Which camera has the most dynamic range in your opinion?

Panopeeper wrote in post #9002152
Let's simplify it to the comparison between FF and APS-C cameras using the very same, FF lens. The lens projects an image with FF size, but the APS-C camera does not use the entire image. The area used by the APS-C camera has the same luminosity for both formats. Isn't it obvious, that the FF camera utilises more light, than the APS-C?

A FF camera uses more of the image circle, a Crop uses less of the image circle. It's still the same image circle. So from a DR perspective the format size does not change the DR of film, it does change the DR handling ability of a digital sensor.

Panopeeper wrote in post #9002152
From a different perspective: the larger format cameras' lenses are larger; the front glass too is larger, it captures more light (we have to think of the same angle of view, not of the same focal length, to achieve the same framing).


The result is, that the level of illumination of the film (or sensor) is the same; however, now we are talking about a larger surface, i.e. the same illumination level means more light.

Facit: the amount of captured light on the entire film or surface area is larger with the the larger film or sensor. Its relevance with film is, that the degree of magnification for a given size of result is smaller if the film is larger (and that means better quality result). With sensor: if we fix the number of pixels, the sensels are larger, the noise becomes less. If we fix the sensel size, we end up with more pixels of the same size (same pixel "quality"), and that allows for larger output or cropping of for dowresing, anyway better than the smaller sensor.

Of course all this is reasonable only when comparing otherwise identical technologies, and that is very seldom the case in the digital domain. The smaller sensors are not crops of the larger ones.

Are we talking about quality or DR? I am referring to DR. I know the quality of LF. As far as DR goes, on film format size does not matter. On dig, format and pixel density play a big part.

chino79 wrote in post #9003848
Are we talking about quality or DR? I am referring to DR. I know the quality of LF. As far as DR goes, on film format size does not matter. On dig, format and pixel density play a big part.

OK - so it's the type of film not its size that affects the DR etc. Fine.

And I think it's the type of sensor (determined by e.g. a type of technology or a sensel's photosensitive area dimension or full-well capacity etc) not its absolute size that matters.

For digital (but not for film) the size and type are highly correlated, so let's not let that fact confound the issue here If I make a print from only the central 50% of sensor A and a print twice the size from the whole 100% of the same sensor, do they have the same DR? Similarly, if I gave you two 1000x1000 crops from images of a test chart (that had e.g. 32 distinct 1/2-stop levels of gray between black and white where the white patch was just blown out) from different cameras - could you not assess the DR of the two without knowing the sensor's size?

Or are you using DR to refer to the DR of a single pixel as opposed to an image or print?

AJSJones wrote in post #9004462
OK - so it's the type of film not its size that affects the DR etc. Fine.

And I think it's the type of sensor (determined by e.g. a type of technology or a sensel's photosensitive area dimension or full-well capacity etc) not its absolute size that matters.

Yes.

AJSJones wrote in post #9004462
For digital (but not for film) the size and type are highly correlated, so let's not let that fact confound the issue here If I make a print from only the central 50% of sensor A and a print twice the size from the whole 100% of the same sensor, do they have the same DR?

If it's the same sensor, absolutely.

AJSJones wrote in post #9004462
Similarly, if I gave you two 1000x1000 crops from images of a test chart (that had e.g. 32 distinct 1/2-stop levels of gray between black and white where the white patch was just blown out) from different cameras - could you not assess the DR of the two without knowing the sensor's size?

Yes, you could.

AJSJones wrote in post #9004462
Or are you using DR to refer to the DR of a single pixel as opposed to an image or print?

Well the original question did not specify the output, though DR could be measured in a small or large print. If an image has blown highlights it does not really matter what size it is. A 10 inch print of an image will have the same DR as a 20 inch print of the same image.

chino79 wrote in post #9003848
A FF camera uses more of the image circle, a Crop uses less of the image circle. It's still the same image circle. So from a DR perspective the format size does not change the DR of film, it does change the DR handling ability of a digital sensor

The DR of the sensor depends on the noise. As the level of acceptable noise depends on the scenery, photographer, etc., the dynamic range is not a single number but a function of the noise level.

When I measure the noise and DR, I do that on pixel level. DPReview does the same, though very unprofessionally. However, there are many photograpghers and reviewers, who are always thinking in terms of print quality. The sensels of a larger sensor with the same number of pixels as a smaller sensor are larger (assumed the same technology, like surface coverage, microlenses), thus the noise will be lower, the DR greater. If the larger sensor contains more pixels than the smaller one, then when printed in the same size, the larger one can be downsized, which means pixel binning or some other method (even averaging helps), which reduces the noise on the level of created pixels, thus the DR of the downresed image is larger than the originally smaller one (from the smaller sensor).

