1 Reply Latest reply on Aug 30, 2007 2:47 PM by Bill_Janes

    ACR and dynamic range tests

      I have been looking at dynamic range tests on dpreview.com for different cameras. I'm particularly intrigued by the "RAW headroom" graphs. They have things like "ACR Best: Exp. -0.85 EV, Shadows 0, Bright. 70, Contrast -25, Curve Linear" (Pentax K10D), "ACR Best: Exp. -1.3 EV, Shadows 0, Bright. 70, Contrast -50, Curve Linear" (Canon 30D) "ACR Best: Exp. -1.0 EV, Shadows 0, Bright. 70, Contrast -50, Curve Linear" (Canon 400D). The "RAW ACR Best" curves for all these cameras are very different, and I don't think I understand their real meaning, so here are a few questions:

      - What is a "linear curve" in ACR? Does it mean no curve at all is applied to the RAW data, and so what we see is the actual sensor response? Or, does ACR apply a standard curve to RAW data from all cameras, even when a "linear curve" is selected? Or even yet, are there different default curves (even with "linear curve" selected) for different cameras in ACR, and what implications would this have when comparing graphs from different cameras?

      - Since these cameras have most likely been processed with different versions of ACR (since they were released at different times), could ACR's algorithms and defaults have been changed in any way that might prevent these tests from being readily comparable? For example, the noise reduction and highlight recovery algorithms might have been improved in the latest versions of ACR.

      Thanks,

      Marcos
        • 1. Re: ACR and dynamic range tests
          Bill_Janes Level 2
          >- What is a "linear curve" in ACR? Does it mean no curve at all is applied to the RAW data, and so what we see is the actual sensor response? Or, does ACR apply a standard curve to RAW data from all cameras, even when a "linear curve" is selected? Or even yet, are there different default curves (even with "linear curve" selected) for different cameras in ACR, and what implications would this have when comparing graphs from different cameras?

          ACR will not give you the actual sensor response with no white balance or gamma encoding; for this purpose you need a specialized program such as DCRaw. If you convert into sRGB or aRGB, the image is encoded with a gamma of 2.2, whereas a gamma of 1.8 is used for ProPhotoRGB.

          If you want to linearize the tonal response in ACR you can set the sliders on the basic tab to zero (Exposure, Brightness, Contrast to and Shadows). Curves was added in ACR ver 3.X and you should also set the curve to linear. The Medium and Strong Contrast in the curves tab adds to the Contrast slider in the basic tab, and I think that the two were not merged for reasons of backward compatibility. In all cases, the resulting image is gamma encoded. To learn more about these controls, I would recommend Bruce Fraser's Real World Camera Raw for PSCS2 book. The Exposure and Black sliders set the black and white endpoints respectively. The contrast control applies an S curve around the midpoint, which is set by the Brightness control (see p 81). Jeff Schewe is revising the book for PSCS3 due out shortly.

          Dynamic range is limited by the noise floor and improvements in noise reduction in the current version of ACR could affect the results. Phil's dynamic range measurement uses highlight recovery with the Exposure slider (negative values). ACR 4.1 also has the Recovery slider, but he apparently does not use this control.