Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: What is "native ISO"?  (Read 21863 times)
Phinius
Newbie
*
Offline Offline

Posts: 31



WWW
« on: January 08, 2009, 11:29:23 AM »
ReplyReply

Michael recently used this term while discussing noise management and dynamic range with the new Sony A900. I always assumed the the lower the ISO the better, but perhaps this isn't true. Can anyone shed any light on what exactly "native ISO" is?
Logged
brianrpatterson
Newbie
*
Offline Offline

Posts: 34


WWW
« Reply #1 on: January 08, 2009, 11:33:44 AM »
ReplyReply

My assumption is the it's the default ISO - 200 ISO on the D300 even though it can go lower. It probably represents the fastest sensor setting that produces top quality images - lower ones affect contrast but are no better for IQ. Haven't seen this term in any spec lists so imagine it is a user-based description...

Here's a table generated to show some info on this subject.

OK, I've bared my ignorance - somebody correct me!
« Last Edit: January 08, 2009, 11:38:21 AM by brianrpatterson » Logged

Brian Patterson
Knoxville, Tennessee
BRP Marketing Design
BRP Publications [Print-on-Demand]

Nikon D300 | Nikon D40
spidermike
Full Member
***
Offline Offline

Posts: 127


« Reply #2 on: January 16, 2009, 04:25:23 AM »
ReplyReply

I have hunted for this before it seems to be a pretty undefined term, with its definition being made of two parts: comparing the level of saturation etc with a film (yes, film. Have we really progressed?) picture; and secondly the signal to noise ratio.
So to me it seems that the so-called experts have created a term that no-one really seems to be able to explain in sensible terms.

These two sites try to give what looks like a comprehensive description. They also explain that the native ISO is not necessarily the same as the lowest ISO setting on your camera.
http://www.cameralabs.com/forum/viewtopic.php?t=4503
http://www.picturecorrect.com/photographyt...tal_cameras.htm

Consensus on this seems to be that for Canon cameras the native ISO is about 100, and I seem to recall someone saying for Nikon it is about 200. Apparently DPReview has some information on native ISOs for different cameras.

Logged
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #3 on: January 16, 2009, 01:31:51 PM »
ReplyReply

Native ISO is the highest such ISO of a camera, at which the analog signal is not amplified before the conversion in digital.

Notes:

1. The "highest such ISO" qualification is necessary, for there are some cameras, which offer an ISO step under the native ISO. For example the 5D2's ISO 50 is identical to the ISO 100.

2. The native ISO is not selectable with some cameras. It can be for example 85. In some cases this is called simply ISO 100 (i.e. cheating). In other cases the result will be adjusted to resemble more closely the "real ISO sensitivity".
Logged

Gabor
Graystar
Guest
« Reply #4 on: January 16, 2009, 03:07:04 PM »
ReplyReply

I believe the native ISO is the point where gain = 1.

If you have your camera set to manual mode and you increase the ISO setting, then the resuting image will be brighter.  Now, you didn't increase the shutter time or open the aperture, so it's the same amount of light as before.  The difference is that the light signal has been amplified (the gain.)

Amplifying the signal also amplifies any noise in the signal.  It also reduces the range of brightness that the camera can capture (dynamic range.)  Taking pictures at the native ISO should result in the least noise and best dynamic range.
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5182


« Reply #5 on: January 30, 2009, 04:15:41 PM »
ReplyReply

As far as I can tell, the phrase "native ISO" is used roughly for the ISO speed at which the signal from a full well gives an output at maximum level, which is usually somewhere between 50 and 200 and is h lowest "normal" ISe speed setting. Higher ISO settings will "blow out" the signal from a full well, but are intended to use when light is inadequate to give any full wells (except perhaps at extreme highlights, like reflections of the sun off water.)

Quote from: Panopeeper
Native ISO is the highest such ISO of a camera, at which the analog signal is not amplified before the conversion in digital.
This seems to suggest that some low ISO value like 100 or 200 is not amplified while others are. This makes no sense, and nor does the subsequent post's related talk of an ISO speed at which gain=1.

