"Native" ISO speeds for sensors?

gooble wrote in post #5755104
My definition? The definition of the word NATIVE in this context. Look it up.

Didn't you read my comparison of the monitor? To have multiple "native" attributes is an oxymoron.

OK. I give up. I was trying to get a consensus definition for the purposes of this discussion.

Forget it.

Out of here.

-js

number six wrote in post #5755039
By whose definition? Yours? Canon's?

That was the point of my post. We need a common definition or this discussion is pointless.

The output of a CMOS photo cell is a very small voltage. Amplification is always required.

-js

six is right here, all ISOs are actually amplified. in fact, all cameras except the mkiii amplify throughout the sensor use, which is why you get hotspots with long exposures (mkiii amplifies after the fact on long exposures, which limits heat generated by the sensor). here's my proposal for "native iso", which i hope people agree with

"native iso- the iso equivalent of the lowest possible amplification level for a sensor)

that would mean software ISOs (50, 3200) do not count.

also, here's another for your enjoyment

"natural iso- the iso equivalent of possible electronically amplified levels for a sensor"

that would mean 100, 200, 400, 800, 1600 (and 3200 for mkiii), but not 50, 125, 160, 250, ..., 3200 (except mkiii), 6400 (in mkiii)

basroil wrote in post #5756983
six is right here, all ISOs are actually amplified. in fact, all cameras except the mkiii amplify throughout the sensor use, which is why you get hotspots with long exposures (mkiii amplifies after the fact on long exposures, which limits heat generated by the sensor). here's my proposal for "native iso", which i hope people agree with

"native iso- the iso equivalent of the lowest possible amplification level for a sensor)

that would mean software ISOs (50, 3200) do not count.

also, here's another for your enjoyment

"natural iso- the iso equivalent of possible electronically amplified levels for a sensor"

that would mean 100, 200, 400, 800, 1600 (and 3200 for mkiii), but not 50, 125, 160, 250, ..., 3200 (except mkiii), 6400 (in mkiii)

Yeah, I agree with that.

I don't think I was trying to say there was no amplification, just that using the term native implies an inherent or original ISO. There can't be more than one original. That's just how I see it but I'm open for any arguments otherwise.

Also, maybe nobody read Chuck Westfall's words that I posted earlier (BTW, he's something like Canon USA's head PR guy so he knows a thing or two) where he said that to get the best IQ and greatest dynamic range it's best to use ISO 100 to 200 which suggests that the native ISO lies somewhere between those two.

C2S wrote in post #5754916
(Sorry about this slightly off-topic question)

Gui: Since the difference between ISO 100 and ISO 1600 is 4 stops, and DPP (for example) only allows up to +2 stops of exposure correction, how did you boost the exposure in the ISO 100 shot by +4 stops? I once tried this myself, but I only boosted an ISO 400 shot by +2, since I don't know how to boost ISO 100 by +4... or did you just set RAW-tab EC to +2, and then modified RGB-tab curves so that it looks approximately as if still brightened up by additional 2 stops? Maybe I missed something.

I'm not disagreeing with you at all, just curious.

It's simple: I don't use DPP but my own software .

I always found the limitation of most RAW developers in terms of exposure correction quite stupid, since correcting exposure by any amount of f-stops is mathematically trivial (just multiply or divide the linear levels by the corresponding power of 2).

In this case I just did a linear development so that a straight curve in PS defined by points (0,0) and (16,255) did the job of pushing the ISO100 shot to the same level of exposure as the ISO1600 shot (16 is 256 divided by 2^4).

Me and other 2 guys are now writing a new RAW developer called Perfect RAW, based on DCRAW's engine, and we have decided to set up an exposure control from -8EV to +8EV. +8EV will blow away any image and -8EV will turn black any image; we also set an optional highlight preservation feature to be used at the same time as the exposure control anyway, both for exposure increase and reduction. This allows to lift the shadows without loosing the highlights or reducing overall exposure without loosing pure white saturated highlights.

HTTP response: NOT FOUND | MIME changed to 'image/png'

fxk wrote in post #5754875
I'm just guessing here - I may be WAY off base...
I woud expect the "in between" speeds to work more or less like a linear volume control - base ISO is zero, a full stop above base, is one, two full stops above is 2, 2 1/3 stops above base is 2 1/3, 2 2/3 stop above is also linear.

I can't see why one can only boost in full stops, and then add 1/3 or go to the next full stop and subtract 1/3. Does not make sense to me.

