Camera Companies: LYING about ISO ratings in digicams

[Ron Parr]

Your condescending tone is misplaced because your argument is based
on a fallacy. The fallacy is that "it was always like this," or
something to that effect. Which it wasn't: I never had a problem
with my Velvia 50 really shooting like 100 or like 25. I never had
to buy whole pallettes of film to get "good runs."

Nobody has to buy whole pallettes of digital cameras to get good ones either. (Well, let's ignore EOS 20D focusing issues for now. ;-)

Each medium has some quirks and each medium has a certain level of consistency once you accept the quirks.

People do argue about the true ISO of velvia too:

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=000Ai0

And especially now with digital, if their ISO is not really 100 but
it's 175, then say 175, so we know. It's not like with film, where
it would cost extra to reprint all the film boxes to reflect the
actual ISO of a run, if it were off by some amount. With digital,
you can just reprogram the text that appears in the menus, so that
it says 175 instead of 100, and it doesn't cost anything.

Why should camera companies lie about their ISOs? Why should review
sites not test the ISOs to see how far off they are? Just because,
according to you, we are addicted to computers, hesitant to use our
equipment, and are lost souls, or something?

I too would like cameras to be more consistent between makes and models, but the consistency within particular models seems as good as film, and the consequences of variation are very easy to deal with. It's a total non-issue from the perspective of getting properly exposed shots once you've spent a few moments understand the quirks of a particular model.

The main issue arises when comparing between cameras and one model can give the impression of having better noise performance than another (or worse) because the graphs use the stated ISO on the horizonal axis and not some normalized sensitivity.

It would be great if Phil could measure the "true" ISO, but doing this in the real, officially sanctioned by ISO, way is actually a very complicated thing.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[dark goob]

I don't understand. If people were forced to adjust things, based upon experience, for their different choices of film and lenses, how is the situation with digital different?

It's different, because with digital, there is a maximum ISO that you cannot go higher than, unlike with film. With film, I buy a camera, and then I can put any type of film, any ISO that I want. If I want 3200 then I can buy some TMAX 3200 and I'm good to go. If that roll of film is actually 2800 ISO, then I'm only out $6.99 or whatever.

But with digital, I have to live with whatever my highest ISO is. And if the company lies about its ISOs, then I'm buying a camera based on false advertising and getting screwed out of my money.

Especially as a salesperson of cameras, I don't want to sell someone something and basically be lying to them like, "Oh yeah, it can shoot at ISO 800" when really it's only capable of ISO 550.

Lets say I sell someone a Sony F828 camera on the premise that it can shoot at ISO 800. But really, the highest it can go is ISO 550, because Sony lies their butts off about their ISOs. So my customer then returns the camera because their shutter speed at f/2.0 and ISO 800 is not nearly as fast as it should be, and so they are unable to use the camera in the way that its statistics would lead you to believe that you could use it: for low-light photography without a flash.

Does it make sense to you now?

-=DG=-

 

[dark goob]

It would be great if Phil could measure the "true" ISO, but doing this in the real, officially sanctioned by ISO, way is actually a very complicated thing.

Why is it so hard compared to any of the other complex tests that Phil does?

All you do is take a digital lightmeter and meter a scene at ISO X, and f/stop Y. Then you shoot that picture with the camera being tested at the same ISO X and f/stop Y. If the camera's shutter-speed is different than the lightmeter's recommended shutter-speed, then you know that it is off from the rated ISO standard by Z number of stops.

That's assuming, of course, that the lightmeter is accurate... which as far as I know, the good ones are.

-=DG=-

 

[Ron Parr]

I agree that it's a problem when comparing different cameras. If this is your main point then I'm with you. In fact, I've argued this very side in other threads.

I don't think it's a practical problem in getting the correct exposure after you've made your choice. You just adjust your scale and then work with it.

Perhaps the problem is that the distinction between these two wasn't entirely clear in what you had written - at least not to me.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Ron Parr]

Why is it so hard compared to any of the other complex tests that
Phil does?