Why? These ideas only apply with a signal that stays in volts throughout, or charge throughout but with sensors, there is a charge to voltage conversion at some point. There is no concept of "amplified" vs "un-amplified" in that conversion process. All there is in a conversion factor, usually quoted in mV/e-.

To complicate things, CMOS sensors typically apply charge amplification in the transfer of signal from the capacitor (well) at each pixel to the sense capacitor at the edge of the sensor which always gives a larger charge in the sense capacitor than that in the photosite. Then every ISO setting has charge amplification with gain greater than one.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2879



« Reply #6 on: January 30, 2009, 08:30:12 PM »
ReplyReply

Quote from: BJL
As far as I can tell, the phrase "native ISO" is used roughly for the ISO speed at which the signal from a full well gives an output at maximum level, which is usually somewhere between 50 and 200 and is h lowest "normal" ISe speed setting. Higher ISO settings will "blow out" the signal from a full well, but are intended to use when light is inadequate to give any full wells (except perhaps at extreme highlights, like reflections of the sun off water.)

To amplify somewhat on BJL's explanation, the determination of ISO with digital sensors is very much like the process used for film, as explained in this article on Wikipedia. Exposure at the plane of the film or sensor is measured in lux seconds and referred to as H. For film, the ISO is determined by the exposure (H) needed to reach a given density when the film is developed to give a stated contrast. Push processing increases the slope of H & D curve, but does not really change the ISO value.

The ISO standard 12232:2006 for digital cameras offers five different techniques for determining the ISO value, but the saturation-based technique is based on the value of H needed to saturate the sensor, allowing 0.5 stop of headroom for the highlights. Sensor saturation is independent of any amplification applied to the sensor output. The amplification is adjusted so that the output voltage  at saturation matches the full scale of the analog to digital converter. This applies to the base ISO. When one uses ISO above the base, the sensitivity of the sensor to light does not change and the sensor is less than saturated. The amplification is increased so that the voltage again matches the full scale of the ADC. This increase in amplification is analogous to push processing of film.

Some cameras can be set to an ISO below base. With the Nikon D3, which has a base ISO of 200, one can set the camera to expose at ISO of 100, but this merely results in overexposure and the amplifier gain doesn't change. Other cameras may use a different strategy.

The Wikipedia article gives equations to relate H to shutter speed and aperture, taking account the effects of lens transmission and cos^4 falloff.

Bill
Logged
pegelli
Sr. Member
****
Offline Offline

Posts: 609



WWW
« Reply #7 on: January 31, 2009, 10:37:40 AM »
ReplyReply

What I understand is that it's the ISO with no signal amplification or weakening, so go lower and you lose dynamic range and go higher you increase S/N and thereby noise.
Logged

pieter, aka pegelli
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #8 on: January 31, 2009, 12:30:14 PM »
ReplyReply

Quote from: pegelli
What I understand is that it's the ISO with no signal amplification or weakening, so go lower and you lose dynamic range and go higher you increase S/N and thereby noise.
1. I have never heard of "signal weakening" in conjunction with image sensors.

2. The cameras with lower than native ISO I know of do never lose dynamic range at the lower ISO, but some cause metering errors, like Canon 5D2, Nikon D3X.

3. Increasing the ISO does not automatically mean increasing the noise. In fact, higher ISO (up to a limit, if associated with analog gain) reduces the noise.
Logged

Gabor
fototrotter
Newbie
*
Offline Offline

Posts: 14


WWW
« Reply #9 on: February 04, 2009, 01:05:26 AM »
ReplyReply

Ok, this is not a very scientific explanation, but this is how Canon explained it at the introduction sessions they did when launching their new 1DmkIII.
This question popped up from the audience (All pro's except for myself).

