Exposure Compensation

[Geoff F]

Canon cannot do it because the EOS Cameras don't have a dial available to use for EC when the Camera is in Manual Mode.

It is most annoying as therefore one cannot use EC whilst using Auto ISO and with the Camera in Manual Mode - a useful combination that one can do with (some) Nikon.

This is noted this previously in a couple of threads here at CiC and also I've been one of a few who have "complained" to Canon via CPS.

No developments in that regard.

WW

True, but my Canon cameras do have an exposure scale which shows the amount of over/under exposure so it is relatively easy to tweak the adjustments until the scale reading is what I require.

Not auto though so you have to check/adjust for each scene which tends to be a bit time consuming and problematic for those quicker shots with changing scenes.

[Saorsa]

Nearly forty years ago now, I was a professional mariner. Back then we learned dead reckoning navigation that required an understanding of compass error, course over ground vs heading, etc. In recent years it's been appalling to me to see people set out on the sea without that fundamental understanding and relying completely on electronics. But in an age when you can have multiple layers of redundancy plus ability to call for help from anywhere in the world, what does it matter?

And so it is with photography. As long as a person can produce the desired imagery, what does it matter how they get there? Us old timers just need to get over it and move on. Nobody cares and it doesn't make any difference anyway.

Not to mention keeping a plot on a paper chart.

[NorthernFocus]

Not to mention keeping a plot on a paper chart.

Plotting? You should have yourself carbon dated. Surely you are a relic

[Saorsa]

Taking a yacht into Port Canaveral the owner was surprised that we arrived before daylight. My crew and I had raced together and got her trimmed into a groove that took us down the coast pretty quickly.

We were furling the staysail and rollerfurling jib when he pops up to ask where we were. We told him it was the entrance to Port Canaveral. He whipped out his Laptop, connected to the GPS and checked his chart. The GPS hadn't found the satellites yet and he insisted we were much further north.

I pointed out the gantries of the space center. Then I pointed out the pretty red and green flashing lights marking the channel in.

Until they come up with a GPS with a 30 inch screen and a compass rose I'll keep a chart on the nav table. Oh, and my bronze pirate as well.

[NorthernFocus]

Taking a yacht into Port Canaveral the owner was surprised that we arrived before daylight. My crew and I had raced together and got her trimmed into a groove that took us down the coast pretty quickly.

We were furling the staysail and rollerfurling jib when he pops up to ask where we were. We told him it was the entrance to Port Canaveral. He whipped out his Laptop, connected to the GPS and checked his chart. The GPS hadn't found the satellites yet and he insisted we were much further north.

I pointed out the gantries of the space center. Then I pointed out the pretty red and green flashing lights marking the channel in.

Until they come up with a GPS with a 30 inch screen and a compass rose I'll keep a chart on the nav table. Oh, and my bronze pirate as well.

That's funny.

I still have to have wide area charts to "keep my bearings" and for trip planning. But gave up plotting several years ago. Many people no longer even put a compass on the boat. I still have to have the old magnetic compass at the helm as the basic frame of reference from which everything else begins. I guess that's a good analogy to the exposure triangle in photography. It's hard to function until one has a basic understanding of that.

[Inkanyezi]

Until they come up with a GPS with a 30 inch screen and a compass rose I'll keep a chart on the nav table. Oh, and my bronze pirate as well.

I have sailed on a commercial vessel, using the GPS, but still plotting. No matter how large the screen, I'd always plot and keep dead reckoning. Redundancy is king. When I sailed a 50' yacht from France to the Scillies, and then from Hugh Town to Crosshaven, my partner checked the satnav now and then, but we arrived perfectly on compass course, log and dead reckoning. I think it is wise to keep plotting, even if you get that 30" GPS screen.

[Saorsa]

That's funny.

I still have to have wide area charts to "keep my bearings" and for trip planning. But gave up plotting several years ago. Many people no longer even put a compass on the boat. I still have to have the old magnetic compass at the helm as the basic frame of reference from which everything else begins. I guess that's a good analogy to the exposure triangle in photography. It's hard to function until one has a basic understanding of that.

I am always amused by those who think the compass is turning.

[xpatUSA]

Plotting? You should have yourself carbon dated. Surely you are a relic

As are I . . . . .

[William W]

I also like redundancy. Indeed in all of life, having Plan B and Plan C is a good idea, even Plan D, I have used once or twice I am sure . . . and I know that I failed miserably when I only had Plan A or worse, no Plan at all.

Back to Photography - on another forum that I attend, we see a lot of 'first time' Wedding Photographers asking questions about their gear . . . they have one camera and (sometimes) one Flash . . . but the emphasis of their question is on which very expensive lenses to use. . .

