DPReview's Dynamic Range measurements

[Jamie Pillers]

I was just browsing around the DPReview site. I often look at their Dynamic Range measurement when checking out the latest camera offerings. In the past, I've assumed that comparing this measurement from camera to camera would give me a reasonable gauge for choosing between models. But from time to time, these Dynamic Range numbers don't make sense. That happened today.

I was looking at the Panasonic LX100 review. It has a modern 12Mp m4/3 sensor. Its Dynamic Range measurement looked quite good when compared to other small compact cameras. Then I compared the LX100 measurement to the Nikon Df... a modern full frame 16Mp sensor-ed camera. According to DPReview, the little LX100 camera provides FAR better dynamic range than the Nikon Df! Is this possible? Or have I been mis-understanding what their Dynamic Range measurement means? Does anyone here think there's something goofy going on here, or can you help me make sense of this? Thanks.

 

[YYV_146]

At base ISO the Nikon DF's DR is nothing to write home about...but it still should be better than any M43 body.

I refer to DXOmark for dynamic range measurements. DP's tests have never seemed to be very systematic or precise.

 

[Jamie Pillers]

... DP's tests have never seemed to be very systematic or precise.

I agree, I think. At least their results often surprise me. Especially when a particular camera has a very high overall rating and lots of praise about image quality around the web, but then DPReview tests show the sensor's dynamic range to be somewhat ho-hum. Curious.

 

[willie_901]

I don't pay much attention to DPReview's measurements.

The analog dynamic range of the sensor when the shutter is open is directly proportional to the sensor system's analog signal-to-noise ratio. After the shutter closes the rest of the data stream can only reduce DR, but it can not increase it. In newer digital cameras the ADCs themselves are not critical as their read noise is low. This is not the case for older cameras.

I compare DR using Bill Claff's data. Claff's measurements are based on statistical modeling of unrendered, raw-file data. The Df does quite well – just as I would predict for a camera with a recent 24 X 36 mm sensor and data stream. Unfortunately Claff has no data for the LX100. However he does compute the maximum possible DR for different sensor areas.

Please note: Claff does not use the engineering definition of DR. Instead he reduces the engineering DR by a constant (two stops of EV if I remember correctly). Cliff calls this photographic DR. While Claff's DR estimates will be less than DR estimated elsewhere, they are useful for the cameras in his data base.