It seems the reason is economic - cost control. The 1D series cameras have two stages of amplification, first whole stops and then 1/3 stops. The cheaper models replace the second amplifier with software manipulation.

First, Highlight Tone Priority is not dynamic range extension. It is exactly the opposite, DR reduction. It works by underexposing by a full stop - when the camera is set to 2oo the actual ISO is 100 - and then software pushing the shadows up by a stop and the highlights only slightly. The result is that by underexposing the highlights you make it harder to blow them - you have more headroom - but since you have all that unused headroom you are using less of the available DR. That is why there is no 1oo. The camera can't underexpose by using a lower ISO.

PacAce wrote in post #5752203
I highly doubt that lower the ISO below the native ISO will generate noise. AAMOF, it should reduce the noise even further. What it will do, though, is reduce the dynamic range of the sensor by approximately however many stops the ISO is decreased by.

Of course, I could be wrong, too. But that's just my 2 cents.

Youre right. Using ISO's that are lower than the native ISO speeds that the camera is capable of will generate less noise (but I dont know of any noise at 100 ISO...) at the cost of losing headroom.

tzalman wrote in post #5757380
First, Highlight Tone Priority is not dynamic range extension. It is exactly the opposite, DR reduction. It works by underexposing by a full stop - when the camera is set to 2oo the actual ISO is 100 - and then software pushing the shadows up by a stop and the highlights only slightly. The result is that by underexposing the highlights you make it harder to blow them - you have more headroom - but since you have all that unused headroom you are using less of the available DR. That is why there is no 1oo. The camera can't underexpose by using a lower ISO.

Quite right, i couldn't remember the actual name for it last night and didn't have my camera with me. I meant Highlight Tone Priority!

I've just stumbled on this post, and read it with interest, OK, may not have fully understood all the techie things. But looking at those ISO comparison graphs on the link has convinced me to leave my camera on 200 ISO permanently.

Hope you more techie people will tell me if my interpretation of all this is incorrect.

Pete

tzalman wrote in post #5757380
It seems the reason is economic - cost control. The 1D series cameras have two stages of amplification, first whole stops and then 1/3 stops. The cheaper models replace the second amplifier with software manipulation.


First, Highlight Tone Priority is not dynamic range extension. It is exactly the opposite, DR reduction. It works by underexposing by a full stop - when the camera is set to 2oo the actual ISO is 100 - and then software pushing the shadows up by a stop and the highlights only slightly. The result is that by underexposing the highlights you make it harder to blow them - you have more headroom - but since you have all that unused headroom you are using less of the available DR. That is why there is no 1oo. The camera can't underexpose by using a lower ISO.

Seems to be a very plausible explanation. Good thought. Thanks.

I determine the shutter speed and f/stop that I want and set the ISO that will give me them, but never less than 200. Since I shoot only RAW, I haven't enabled the 1/3 stop ISOs or H - they are just more to scroll through and I get the same thing when I tweak the RAW conversion, except that the converter allows 0.01 stop increments, not just 0.33 like the camera. So whether your interpretation is correct depends on your shooting style, quality demands and subjects. Staying at ISO 200 seems to me very limiting.

gjl711 wrote in post #5751859
Found the link. Has some nice graphs showing noise at different ISO settings.
http://forums.canonpho​togroup.com/showthread​.php?p=5225

And here's the link for the same plot for the 30D.

http://forums.dpreview​.com …19&thread=19721​647&page=1

There is some interesting info in this thread - particularly by John Sheehy.

I'd personally like to know who the pinhead was that started calling it "Native ISO".

Budley007 wrote in post #5759365
I'd personally like to know who the pinhead was that started calling it "Native ISO".

Hey, even Bob Atkins uses it ...
http://www.bobatkins.c​om …non_eos_40D_rev​iew_6.html

although he does put it in "quotes"

I don't see where the debate is. The "native ISO" should be what the CMOS sensor was designed to do without further amplification or mucking with the data. Of course they all have some signal processing performed, we're talking electronics here anyway. For example, Canon bodies are "typically" ISO 100. Using ISO 200 implies additional gain/amplification.

Double Negative wrote in post #5759622
I don't see where the debate is. The "native ISO" should be what the CMOS sensor was designed to do without further amplification or mucking with the data. Of course they all have some signal processing performed, we're talking electronics here anyway. For example, Canon bodies are "typically" ISO 100. Using ISO 200 implies additional gain/amplification.

I agree but am not sure the native ISO is 100. Based on the results of the tests and Canon's comments I think it's between 100 and 200. Most of those graphs show the lowest noise at 160 which is between those two.