Is dynamic range a camera spec?

Helvegr

I've been continuing my learning and had been learning more about Dynamic Range. I've also heard that medium format camera, have more of it. Which is one of the reasons for their $$$$ price tag.

Just curious, is the dynamic range of a camera sensor translated as some type of spec? like a contrast ratio or something?

I was just wondering if there was something more quantitative about it between cameras. If you compared a powershot point and shoot to some 80MB Phase One camera, how would know the range of each?

Hope the question makes sense.

Thanks!

ziggy53

Dynamic range can be relatively qualified, meaning that a dynamic range measurement from one review site may not directly compare to the dynamic range measurement from another review site.

Dynamic range is generally measured in terms of "stops" and the measurements are generally limited at the top by when highlight detail is obliterated and at the bottom when shadow detail is equaled by random sensor noise. Manufacturers are now somewhat sensitive to the reviewers measuring dynamic range and even RAW files will often have some noise reduction applied to the image data to improve the dynamic range numbers.

Since all color sensors have multiple color channels dynamic range is also affected by white balance and the individual sensor's sensitivity at the particular white balance and the available color spectrum.

In other words, there is no singular dynamic range definition and no singular method of measurement. All of that makes comparisons risky between different reviewers and sources. As usual, you cannot depend on accurate data from the manufacturers unless they specify the exact methodology and you understand the differences between methodologies.

Helvegr

Great explanation, thank you. When it comes to typical gains in range, are there some "standard" jumps that are agree upon? For example, is there a jump going from say a point and shoot, to a DSLR, or a jump from a crop sensor to a full sensor? Or maybe to a medium format? Or are the measurements between these types of groups still very subjective?

ziggy53

The most consistent testing of dynamic range to use for "comparative" purposes is at DXOMark:

http://dxomark.com/index.php/en/Camera-Sensor/Sensor-rankings/(type)/usecase_landscape

They don't cover P&S/Digicams much and only a few digital backs but very good coverage of dSLRs and many of the EVIL cameras. They are consistently higher in DR ratings than, for instance, DPReview.

chrisjohnson

One thing I like with the new lower end cameras is that they increasingly integrate an hdr function - basically taking multiple shots with stop differences and then integrating the extremes in-camera to give a wide dynamic range in the finished result. On DSLR you have to do this "manually" using specialized software.

I don't know of any camera that gives the dynamic range perceived by the human eye, which is the result a lot of shooters are aiming for.

Overfocused

Wiki suggests humans can see a 15 stop range at one time. With optimal settings current digital cameras can capture ~10-13 stops in higher end prosumer dslrs. I would be very interested to see photos from a camera with a true 14 or 15 stop range!

Richard

goldenballs wrote: »

I don't know of any camera that gives the dynamic range perceived by the human eye, which is the result a lot of shooters are aiming for.

It's not only a matter of cameras, but of output devices, which have even less dynamic range.

colourbox

Yes, even if a camera could see the full range of the human eye, we would only see it if our monitors and printers could show it! Can any monitors display that full human eye range now? I know paper can't come close...even if such a camera existed it would not save us from having to make edits to get what we want within the dynamic range of paper. If you're printing.

The first benefit of a "full" (compared to human eye) dynamic range camera will be to have "one shot HDR" in other words we'd get everything from dark to light not intended as a final product, but as a great starting point (no clipping) for the final print. As with white balance and exposure, "automatic" camera modes will probably take this one-shot full range frame and make a passable HDR out of it, while pros will take the unprocessed raw and make a more traditional dodged/burned/masked image out of it but without having to take multiple frames to get all the levels.

kwalsh

Helvegr wrote: »

When it comes to typical gains in range, are there some "standard" jumps that are agree upon? For example, is there a jump going from say a point and shoot, to a DSLR, or a jump from a crop sensor to a full sensor? Or maybe to a medium format?

In general there are of course a lot of variables and sensor technology can change things significantly within a give category. For example, at the moment there is a sensor/analog-to-digital converter combination by Sony being used in a few APS-C cameras (K5 and D7000) that has very significantly improved the dynamic range at base ISO to the point of exceeding some full frame solutions. At higher ISOs different issues arise and the advantage is lost.

So, with that in mind, if one could make "all things equal" (which they never are) you would expect dynamic range performance to scale with the linear sensor dimension almost directly (for example a FF sensor would have about one stop more dynamic range than a 2x crop format like micro-four-thirds).

Or are the measurements between these types of groups still very subjective?

In a way, yes. Measuring dynamic range is actually very hard to do "right". Take DxOMark for example. They have a well documented, fairly reasonable methodology for measuring dynamic range and they've used it on a lot of camera models so there is a large database. Seems ideal, right? Well, it is a very good effort but still falls short. For instance, they do not differentiate between pattern noise and random noise in their DR measurement. Pattern noise is far, far more noticeable than random noise. For cameras that exhibit pattern noise their numbers are easily off by a stop. Similarly, some more recent cameras make a trade off in the color filter array - broadening the filters to allow in more light to get better DR/SNR numbers. Problem is that when they have to convert to the end color space they actually amplify color noise in the process on account of the overly broad filters with the result that the apparent DR to the viewer is worse than DxO's simple measurement would imply.


Bottomline, dynamic range tends to improve with larger sensors. In general, with most modern cameras beyond the tiniest compact sensors you'll have to shoot RAW to ever begin using the available dynamic range and then you'll probably have to do a fair bit of PP to get that dynamic range back under control such that you can actually display it in a photogenic manner.


Ken

rpcrowe

ziggy53 wrote: »

Dynamic range can be relatively qualified, meaning that a dynamic range measurement from one review site may not directly compare to the dynamic range measurement from another review site.

However, would not the comparison between two or more cameras on the same site, done using the same methodology; indicate which cameras have greater dynamic range.

Of course, dynamic range doesn't sell cameras. Gazillion mega pixels and quantum ISO capabilities sell the cameras...

On the other hand, reviews like this from DPReview:

http://www.dpreview.com/reviews/canoneos7d/page17.asp

These sites seem to review the dynamic range of cameras (there are many reviews on different cameras) using JPEG as their capture mode. I wonder if these reviews indicate the inherent capabilities of the sensor to capture a dynamic range or the capability of in-camera processing to the JPEG format.

Since I always shoot RAW, is the dynamic range of the JPEG capture any indication of its capability in RAW capture. Also, is the conversion method from RAW (manufacturer's supplied software, Lightroom, Capture One. Photoshop, Photoshop Elements, etc. etc.) a determining factor in the resultant dynamic range capture capability?

ziggy53

DXOMark may, or may not, use the same demosaicing engine from version to version and that can affect results. There is no singular default method for interpreting RAW to full-color representations so things like contrast settings, etc. can, and do, affect DR.

DPReview does sometimes use RAW files and ACR to determine default conversion DR as well as "optimal" settings for extended DR, with the caution that with optimal settings for DR color channels may be saturated/clipped resulting in uncertain color rendition.