WW

[Inkanyezi]

Oh, maybe the best then would be to stay away from it?

Can I join the club of geezers?

[xpatUSA]

I do that sometimes too Ted. 2 reasons - more saturation out of the box which ever way that can be easily adjusted up, flower colour adjustment and maybe things like white swans. Also if I am definitely going to work on a jpg to help get round it's camera curve easily if needed. Generally 1/2 a stop or so. I experiment with that from time to time. On jpg's I aught to see what benefit changing the camera curve has.

John
-

Interesting. Do I understand correctly that reducing exposure increases color saturation?

[Inkanyezi]

Interesting. Do I understand correctly that reducing exposure increases color saturation?

You're about to open a can of worms Ted.

When shooting on film, both negative and positive, holding back exposure was one way to get more saturation in colours, so that they would not be washed out. Of course it shouldn't be overdone.

We carried the conception of saturation versus exposure with us into the digital age, and holding back exposure is indeed a way to keep saturation in flower shots that tend to wash out -- provided you save your images as jpeg, which in a few respects is a bit like film and standard development.

Now, when digital is maturing, the mere concept of "saturated" is different, as we use it for a primary colour that hits the roof, the maximum possible bits in a channel, with the other two low or not present at all. However, also a much darker luminosity can be saturated in a technical sense, in the absence of the other two colours.

There is a common problem associated with this, and the typical exponent is a bright red flower, as a poppy or a rose. If the petals of a red flower are close to clipping in RAW, it will definitely clip when the image is white balanced, as the coefficient for the red channel in white balancing is larger than one. When shooting jpeg, the only way to cope with this problem is to expose less, minus compensation. In RAW, it is different, as you can use the "exposure" slider to hold back all three channels, before white balancing.

Before digital cameras, there were analog cameras, both still cameras and movie. Analog cameras white balanced in a different way, as their signal was analog, balancing was done by amplification, which we still can have in our digital cameras, in which case it is called gain. I would have loved if PP software also had called the control gain instead of exposure.

There are scores of misconceptions about the concept of exposure, and also about how best to cope with saturated colour. Combined with the common misunderstandings of exposure as a whole, with further misconceptions of the ETTR idea, you have your can of worms served on a silver plate.

So -- if we conceive "saturated" as in the past, that the colour is very "deep", you can indeed have red flowers (or yellow and violet as well) more saturated by holding back exposure. That will render a "deeper" (more saturated) colour on the computer screen, or printed on paper. If on the other hand we conceive the idea of a numeric channel reaching its maximum possible value as saturated, when the other two are very low or absent, we can easily use the RAW file to create a good image, if the channel is not clipped.

The first method, exposing less and letting the camera convert to jpeg, creates an image with the saturated colour still having tonality, but as the other two channels also are lower, more noise will appear if we try to adjust the curves to make the image a bit brighter, and tonality might get lost in the darker parts, where posterisation can also take place. So there are drawbacks with the minus compensation way.

When holding back gain ("exposure") in RAW processing, we have more data in all channels, and the darker parts of the image will not be posterised. Colour of green leaves will appear more fresh as well, when we adjust curves so that darker parts get brighter (akin to Nikon's D-Lighting).

So ETTR has an edge in this situation, as well as UNI-WB, which is the technique to get camera histograms show the channels more close to true for the RAW file.

Now, there is your can of worms.

[ajohnw]

Interesting. Do I understand correctly that reducing exposure increases color saturation?

Might be a miss use of words Ted - colours are darker and I haven't measured what it produces. You have an example and it is from jpg. I find it easier to get to the correct tones this way on some things. I feel the same way about raw.

There is another effect on an E-M1. The camera uses a sort of S curve for jpg's. Some people describe it as a film curve. I sometime hold back 1/2 a stop so that the highlights can be uncompressed. The curve is shown here.

http://www.dpreview.com/reviews/olympus-om-d-e-m1/19

On film the best way to get glorious colours slide or film was to ignore the speed updates and stick to the earlier development regimes that went with them.

John
-

[xpatUSA]

Urban and John,

Thank you both for your clarification.

I am familiar with color saturation and include ColorThink amongst my 'must have' apps, hence no can of worms for me I'm pleased to say. But I was interested in finding out John's reasoning.

Y'all might find this thread on DPR of interest:

http://www.dpreview.com/forums/post/54889359

I have a post there going on about a flower shot where indeed it took a while to persuade the OP that his rose petal image was quite badly color-clipped (by which I mean out-of-sRGB-gamut, hence clipped to the gamut boundary when saving as sRGB JPEG) - but he agreed in the end.

[Inkanyezi]

In essence, professional printing always has to cope with colours out of gamut, and the standard method is to decrease brightness (aka saturation). It can be done in several ways, among them using masks made with complementary colour. But the end result always is to shrink the content of the file to the colour space where it is to be rendered.

The term "colour clipping" or gamut clipping, has been coined for the out of gamut colours, but it is not really an effect of any clipping, as it takes place in an analog space, the colour space.

What we do to cope could also be seen as making the colour "darker" by adding black, which is actually suggested by the three axis Lab space. Of course our digital files don't have any black, only absence of luminosity where you have zero data.

So in effect we decrease the value of all colours to fit within the space allowed by the presentation medium. The image becomes darker - "black added" - and colours come out as more saturated (in the old sense). Then we may increase brightness by lifting the curve, as long as we keep its white point fixed in the upper right corner. the hump lower down to the left brightens shadows, increasing contrast, and brightens mid-tones, decreasing contrast in the brighter parts of the image. Then there are also other tricks that can be used, but seldom are, as tone mapping etc.

I think ETTR and UNI-WB has a lot going for it, and the main problem with wide gamuts is that you cannot print it or display it on screen. Probably we will see more efficient tone mapping algorithms in the future, because with paper base or a monitor, the limits are too narrow for wide gamuts. We might also in the future see cameras that are better at exposing to the right without blowing out any channel, and default jpeg converters in the camera to make use of ETTR files, as well as RAW converters with such algorithms. Meanwhile it is rather easy to make one's own profiling of the converter to the camera, although it is a tad more difficult to use ETTR. I think we all should be thankful to Guillermo Luijk for his work on the topic as well as those that initiated the ETTR fad.

And there will always be those that won't subscribe to such an odd idea, when you can use the camera just as it is. After all, the dynamic range is sufficient, and just get yourself a color-rite passport or whatever it's called and profile your workflow.

Remember I told you it's a can of worms. There will be contenders of all types.

[george013]

...
There is a common problem associated with this, and the typical exponent is a bright red flower, as a poppy or a rose. If the petals of a red flower are close to clipping in RAW, it will definitely clip when the image is white balanced, as the coefficient for the red channel in white balancing is always positive.
...

That means that when they clip in the raster-image, then there is also no room anymore in the RAW-image, independent of the sensel value. Just coming back to another thread. A correction-factor of 1.1 for red in the WB gives a maximal value for the red of 238, instead of 255 as said before.

....
When shooting jpeg, the only way to cope with this problem is to expose less, minus compensation. In RAW, it is different, as you can use the "exposure" slider to hold back all three channels, before white balancing.
....

What you're saying here is that everytime I change the "exposure" slider, that happens on the RAW-data with a WB-calculations following. I'm not sure about that. However there is something strange in Capture, I can't use the exposure-slider when loaded a JPG. I'm still wondering why.

George

[ajohnw]

Urban and John,

Thank you both for your clarification.

I am familiar with color saturation and include ColorThink amongst my 'must have' apps, hence no can of worms for me I'm pleased to say. But I was interested in finding out John's reasoning.

Y'all might find this thread on DPR of interest:

http://www.dpreview.com/forums/post/54889359

I have a post there going on about a flower shot where indeed it took a while to persuade the OP that his rose petal image was quite badly color-clipped (by which I mean out-of-sRGB-gamut, hence clipped to the gamut boundary when saving as sRGB JPEG) - but he agreed in the end.

Part of my reason for trying it is down to channel clipping, not that I was getting it more that I might especially in jpg's. Thinking in terms of more extreme rgb mixes and the dreaded yellows. I found it's easier to brighten, same from raw. If dynamic range is low loosing some dark dynamic rage is irrelevant. The flower is a good example of that.

Cameras vary though so there can't be any hard and fast rules. I posted a high iso shot of a Nikon V2 in the wide open thread which is a good for instance. It tends to give a result ( so far ) that may as well have been fairly brightly lit. The raw 6400 iso shot posted is similar to the 200 ISO jpg. The 6400 ISO jpg is rather flat and muted but still "well lit". Olympus tend to reflect the state of the lighting as it was. Canon at least historically love to clip highlights in jpg's but colouration is more like Olympus. Rant - I saw a stupid comment on Dpr about Olympus and tungsten lighting - idiots, turning that off is an option in a menu called keep warm colouration for tungsten lighting. This area may well relate to why Nikon are often reckoned to over expose. They do this some how or the other to try and bring up shadows - the digital sensor problem area. If a shadow is visible under many circumstances - well lit elsewhere the camera exposure will concentrate on the bright part and give much denser shadows. Just as it needs to. Clip them and they are gone, shadows may be recoverable.

I feel uni-wb might be a nice doesn't give a clue name but has something of a problem. If a channel gets clipped correcting for colour temperature then that will need changing away from reality or the image will have to be generally darkened, In my view it would be better to calibrate the camera to ensure that it produces sensible numbers. My cameras don't seem to temperature correct raw files so what's the point of producing green jpg's other than trying to meter raw. Not that I read that carefully. It looks to be yet another red herring to me. Auto white balance gets very close most of the time for me so the clipping is going to be a problem at some point. Actually metering for raw headroom doesn't make much sense either. The spare is there for a reason - errors in the system and is probably too much on Canon. I haven't checked raw headroom but from results I know that the spare is often needed - unless I feel I might need to compensate. Next time I'm bored I think I will check it but not with a white card. I have a colour card some where. That will check what the channels are doing. White wont.

John
-

[Inkanyezi]

That means that when they clip in the raster-image, then there is also no room anymore in the RAW-image, independent of the sensel value. Just coming back to another thread. A correction-factor of 1.1 for red in the WB gives a maximal value for the red of 238, instead of 255 as said before.

Don't mix issues. The RAW data resides in a 16 bit file with up to 14 relevant bits, far more than 255. RAW data may technically utilise floating point numbers.

And don't forget, that also the RAW data refers to a "raster image" even if it is not yet turned into a colour corrected raster image. It is not the software that makes it "raster", but the fact that the sensor is a raster.

What you're saying here is that everytime I change the "exposure" slider, that happens on the RAW-data with a WB-calculations following. I'm not sure about that. However there is something strange in Capture, I can't use the exposure-slider when loaded a JPG. I'm still wondering why.

If you read the works on UNI-WB and ETTR, it might dawn upon you, that all RAW converters won't do the trick, although some do. The "exposure" slider changes the gain, and with a lower gain, the >1 factor for red in white balancing will not push it to clipping. The gain is fixed when the image has been white balanced, you cannot change it in a jpeg.

[xpatUSA]

Out of interest, I shot the Munsell mini-card over a 2 EV range of exposures. I found that the saturation (according to the HSV color model, measured in RawTherapee) did indeed increase as the exposure reduced. Some results:

(Red Patch)
Speed(frac. sec.), Saturation(%), Value(%)
1/4, 79.1, 92.2
1/6, 81.7, 77.3
1/8, 87.9, 67.8
1/10, 90.0, 58.8

The trend is clear, even though the camera was handheld (distance might have varied a bit between shots).

I also made the value of the 1/10 shot equal to that of 1/4 shot two ways:

By increasing the lightness slider, there was a significant hue shift of some 5 degrees !

By increasing the exposure compensation slider to 1.46, there was higher saturation (87.2 compared to the 79.1 above), and the higher parts of the histogram disappeared to the right.

As with all things photographic, nothing is ever simple

In this thread, please can we stop calling saturation "brightness"? It confuses my pedantic mind

[george013]

Don't mix issues. The RAW data resides in a 16 bit file with up to 14 relevant bits, far more than 255. RAW data may technically utilise floating point numbers.

And don't forget, that also the RAW data refers to a "raster image" even if it is not yet turned into a colour corrected raster image. It is not the software that makes it "raster", but the fact that the sensor is a raster.



If you read the works on UNI-WB and ETTR, it might dawn upon you, that all RAW converters won't do the trick, although some do. The "exposure" slider changes the gain, and with a lower gain, the >1 factor for red in white balancing will not push it to clipping. The gain is fixed when the image has been white balanced, you cannot change it in a jpeg.

There is no essentiel difference between 8,10,12,14,16 bits. Only a difference in accuracy. Any value within the range you can replace with a percentage. I thought have seen that RT is using percentages in stead of values in their histogram. I must say I don't know how to recall that histogram anymore. Using floating point numbers is the same. Only a difference in accuracy and programming technique. The number 254 in 8 bits is 0.996....%, in 12 bits it is 4080. Both are representing exactly the same color and luminance.

When I'm talking of a rasterfile I mean that file that's the most important in digital photography but just doesn't has an own name. It's the file created out of the RAW-info and written in memory as a RGB-rasterfile. It's the file that's written to disk or to printer. A JPG-file or TIFF-file is a DISK-file.

I can't think of another benefit of the UNI-WB then to explore the behaviour of the sensor. I don't see its relevance for an image.

George