You aren’t describing my experience. At least, with my cameras, the ISO setting changes the RAW data. While I can make the ISO 100 photo brighter than the ISO 800 photo in post, the resulting image is terrible, and has lost its detail.
Your example of ISO 100 vs 200 is probably true because one stop is very little latitude.
I think there is a misunderstanding here and maybe it's me :-) but that statement is actually correct because my understanding of the ISO standard is that it states what the effect of ISO must be but in no way does it specify how that effect is to be achieved.
As Michael Fryd correctly pointed out earlier in reply to your query:
My understanding of Michael's reply is that camera manufacturers can implement the effect of ISO pretty much as they like as long as the effect is as specified in the ISO standard.
ok, maybe I wasn't clear. Both my test shots were at base ISO, ISO100.
I haven't made test shots yet at higher ISOs to see how much the extra headroom is above the in-camera jpeg histogram clipping point. I suspect the extra headroom will still be around 1 stop maximum.
I still have 29 days free trial on my RawDigger :-)
Actually, it is not lightness. It is intensity of that color component. When you consider the other color components you may find that the brightness/lightness is not correlated to a single color component.
Let's try a simpler comparison. Let's consider two pixels in the raw file that have the same intensity value. Assume both are pixels behind the same Bayer Pattern color filter.
If pixel values indicated brightness/lightness, then we would expect that if two pixels had the same value, that the original scene had the same brightness/lightness at those points. However this is not the case. It is quite possible that the two points in the scene had different brightness.
For instance, consider two pixels that are behind red filters. Both have the same high intensity value. One also has high intensity values in the nearby green and blue pixels, and the other has low intensity values in the nearby green and blue pixels.
The former may produce a white pixel in the JPEG (r=255, g=255, b=255), while the later might produce a red pixel (r=255, g=0, b=0). Generally, we would say that the white JPEG pixel is lighter than the red JPEG pixel. This is a simple example of pixels having the same value, even though the scene had different brightness at those two points, and the resulting JPEG has different lightness at those two points.
The process of going from raw data to a JPEG image is not simple. It is not correct to say that individual raw pixel values correspond to how bright the scene was at that point.
But that's not what ISO setting is, and ISO setting isn't what ISO is.
For example, in some cameras / for some ISO ranges changing the setting to a 1/3 or even 1 stop higher doesn't change raw data in the file.
Lightness is determined through a colorimetric observer, quite often the one that is called "standard observer". Camera observer isn't colorimetric. Before the data is converted to a colorimetric observer, and there are multiple ways of doing that, with different pictorial results, there is no lightness. Raw data has is presented in a camera, non-colorimetric observer, and as such it has no no lightness.
Yet another problem is that lightness predicts perception, and before we apply (non-linear) tone curves we deem suitable, lightness and lightness ratios are not determined.
I would not dispute what you say, but think it is getting into far more depth than I would consider necessary for a "beginner". We are supposed to be in "Beginners' Questions"!
I was originally assuming that Jim Kasson's comment about raw files not containing lightness data could be taken as meaning that the data was not related to what a beginner understands by lightness. Which I think is not the case. The data in a raw file is related to the lightness/brightness/intensity/whatever of the captured image, although the exact relationship between the raw file and the final image is typically very complicated.
For a beginner’s introduction to exposure, I think it is sufficient to suggest that they think of sensor pixels as being like little buckets that collect and count photons (which can be thought of as little droplets of light). I see no need to go to far into the weeds.
Suggesting that they are related to brightness introduces doesn’t hold true when you are comparing data from one shot to the next. Take two shots of the same scene, one at 1/10 and one at 1/500, both a f/8. In both the scene has the same brightness, in both the sensor sees the same brightness, but the pixels see many more photons in the one with the longer exposure.
Yes, scene brightness is an important factor in the raw values, but it is not the only one.
There's a lot to unpack there. First, it really does matter if you set the lightness in processing or post-processing. Don't do the latter. It's in a perceptual space, and if you stretch too much you get the effects you describe. There are many cameras where you can expose for 800 ISO when 100ISO is set and get a very good result. There are some where you can't, because they have a load of read noise at low ISOs.
Not really. What's directly after the photodiode is a strip of floating diffusion, the capacitance of which directly affects the 'conversion gain' and on most modern sensors can be changed to a high or low value. After that is a source follower, after that a column amplifier and then you get to the variable gain amplifier, usually two stages. Not no these things is directly controlled proportionately to ISO setting in any camera. I say this not because it's something any beginner wants to know, just to point out that most 'ISO is gain' arguments are based on not knowing what the electronics is actually doing.
I think that it is a error to teach a beginner something that is wrong and will have to be unlearned -- or will not be unlearned -- later. There are other terms that could be used, at least one of them mentioned in this thread. Raw data is not colorimetric. Lots of people are confused on that point. Why add to the confusion? As to the "related to" part, that is an extremely loose connection, and that term was not used in the original statement, if I'm not mistaken.
Lots of luck getting thousands of potentiometers on a CMOS chip. Would be great for the ultimate in control. One problem is where to put all the knobs on the camera. 😉
Bobn2 has already explained that CMOS sensor architecture is different from the way you described it.
Good explanation. Don't forget the muxes.
Simplifications, fine. Confusions, not fine.
Correct is pretty easy to explain. What becomes difficult is explaining to people whose thinking has been influenced by false mental models.
Perhaps semantics are getting in the way and I misunderstand. It seems several of you have stated that EC does not affect the raw files.
That does not appear to be the case for my camera or any previous digital camera I have used. A very simple test will show that the raw file will be affected by a change in EC. Set the camera to shutter priority, fix the ISO and watch what EC does to the aperture setting. If you dial down the EC, the aperture will close down admitting less light which will certainly affect the raw files. Another test is to up the EC to the point of blowing out highlights. No amount of tinkering will allow recovery of those blown highlight for jpegs or for raw files.
As to brightness, lightness or even luminance and luminosity, there may be theoretical precise definitions for these terms, but it hardly matters since almost no one seems to care or be precise in using the correct word.
Set the camera to Manual Exposure mode and the exposure will only change when you change it, and the EC setting will have no effect on the Raw file.
Who said that? EC affects metering. Metering affects exposure in some modes.
I question your statement about what almost no one cares about. If the definitions of the terms used don't matter, I think this exercise is pointless.