For example DxO shows numbers relating to some uniform print size.

Panopeeper wrote in post #9004905
The DR of the sensor depends on the noise. As the level of acceptable noise depends on the scenery, photographer, etc., the dynamic range is not a single number but a function of the noise level.

When I measure the noise and DR, I do that on pixel level. DPReview does the same, though very unprofessionally. However, there are many photograpghers and reviewers, who are always thinking in terms of print quality. The sensels of a larger sensor with the same number of pixels as a smaller sensor are larger (assumed the same technology, like surface coverage, microlenses), thus the noise will be lower, the DR greater. If the larger sensor contains more pixels than the smaller one, then when printed in the same size, the larger one can be downsized, which means pixel binning or some other method (even averaging helps), which reduces the noise on the level of created pixels, thus the DR of the downresed image is larger than the originally smaller one (from the smaller sensor).

For example DxO shows numbers relating to some uniform print size.

That is all good, but if I were to under expose an image the noise will be greater. So you cant measure on Noise. I think we are getting away from the main question, "What camera has the Greatest DR?".

chino79 wrote in post #9005179
That is all good, but if I were to under expose an image the noise will be greater. So you cant measure on Noise. I think we are getting away from the main question, "What camera has the Greatest DR?".

I wondered if the definition of DR would be an issue! I think the essence of DR is the distance between the highest tone value that can be accurately recorded (i.e. just below clipping) to the lowest that can be reliably distinguished from the noise at the bottom end. So some way of determining how noise plays a role in measuring DR has to be included, right? Once we settle on an agreeable definition then the main question can be addressed.

chino79 wrote in post #9005179
That is all good, but if I were to under expose an image the noise will be greater. So you cant measure on Noise. I think we are getting away from the main question, "What camera has the Greatest DR?".

The topic was the dynamic range of the camera, as in contrast to the DR of the scenery. This depends only on the camera (sensor) and ISO, but not on the exposure. (Assumed that the firmware does not change the noise by manipulating the raw data, like the Sony A900.)

Following is a capture from the noise measurement of the 5D2 with ISO 1600, the "scenery" is a color checker card. The measurements were done on the raw channels, i.e. before demosaicing. I included the average pixel intensity and the noise in each patch. The pixel intensity is measured in EV from clipping backwards, i.e. the greater the number, the darker the channel on that patch. The noise is measured as the standard deviation of the pixel values on a selected area (like on the top left patch, marked with an orangey rectangle), in percentage of the average pixel intensity on the selection. This is in effect the reciprocate of the signal-to-noise ratio. 50% noise is SNR=2. Note, that the noise depends on the intensity but not on the channel (it is a myth, that the red or blue would be more noisy than the green; rather, the intensity is usually higher in the green channel, thus the noise is less).

MIME changed to 'text/html' | Content warning: script

Here is a sample for how I describe the DR characteristics of a camera (here the 5D2). One picks the level of acceptable noise on the y axis; the x value shows the dynamic range, depending on the ISO.

However, this is not all to consider. The appearance of the noise may further restrict the dynamic range; for example the 5D2's pattern noise does not increase the noise as standard deviation, but it makes the noise ugly.

Note, that each point is the result of an individual measurement, there is no curve fitting involved here.

MIME changed to 'text/html' | Content warning: script

According to Clarkvision.com -

Scroll about halfway down.

http://clarkvision.com …rmance.summary/​index.html

And if you're trying to calculate dynamic range, here's a good link. See page 8. Lower right hand corner. Pixel well depth divided by the noise. That'll give you the dynamic range number for the sensor.

http://clarkvision.com …rmance.summary/​index.html

Uh oh, here comes the graphs.

FlyingPhotog wrote in post #8987830

[GIFS ARE NOT RENDERED IN QUOTES]

yeah me too...lol good one

Collin85 wrote in post #9014355
Uh oh, here comes the graphs.

I was just looking around the room to see where I'd hange that RAW Analyze chart...

Unfortunately, it clashes with everything.

Great charts by the way. I still think that if I take a 1DIII and underexpose a shot, the noise will be worse than a correctly exposed image.
OK, so what camera has got the highest DR?

chino79 wrote in post #9017053
OK, so what camera has got the highest DR?

That's easy. D3X.

Daniel Browning wrote in post #9023099
That's easy. D3X.

Why the D3x?