The answer sounded like this: It's the ISO setting where the camera's chip was designed for and test results are based on this ISO sensitivity. For the 1DmkIII this is ISO 200,... It's the ISO setting where you should get the best results with as well, and the ISO setting it will (try to) be using whenever possible in any (semi-)automatic mode.

Follow-up question from the audience: So with film I get finer detail if I use lower ISO,... How does that work with this camera? Can I use lower than 200 ISO?
Answer: Yes, you can use lower ISO, but with every step down you will also move away from the design-ISO and will have a ("minor") loss in image quality.

They never went into further detail so cannot elaborate more, but to me the gain-factor being 1 at this point, sounds plausible.

greetings,
Steven
Logged

Learning amateur with a canon 7D, 400D. Portfolio at http://www.fototrotter.com
pegelli
Sr. Member
****
Offline Offline

Posts: 609



WWW
« Reply #10 on: February 04, 2009, 03:24:56 AM »
ReplyReply

Quote from: Panopeeper
1. I have never heard of "signal weakening" in conjunction with image sensors.

2. The cameras with lower than native ISO I know of do never lose dynamic range at the lower ISO, but some cause metering errors, like Canon 5D2, Nikon D3X.

3. Increasing the ISO does not automatically mean increasing the noise. In fact, higher ISO (up to a limit, if associated with analog gain) reduces the noise.

Maybe my understanding of this is not yet complete, but this is what I have learned sofar from reading other forums:

1: Maybe a confusing term, what I meant is that the way ISO below native is implemented in KM and Sony (don't know about the other brands) is that for ISO 100 vs. native ISO 200 they basically do a +1EV exposure correction at the native ISO and then devide the signal by 2 (i.e. reduce 1 stop again).

2: This is why in this set up you lower DR. An oversaturated pixel will still be oversaturated with the signal devided by 2. If you don't need the highlight headroom it can still be a good option, because it still increases S/N in the rest of the picture (especially visible in the shadows)

3: Don't understand, never seen this before, can you explain more as I would like to understand this better.
« Last Edit: February 04, 2009, 03:31:33 AM by pegelli » Logged

pieter, aka pegelli
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #11 on: February 04, 2009, 01:30:49 PM »
ReplyReply

Quote from: pegelli
what I meant is that the way ISO below native is implemented in KM and Sony (don't know about the other brands) is that for ISO 100 vs. native ISO 200 they basically do a +1EV exposure correction at the native ISO and then devide the signal by 2 (i.e. reduce 1 stop again)
I don't know how KM cameras are doing this. However, I do know that neither the Sony A700 nor the A900 is doing this. Actually, the DR @ ISO 100 is somewhat greater than @ ISO 200 with both cameras, but one needs to know how to expose: the exposure @ ISO 100 should be only 0.5 EV higher than @ ISO 200 with the A700, and only 1/3 EV higher with the A900.

Quote
This is why in this set up you lower DR. An oversaturated pixel will still be oversaturated with the signal devided by 2
Even if it worked this way, that would not reduce the dynamic range. The dynamic range does not depend on how you are arriving at a certain exposure. The camera's metering is incorrect in these cases in principle, but it may be the best by chance (i.e. by circumstances not recognized by the metering); beside, you can use exposure bias or even manual metering.

So, it boils down to the correct exposure.

However, there are cases, when digital adjustments cause loss of DR:

1. the fake ISOs in the high range, as adopted by DSLRs: doubling the pixel values causes clipping those pixels, which are in the top EV but not clipped,

2. some cameras create 1/3 stop ISO values from the next lower respectively next higher full stop ISO. The x-1/3 ISO reduces the number of levels (not an issue with 14bit anyway), but the x+1/3 ISO "outmultiplies" the pixels in the top 1/3 stop.

Quote
Don't understand, never seen this before, can you explain more as I would like to understand this better.
I am working on a complete set of demonstration; I will post it soon.
« Last Edit: February 04, 2009, 01:32:05 PM by Panopeeper » Logged

Gabor
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad