When I'm writing the code and want to display multiple channels, I find a scatter plot the most useful. That preserves the "buckets".
May 18, 2023
305
-
-
I think that's for people who started out with analog video. Kinda like D-logE sensitometry curves make the most sense to people who have Zone System experience.
-
The simplest possible answer to the question of what a raw histogram can be used for is that it can reveal if my exposure might be too little or too much. Both of those scenarios can lead to images that are not satisfactory for my purposes. In a nutshell, this is about maximizing this one aspect of image quality.
Underexposed images will look too dark unless they are made brighter, but if I bring the brightness up I also reveal the inherent noise more prominently. On the other hand if the exposure was too much there will be portions of the image that are blown out to pure white and have no detail, or just as bad, some but not all of the individual color channels will be blown out, which throws off the ratio of, say, red to blue, and then the colors of that part of the image will be wrong… sometimes by a lot.
I could set myself a goal of exposing as much as possible (to avoid excessively noisy image appearance) without going even just a little bit too far.
I really like the Blackjack analogy Jim used in his article. The goal is to beat the dealer so the higher the number total from my cards, the better my chances will be - up to a point! When I used to play it the dealers seem to be magnetically fated towards obtaining a count of 20. In that case I would need 20 to break even, or 21 to win. That’s enough to make me sweat. But if I draw a count of 22 or more I will have immediately lost the hand.
I could take a cautious approach by not asking for another card once I’ve reached a count of 16 and then I won’t “go bust” but that won’t fare well against a dealer’s total of 17, 18, 19, 20, or 21. That’s the conservative approach taken by most camera meters, in spite of the fact that the in-camera histogram is proudly claiming that I am at “20”. The camera hasn’t been calibrated to take me to “20” or “21”, because it might accidentally get to “22” which is very bad. So I have to settle with a “16” image and I will probably be bumping up the brightness, so here comes the noise.
This is where I bend the Blackjack analogy’s numeric model a little: if I had a raw histogram in camera I could bring my exposure up to “20.9” which would be a “winning hand” a very high percentage of times.
-
We do not need raw histograms when shooting at higher ISOs. Is that correct?
-
If the statement applies to all digital cameras, it is not correct - sorry.
It does not apply to cameras where the sensor output is amplified according to the ISO setting before being input to the ADC. Therefore, it is incorrect for most digital cameras.
-
It's not as important at higher ISOs because you can leave extra headroom with little penalty, at the same exposure.
-
For the latest Sony, Canon, Nikon, and Leica cameras (all the latest digital cameras?), there is very little or no penalty when "underexposing" by keeping the ISO a few stops lower than the maximum possible while keeping the exposure unchanged.
ETTR makes sense for base ISO and raw histograms are necessary when applying ETTR. Otherwise, the JPEG histograms are good enough. -
I guess the word "need" muddies the issue.
So, "We do not need raw histograms when shooting" would also be true. I withdraw my response/
When Sigma introduced an AFE into some models, there was a lot of whining about blown highlights compared to the sensor headroom provided by non-AFE models. My SD9 ISO "knob" remains super-glued at 100 ... LOL.
-
'Fake ISO' is a little misleading. These settings in general conform to 12232 - they might not allow enough headroom for a full brightness range, but 12232 doesn't specify how much headroom there should be. Some Panasonics had negative headroom at base ISO.
-
Over the years I've taken my share of grief for calling them that. I remain unrepentant.
On the Sony a7RIV, when you select iSO 50, it gives you the same raw file as if you had picked ISO 100, and it dims the image in the finder and the JPEG preview image and makes the histogram lie to you. It also tells the raw developer to pull the image by a stop. You could get the same raw file by just setting the EC to overexpose by a stop.
It gets worse. On the Sony a7RIV, when you select iSO 80, it gives you the same raw file as if you had picked ISO 160! When you select iSO 64, it gives you the same raw file as if you had picked ISO 125!
-
I assume this was the definition of 'exposure'. Consider the definition of exposure that most will be taught at photography school, H=E*t. Here 'H' is exposure, 'E' is image plane illuminance, often called more simply 'intensity', and 't' is exposure time. E*t is usually explained as an 'amount of light', but lets think a bit more carefully. 'Intensity' doesn't refer to an amount of light, it refers to how intense is the flow of light, that is the amount of light per unit area, so in fact exposure is amount of light per unit area.
We can also come to the same conclusion using a units analysis. The SI unit of luminance or intensity is the lux, so the unit of exposure is lux seconds, directly from the equation above. If you look up your SI units, you find that the lux is a name for lumens per square metre. so in other terms the unit of exposure is lumen seconds per square metre.
You can carry on this type of units analysis. You find that the lumen is the unit of luminous power (which is why light bulbs are now rated in lumens). High school physics tells you that power * time = energy, so exposure is luminous energy per square metre (or any other unit area, if you adjust all the units appropriately). -
But ISO doesn't define anything about the raw file.
-
Touche. But I still think it's a shuck, for raw shooters.
-
But when shooting raw I still need to deal with ISO setting ;)
-
Yes, but the number on that setting won't tell you very much useful about what it actually does.
-
It was never intended for raw shooters, that's the problem.
-
For the latest Sony, Canon, Nikon, and Leica cameras (all the latest digital cameras?), there is very little or no penalty when "underexposing" by keeping the ISO a few stops lower than the maximum possible while keeping the exposure unchanged.
<>Interesting how things change. In my world, lowering the ISO increases the sensor exposure, not "underexposes". But then, in my world, it is not possible to change the ISO without changing the exposure ... the sensor exposure, that is.
-
Lowering ISO does not change exposure if you keep exposure fixed, as I wrote above.
Lowering ISO can increase exposure if you let the camera change the exposure automatically.
