely mid tone image is taken like that it can be corrected in post processing and in that case it may improve the capture of the dark end if there is any there. Personally I think that's something to try for yourself. Often discussions about shooting to the right involve pictures with white in them. Really it applies to well lit colours - high tone levels in other words. Some cameras offer RGB histograms. making that effect easy to see some just offer what is essentially a black and white one. Hope that explains tone levels. For this method to work the camera needs to meter the entire scene.

I find that a centre weighted reading is best for general use but may blow highlights that aren't metered. Typically that will give decent exposures of peoples faces and clothes - well at least the one that is metered. If it's a group you can scan the crowd with it too. The histogram is the best option in this mode too really but if a lot of white I would tend to leave a bit of a gap at the right end but that depends on the camera. On some clipping it a bit wont matter.

What you will find is that you are taking shots where the entire scene can't easily be fitted into a sRGB colour space which just has 256 tone steps. What happens depends on the camera and increasingly how it's set up. If you use raw there is still a need to know what the camera will make of the scene and how to persuade it to capture the tone levels you want and then how to post process them as you want again into a jpg/print or view on a screen. For raw I would say usually avoid clipping highlights.

If the camera just can't capture the tone range that this there and you need it to in come HDR. Several exposures with say 3 stop differences that are merged together. Sometime if the tonal range just fits into what the camera can capture it's possible to produce 2 different exposures from the raw file and merge those. Often called fake HDR but it can actually cram another 3 or 4 stops into an sRGB image in a completely natural way so is true hdr really. The shot just looks normal and more like what would be seen by the eye as it looks around.

John
-

[Dave Humphries]

Hi Rick,

My advice is; keep it simple.
e.g. at this stage of your experience, you probably want to capture the scene in a single exposure

Use evaluative/matrix metering and take a shot, then review the histogram and as Glenn advises in his 1 or 2, use EC to ensure you don't clip what's important to you. If you're not sure and the highlights are not specular, then protect the highlights.


If you start dabbling with spot metering and applying large amounts of EC when you're (perhaps) not fully understanding exactly you should be spot metering off, I can almost guarantee you'll waste more time and potentially spoil more shots than is necessary.

However, if you're of the disciplined mindset that can conduct a series of experiments on a few shooting scenarios to gain the necessary HDR/bracketing and blending experience, then "go for it".

Hope that helps,

[Glenn NK]

Zone metering or exposure fusion.

If the dynamic range is manageable and don't need multiple images, you can find the brightest area you don't want to blow out (as Glenn said, forget about specular highlights), meter off that, and then try for a start opening up 2 stops. You can do the same with the dark end instead, as Glenn suggested, but obviously closing down rather than opening up. Depending on the illumination of the scene, 2 stops may not be exactly right, but you can quickly adjust after a test shot or two.

For blending multiple exposures, I much prefer exposure fusion to HDR. Google it if you are not familiar with it. For that purpose, I use a Lightroom plugin called Lightroom Enfuse.

Good point Dan. I've used Enfuse with good results, but mostly I've been fortunate (using ETTR and adjusting with the histogram), that the dark areas are OK. But there will be times when the DR is too large - sunset with foreground in shadow - quite often I resort to a silhouette or a "partial silhouette" rather than blend.

But I should warn Rick (OP), that in-camera settings for contrast will have to be adjusted to a negative value in order to get the JPEG generated histogram closer to what the RAW file will look like (this of course assuming that he's shooting RAW).

If he is interested, he can ask more about this approach.

Glenn

[ajohnw]

.
But I should warn Rick (OP), that in-camera settings for contrast will have to be adjusted to a negative value in order to get the JPEG generated histogram closer to what the RAW file will look like (this of course assuming that he's shooting RAW).

If he is interested, he can ask more about this approach.

Glenn

I find that the standard settings are fine for high lights and the histogram in a camera isn't exactly a precision instrument anyway. Perhaps it depends on the camera that is being used. So much of the dark end is compressed into it that this tends to look after itself if the camera is capable of capturing it.

Perhaps the best option really is to look at the jpg tone curves on the Dpreview site for a particular camera and make a judgement based on that. The best option however is play with any settings after some experience of actually achieving 100% highlight capture with excellent gradation in various situations.

Even Canon offer a curve that compresses highlights a little now. Dpreview often use a particular tone curve and make comments about how Cannon chop off highlights. I'm inclined to think Canon were right to do that really if some one wants to completely avoid washed out / pooor gradation in whites in particular. It does leave them a problem though. It's oh so easy to get them in jpg's - but there is a very good chance that they will be fine in raw.