All you do is take a digital lightmeter and meter a scene at ISO X,
and f/stop Y. Then you shoot that picture with the camera being
tested at the same ISO X and f/stop Y. If the camera's
shutter-speed is different than the lightmeter's recommended
shutter-speed, then you know that it is off from the rated ISO
standard by Z number of stops.

That's assuming, of course, that the lightmeter is accurate...
which as far as I know, the good ones are.

This experiment only tests the meter and not the sensor. Two cameras could agree with the meter, but one could produce a totally black image and the other a totally white image for the selected exposures.

A better, but still flawed, way to do this would be to pick a standard set of lighting conditions and then define a "correct" output value. The ISO could then be adjusted with respect to this reference value. However, there are still two problems with this:

1. Different cameras produce different tone curves, so if you pick, say value 128 to be correct, cameras A and B could both hit 128 while for the same real world image, A could have blown highlights and B could have black shadows because of the different tone curves.

2. This isn't actually how ISO defines the ISO rating for digital cameras. Following ISO's steps, requires: 1) calibrating the camera to a standard tone curve, 2) taking a long sequence of photos, 3) measuring SNR in these different photos and determining when SNR crosses a certain threshold.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[beyondalldoubt]

Ohhhh Nooo!!!! So you sell things that you basically know nothing about? What a self-demeaning admission! The proof is in the pudding, sir, and if one has not tested the recipes, one is in no way qualified to refer me to one pudding or another!
Surely, this must be some ill-conceived attempt at humor?

 

[Ron Parr]

Ohhhh Nooo!!!! So you sell things that you basically know nothing
about? What a self-demeaning admission! The proof is in the
pudding, sir, and if one has not tested the recipes, one is in no
way qualified to refer me to one pudding or another!
Surely, this must be some ill-conceived attempt at humor?

Ooh... I missed this when he first posted it. It's quiet funny.

OTOH, the fact that he's on dpreview and reading about the stuff that he's selling puts him ahead of the people out there who don't even bother.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[dark goob]

The issue of matching to an objective standard is difficult. The "official" procedure for doing this is quite complicated.

What's the official procedure? Got a link to a description of it or anything? Just curious.

-=DG=-

 

[Ron Parr]

What's the official procedure? Got a link to a description of it or
anything? Just curious.

It's described in the document ISO 12232:1998. If you google it, you'll find a link where you can pay to download the official document. I paid and downloaded it. It's not legal for me to redistribute it.

The short version of what must done is something like this:

• Calibrate the camera to establish a mapping from sensor output values to image brightness values. (This must be done in accordance with yet another ISO standard.)
• Take a sequence of images from a light source of controlled spectrum and brightness.
• Measure the change in signal to noise ratio in this sequence of images.
• Note where the SNR crosses two threshold values specified in the document. From the light intensity levels when the thresholds are crossed, there are equations for figuring out both the base ISO and the max ISO for the camera.

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Ian Yаtes]

Take ISO 400.

What its says is that the chemical composition of the film should permit this shutter speed and this aperture to correctly expose the image and so on and so forth. Different chemical compositions allow for fast shutter speeds in failing or low light situations.

Digital cameras borrow these numbers essentially claiming that they can correctly expose a picture under similar conditions. It's as if the cameras actually contained a fast film in the camera. It's a guide , a metaphor.

Instead of chemicals the digital cameras use signal voltage manipulation or some such device to capture the the low light. It is going to vary from camera to camera, manufacturer to manufacturer.

You see, film is a constant accross all cameras and manufacturers so the ISO number has meaning.
This cannot be the case with digital because each camera differs from the next.
--

 

[Ian Yаtes]

there is the new ISO for digitals.( just mirrors the film)
Read the link on Ron Parr's reply.
--

 

[Ron Parr]

Digital cameras borrow these numbers essentially claiming that they
can correctly expose a picture under similar conditions. It's as if
the cameras actually contained a fast film in the camera. It's a
guide , a metaphor.

It's not a metaphor. It's a genuine international standard with testing procedures specific to digital cameras.

http://www.cs.duke.edu/~parr/photography/faq.html#isomeaning

--
Ron Parr
Digital Photography FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/