"Settings for an Accurate Histogram": WB?

[Ray]

I don't believe, that the accuracy of highlight warning has to do with the way the contrast adjustment works.

Why not? In my experience with the 5D, taking a series of shots of the same landscape scene with lots of blue sky, using the same exposure for a correct ETTR according to ACR, a maximum contrast setting will result in the highlight warning covering the whole sky. With a zero contrast setting, about half the area of sky will flash. With a -4 (minimum) contrast setting, just a small portion of the brightest part of the sky will flash.

What's the problem? Are you guys just bored with simple solutions?

[Jonathan Wienke]

The flashing highlight warning starts before JPEG clipping occurs; on my 1Ds and 1D-II it seems to start at about 245 or so. Anything above that will flash even if it isn't clipped in the JPEG. So by adjusting the JPEG contrast setting, you can change the RAW level where the JPEG clipping indication begins. If you have contrast set too high, then you'll find yourself unnecessarily reducing exposure in situations where the RAW isn't clipping at all, just because you see a lot of blinking highlights.

[Ray]

If you have contrast set too high, then you'll find yourself unnecessarily reducing exposure in situations where the RAW isn't clipping at all, just because you see a lot of blinking highlights.

Exactly my experience too, Jonathan.

[Panopeeper]

So by adjusting the JPEG contrast setting, you can change the RAW level where the JPEG clipping indication begins. If you have contrast set too high, then you'll find yourself unnecessarily reducing exposure in situations where the RAW isn't clipping at all, just because you see a lot of blinking highlights

My goal is to make ETTR more reliable by making the histogram and the clipping indication more closely reflecting the raw data. I don't care the least for the clipping in the embedded JPEG; I am comparing the raw clipping in Rawnalyze with the clipping indication in the camera (the in-camera histogram is good only for a coarse judgement). With a neutral WB and neutral other settings (including contrast=0), they match within 1/3 EV. If I reduce the contrast to the minimum, i.e. I apply negative contrast, the camera hides some of the raw clipping.

[Jonathan Wienke]

If I reduce the contrast to the minimum, i.e. I apply negative contrast, the camera hides some of the raw clipping.

I don't doubt that this is true of YOUR camera, but is definitely not true of ALL cameras. You are generalizing the behavior of your particular camera to all cameras, which is demonstrably wrong. It's ridiculous to assume that every camera manufacturer uses the same RAW-to-JPEG processing and highlight indication algorithms in every model of camera they've ever made. It's not even true that every camera model has the same JPEG adjustment parameters; the adjustment range and the effect a particular setting value has vary widely from camera to camera.

As I've said before several times now, the only way to know for sure what the relationship between your camera's JPEGs and RAW is to shoot a series of RAW+JPEG frames containing both neutral and saturated colors, making careful notes of which frames show JPEG clipping in-camera and comparing that data with RAW clipping (or lack thereof) shown by Rawanalyze or an equivalent tool. That test will tell you what settings work best for your camera, but generalizing your results to a different camera model or especially a different camera brand is simply idiotic.

[jljonathan]

Sorry, subject moved to another board.
Jonathan