Of late due to camera changes I use a different method. If the preview shows any clipping in the highlights I adjust and shoot again if needed. I usually use centre weighted for that and if suitable nudge the exposure by metering with more sky in the view. This method is more less metering at a pixel level and doesn't miss anything.

John
-

[Inkanyezi]

I think my approach is a bit simpler. When I have a sunlit landscape, regardless of dynamic range, I use the largest dynamic range that my camera offers and expose after the Sunny 16 rule. I have checked with the cameras I use, and generally they do blow the highlights whatever setting I use for the light meter - except if I compensate to minus.

With the Panasonic G1 that I used until it went AWOL, it also worked setting compensation to +2 and spot metering the brightest highlight that I wouldn't want to blow, but Sunny 16 is faster and simpler and works 100 % for landscapes.

[rpcrowe]

Originally bracketing exposures was not a technique used for capturing images to composite into an HDRI.

It was done by professional photographers who were shooting with the original 35mm Kodachrome a very slow film (= to ISO 12) which had little or no latitude. The bracketing was simply to allow the photographer three choices in exposure.

Bracketing with a digital camera (especially with Canon DSLR cameras) is exceptionally easy and costs nothing to accomplish.

It can accomplish one or both of these aims:

Ensure that there is at least one correct exposure of a chancy scene...

Provide a new photographer a graphic example of how exposure should be managed over and below what the meter reads for light and dark subjects. (The snow covered landscape or the black cat in the coal bin).

[Glenn NK]

This is what I was looking for when I first posted:

http://www.guillermoluijk.com/tutori...b/index_en.htm

The technique has been around for several years, and was widely discussed on POTN.

The full name is Unitary White Balance.

The method works well with the following caveats:

1) Specular highlights will still blow out, so must be taken into consideration.

2) The LCD will not look "normal". For those that use the camera's LCD to judge an image, it could be disturbing to see a greenish cast.

In the article by Guillermo Luijk, he goes into great detail on how to set the UniWB, but later on he describes a quick method (which doesn't work for some cameras including mine):

Shoot at some brilliant source of light for a couple of seconds, so that all three channels get blown in all pixels. Use the resulting RAW file that will be in the memory of the camera to set custom white balance.

I will use this method for my old 30D as it's very easy to use - select a long shutter time with lens wide open (Manual settings obviously), and shoot the sky.

Glenn

[jcuknz]

There is the 'old timers' way of metering off the back of your [caucasian ] hand and opening up one stop? There is nothing to stop you using your camera as one used a hand held meter in the old days.

[Glenn NK]

There is the 'old timers' way of metering off the back of your [caucasian ] hand and opening up one stop? There is nothing to stop you using your camera as one used a hand held meter in the old days.

True, and it worked for film because the response of film was not linear. When digital clips on the high end the result is no detail, no colour, nothing at all.

With digital, the best indicator of clipping is the histogram - the trick (if shooting RAW) is to get the in-camera JPEG which displays on the LCD to "tell the truth".

If shooting JPEG, then the histogram on the camera is a very good tool.

The UniWB approach puts a JPEG on the camera that matches the exposure of the RAW.

Glenn

[ajohnw]

Actually back of the hand and leave it alone but it wont stop blown highlights.

Bracketing is relatively small steps each side of the apparent exposure or maybe 2 above etc

HDR needs much bigger steps if needed and a number of correct exposures. I would have thought 3 - 4 stop steps or maybe even more depending on the scene.

Adjust the histogram to mimic raw? I find that there is very little head room available over the jpg histogram display so no harm in leaving it alone. It isn't that precise anyway. Some slight gradation at the bright end could easily be missed.

Each ot their own and cameras vary from make to make anyway..

John
-

[FootLoose]

It sounds like developing a more reliable histogram might be a better path for camera manufacturers rather inventing gazillion-megabyte sensors

[Glenn NK]

It sounds like developing a more reliable histogram might be a better path for camera manufacturers rather inventing gazillion-megabyte sensors

Oh boy have you got that one right.

This point is made in the link by Guillermo Luijk.

They give us an RGB histogram, then unless you shoot JPEG, it's not very accurate.

Go figure.

[jcuknz]

True, and it worked for film because the response of film was not linear. When digital clips on the high end the result is no detail, no colour, nothing at all.

With digital, the best indicator of clipping is the histogram - the trick (if shooting RAW) is to get the in-camera JPEG which displays on the LCD to "tell the truth".

If shooting JPEG, then the histogram on the camera is a very good tool.

The UniWB approach puts a JPEG on the camera that matches the exposure of the RAW.

