Equivalence and dynamic range

Started 6 months ago | Discussions thread
alanr0 Senior Member • Posts: 2,049
Equivalence: what is it for?
1

FingerPainter wrote:

alanr0 wrote:

FingerPainter wrote:

<snip>

You might want to check this out.

I have. Have you understood it?

From your link: http://www.josephjamesphotography.com/equivalence/#exposure

  • "The role of the ISO setting on the camera in the exposure is in how the setting indirectly changes the actual exposure by having the camera choose a different f-ratio, shutter speed, and/or flash power."

Original emphasis preserved.

  • "Brightness (RGB value) = Total Light Collected mapped into a particular color space and recorded in the image file."

ISO setting is relevant, but only indirectly. Brightness, and the relationship between photons captured and pixel RGB value is a function of image processing.

ISO sets the analogue gain and ADC operating regime, but is only indirectly related to the light captured by the sensor.

From: http://www.josephjamesphotography.com/equivalence/#brightness

  • "Other factors, such as ISO and pixel count / size play a minor role in relative noise compared to the above three factors."

See also: https://www.dpreview.com/articles/8148042898/exposure-vs-brightening

Note that I was referring to equivalent exposures.

Not in our post title.

Brightness and display size matter when it comes to equivalence of photographs,

Which is what I was talking about, and which is what I assume when the term "equivalence" is used unmodified by "exposure".

but they are a function of image processing, and only rather loosely related to exposure.

I was reacting to the title of your post, "ISO setting is not part of equivalence", which did not mention exposure, just equivalence.

iSO is part of equivalence but has no part in exposure (including equivalent exposures).

See links above and GB's post. The relationship between ISO and equivalence is indirect and not rigidly defined.

Consider 3 identically framed exposures from the same camera, in RAW + JPEG mode.

  1. Select average metering: f/4 exposure at 1/1000 s at ISO 1600.
  2. Blown highlights were detected. Select spot metering and capture a second shot at f/4, 1/1000 s and ISO 400
  3. Now select high dynamic range mode with 2 stops of highlight protection. Re-select average metering and expose at f/4, 1/1000 s with ISO 1600 indicated.

Out-of-camera exposures 1 and 3 have similar brightness, but are not identical. Contrast is lower in 3 and highlights are blown in exposure 1.

Exposure 2 is darker out of camera, but has identical RAW content to exposure 3. In HDR mode, the camera internally selects a lower gain than normally expected for the indicated ISO, and changes the tone curve to deliver appropriate brightness.

Exposures 2 and 3 can be processed from RAW to give essentially identical printed images.

Exposure 1 can be processed to a similar image size and brightness, but may lack highlight detail. The shadow noise will be different, but so will the noise in an equivalent exposure from a different camera.

Now tell me whether the three final images are equivalent, and if not, explain why not.

While their exposures were equivalent (indeed identical) , the final images are not equivalent, because they do not have the same brightness at all points in the frame.

So none of the out of camera images are equivalent.

By your criterion (same brightness in all points) raw processed image 2 (ISO 400) and raw processed image 3 (ISO 1600) are equivalent. Raw processed image 1 has lost highlight detail, and is not equivalent.

I see some rationale behind this approach, but if you take it to its logical conclusion it becomes very difficult to compare equivalent images from different cameras. In principle, differences in tone curve and brightness can be corrected in post-processing, and different pixel counts managed by interpolation.  However, it becomes more difficult to create identical brightness when cameras use different colour filter arrays which deviate from the Luther-Ives condition.

To my mind, the value of the equivalence concept is in abstracting the physics of the image formation process at the sensor plane from the engineering implementation of the sensor architecture, pixel count, quantum efficiency, analogue amplification, ADC design, de-mosaicing algorithm, colour balance, tone curve etc. etc.

Comparing "equivalent exposures" emphasises this separation of optical physics from engineering implementation.

Equivalence allows one to define conditions under which ideal photon-counting sensors of different sizes could (with appropriate post-processing) produce essentially identical images.

A real world comparison of different cameras would then identify regimes where equivalent images are not possible. A more refined comparison would assess how engineering details such as pixel count, QE, read noise, CFA transmission and tone curves, impact image quality.

Similarly, for a particular camera, there is no reason to restrict the choice of ISO when setting the exposure.

Just my 2c.

-- hide signature --

Alan Robinson

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow