In the digital camera domain, yes, 12232.
I wouldn't say "describe". It defines results, and only under a subset of conditions.
See, but not necessarily register, however. With dual gain 10,000 maybe more than the full well accommodates.
Maybe it's worth mentioning that neither of the words (photon, lightness, map) occur in 12232 or the amendment.
The notion that ISO can be changed after the shot was taken is an important one.
All reasonable points.
I especially agree that I should have mentioned that ISO can be changed after the shot was taken. If you have the raw file, you can adjust the mapping from raw values to lightness in the resulting JPEG. This effectively changes the ISO. Unfortunately, not all raw converters label this appropriately. For instance, some raw converters use the label “Exposure” for the control that alters the ISO.
I also want to mention, that I used the term “JPEG” to refer to the image produced from processing the raw data. This was a short cut to keep the explanation short. While JPEG files are often produced when processing raw files, this is not the only possible image format. JPEGs have 8 bits per channel. One can also process raw data to produce images with more bits per channel. Furthermore, there is no requirement that the resulting image be stored with JPEG compression. The image could easily be stored with a lossless compression, or no compression at all. Images can be stored in a variety of formats. TIFF, PSD (Photoshop), PNG are commonly used image formats.
This sentence confused me a bit at first. That was because of this previous sentence:.
They both use the term "mapping". I guess the distinction is that in the sentence above it is "mapping" photo counts. In the other sentence in which no mapping occurs you are mapping "lightness" from sensor to jpeg. I guess that's because there is no such thing as "lightness" at the sensor. It is only photon counts as stated in the earlier reference to mapping. Right?
Perhaps the problem is that I may not fully understand the process of "mapping".
By the way, I should have included in my previous post a "thank you" for including some personal information about yourself under "details". I have done the same and written about it in a separate thread related to anonymous posting -- which received some push-back.
The Exposure control in Lightroom produces a different effect than changing the ISO setting. Last time I checked, the Exposure control in Lr (not the one in ACR) worked the way you say. And then there is profile twist, which, if it occurs, also makes moving an Exposure control in a raw converter different from changing the in-camera ISO.
And, while I'm on the subject, the black points can be different between a shot with an in-camera ISO boost and the same tweak to Exposure in a raw converter.
My understanding of ISO as used by the digital camera is that it is a setting telling the demosaic function how to interpret those well counts.. The sensor itself is like film and has an optimum ISO value.... Settings are basically a user or meter interpretation of a preferred setting for the given light seen/measured …. am I wrong?
WhyNot
"Mapping" is a mathematical term. It's defined as an operation that associates elements of a set with elements of another set. It's easy to understand most simply for small sets. Let's suppose that we have a set of times that buses arrive at a bus stop. We can make a table of the times, then put a column next to it and make a table of the bus destinations. In mathematical terms what we've created is a relation between bus times and destinations. One important thing to note here is that 'times' and 'destinations' are completely different types of thing. You can't just apply some arithmetic to a time which would turn it into a destination.
Similarly (if we restrict ourselves to monochrome to keep things simple), one could make a table of all the charge levels in a pixel and and then make a second column which gave lightness in an output photo. That's what is commonly called a 'look-up table' or 'LUT' and again defines a relation between two different types of thing - though in this case it is usually possible to express the relation by means of arithmetical operations.
By demosaic function, I'll assume you mean the raw developer. It depends on the camera. Most cameras don't work that way for most of the ISO range. Most cameras scale the raw data using analog techniques, digital techniques, or both before writing it to the file for at least part of the ISO range.
Let's just say 'not completely right'. But not completely wrong, either. First, fundamentally all ISO is, is a guide to setting exposure - in fact it's inversely proportional to the exposure that it's recommending you to set. It's not the film that had an optimum exposure (i.e. optimum ISO) - it was the film/development combination - the way you developed a film affected the optimum exposure. Also, film had a limited region where the density was linearly proportional to exposure, and finding 'correct' exposure meant getting the range of exposures in the photo spread over that linear region. Digital has a much larger linear region (enough larger that sometimes we forget that the top and bottom are in fact not linear), so 'correct' exposure in that sense matters less. What you're effectively doing is using the large exposure latitude to provide the illusion of having different speed films available.
I guess my point is that the sensor is basically a photon counter and ISO setting is not a factor (just like film use to be.) The ISO as seen by the user setting it manually or by meter is a little like a volume control that is used in producing what the user sees on his LCD or monitor after downloading ...
WhyNot
You mean that ISO setting is not a factor in the number of photons counted? Ignoring clipping introduced by the electronics and ignoring read and quantization noise, that is correct.
Yes .. My thought was that for the photographer using a camera it is better and necessary to simplify that thought process to what he needs to consider when making that adjustment ... Noise introduced by random photons and heat will get amplified along with other counts for presentation as he adjusts away from the sensor optimum ( with film we could select different optimum ISO values) Clipping is simply overflow of the well ... I like to simplify the process to basics ..
WhyNot
.
In modern CMOS cameras, clipping can occur at several places in the image processing chain.
You mean away from base ISO? True at the output of the ADC, but input referred noise can drop as the ISO is increased.
GOSH
( film was better )
( * _ * )
Well ….. Noise is introduced at the well photon count … any change in that is due to Signal to Noise processing and other processing introduced in conversion from the Raw count to the RAW digital that is shown to you for processing considerations … (are you suggesting that the signal value changes due to additional noise along the processing path and not introduced by the processing of that signal?) ….. but then again I have to remember that I am technologically illiterate as I haven't looked at an equation in 30 years!!! …. and this conversation is getting too technical for me … I have to keep it simple or I have trouble taking that picture. …
WhyNot
The photon count is indeed noisy. The noise statistics are Poisson, which means that the more light, the more noise, but the more light, the better the signal to noise ratio. For a lot of photography, you can forget about all other sources of noise, since noise adds as the square root of the sum of the squares of the components, meaning that the dominant component tends to determine the noise. .
Introducing read noise makes for a lot of complications, especially with dual conversion gain sensors.
'Well' is another term that I think is unhelpful in this discussion. 'Pixel photon count' probably means more to most people, and in most of the cameras you use today the saturation photon capacity has very little relation to the full well capacity. And of course it's an electric potential well, not a gravitational potential well, which a lot of non-electronics types don't get either.