Glenn

I never look at the histogram and rely on 'blinkies' to warn me of blown highlights with digital having read the Luminous landscape article way back and other advice sources. I used to have my camera set to one stop under-expose but not since I use MFT. One of the advantages of live view, proper live view, is the facilitiy to have a short review in the EVF following the exposure which is long enough to see the blinkies [ 1 second ].

But I guess watching the histogram is a good idea for some. I just don't think I need to.

While I shoot raw plus jpgFINE these days [ credit/blame CiC ] I have yet to need to use a raw file so far ... but I probably look for different things in a photograph which was brought home to me recently when somebody was concerned about a colour cast in their photo while I immediately saw a bad composition.

In the 'environmental portrait' I would have thought the person was the important aspect and blown skies, dark foliage is neither here nor there. When I got the blinkies I would adjust exposure or change composition depending on how much of the composition they occupied.

[ajohnw]

I never look at the histogram and rely on 'blinkies' to warn me of blown highlights with digital having read the Luminous landscape article way back and other advice sources. I used to have my camera set to one stop under-expose but not since I use MFT. One of the advantages of live view, proper live view, is the facilitiy to have a short review in the EVF following the exposure which is long enough to see the blinkies [ 1 second ].

But I guess watching the histogram is a good idea for some. I just don't think I need to.

While I shoot raw plus jpgFINE these days [ credit/blame CiC ] I have yet to need to use a raw file so far ... but I probably look for different things in a photograph which was brought home to me recently when somebody was concerned about a colour cast in their photo while I immediately saw a bad composition.

In the 'environmental portrait' I would have thought the person was the important aspect and blown skies, dark foliage is neither here nor there. When I got the blinkies I would adjust exposure or change composition depending on how much of the composition they occupied.

That's what I do mostly now and often process the jpg rather than the raw. I read something on the OMD EM-5 that suggest it will do something like that in live view. I'm still working my way through the menu's. It can set the over and under exposure indication limits.

As to jpg histograms - usual problem people are not aware just how many stops cameras stuff into jpg's or what the actual tone curve used covers so they read something some where and do it rather than find out for themselves.

Prejudices come into it as well. I have one, justified in my view. I tend to favour centre weighted metering. I find evaluation metering doesn't always avoid washed out highlights and can easily give incorrect exposures under a number of circumstances. Centre weighted will at least give a passable exposure on the subject. A good place to start from. The rest is up to me. What ever people do I feel that last aspect always crops up in the end.

John
-

[bambleweeney]

If I'm shooting landscape I'll use a hand held meter and depending on the direction of the light it'll either use incident light metering which nails the exposure 99% of the time or a spot meter I spent some time calibrating a while back. The incident meter you just kind of wave it around at the scene you're taking, the spot you measure the brightest point in the scene and for my meter I add 2.3 stops to get the highlights high without blowing out. Effectively ETTR I guess but the highlights are good and the shadows retain detail. I usually spot the darkest part of the scene to check the dynamic range and decide whether HDR is a goer or not. Call me a luddite if you will but if I've taken the time to get my sorry ass out of bed and drive to a location I figure it's worth the effort.
Oh yeah, and bracket like crazy

[ajohnw]

Just for a for instance I exposed this to ensure that the rather subtle detail in the clouds was captured and adjusted it for what I thought was a nice shot. On this one I just adjusted a couple of regions to show what is in a jpg. Just brightness adjustments to show that light levels can be recovered. This is about the limit from a jpg off this camera.



Out of curiosity I decided to see what was in the dark end of raw - but unfortunately didn't bother about keeping the sky - which in this case would be tricky due to the use of using a very extreme tone curve for raw conversion mainly aimed at the stream forgetting the rest.



A better raw conversion but still some cloud lost and again a very extreme tone curve which would make that aspect difficult due to similar light levels in the church,




The raw conversions are just that. No noise reduction or anything else.

Whoops. The exif on the last one shows Fotoxx mods, looks like it didn't remove a couple of things i tried as this is not HRD or blurred. It may be tone mapped a bit.

John
-

[New Daddy]

Just for a for instance I exposed this to ensure that the rather subtle detail in the clouds was captured and adjusted it for what I thought was a nice shot. On this one I just adjusted a couple of regions to show what is in a jpg. Just brightness adjustments to show that light levels can be recovered. This is about the limit from a jpg off this camera.



John
-

I'm confused. So in that first scene, the photo is from RAW, and the two pictures-in-picture are from the camera's JPEG rendition of the same RAW? The camera automatically adjusted the brightness in creating the JPEG?