What do in-camera histograms reflect

[larsrc]

When I take a picture in RAW mode, I can get a preview and a histogram on the camera.  The preview obviously has some exposure curve applied, but is that the same that is shown in the histogram?  Is there a way to make the histogram reflect more accurately what data is in the RAW file, esp. at what point highlights are blown?  If not, what are peoples experiences/guidelines with how much of what appears overexposed in the camera histogram can be rescued from the RAW file?

For the record, I use a 350D.

Thanks,
-Lars

[John Sheehy]

Is there a way to make the histogram reflect more accurately what data is in the RAW file, esp. at what point highlights are blown?[a href=\"index.php?act=findpost&pid=83966\"][{POST_SNAPBACK}][/a]

Not precisely.  You can get closest by setting contrast to minimum, and saturation to minimum.  Even closer by custom-white-balancing to a magenta sheet of paper, but your review images will all have a cyan-ish green cast.

The RAW data in most Canons clip about 1/3 stop above where the review JPEG flashes (upon which the histogram is based).  This is only true of white highlights, though.  Red and blue do not affect the histogram/review luminance as much as green, and may be under-stated in the review.  On the other hand, saturated red flowers may clip in the JPEG, but not in the luminance histogram, but still be far from clipping in the red RAW channel.

I guess Canon thinks we're too stupid to use RAW RGB histograms.