Well, just completed round one of DR comparison. I converted everything
exactly the same way (flat, zero corrections of any kind) in Picture Window
Pro. Yes, it's pretty brutal in the uncorrected format, but that's what I
wanted to see here.
Out of entertainment we shot the 7D at 100 and 3200. I've got an observation
about this. I'm not aware of the 7D pulling exposure for ISO 100, but there
was an apparant DR hit on it. We may retest at ISO 200 to study this
further. Regardless, Canon definitely fixed the color and dynamic range
issues with the 7D. I'm quite impressed with the ISO 3200 tests. I can tell
there is a lot of noise-reduction going on with the 7D's 3200 setting
because the bottom end was lifted and the top end supressed. Inotherwords,
the noise is being masked in-camera (even on the CRF files) and then
noise-reduction is actually extending the useful dynamic range of the
image!!!
Of major interest to me was how the venerable E-1 stood up to these two
cameras. Just as I expected, the top end takes a hit. But so did the 7D at
ISO 100. The 7D at 3200 almost perfectly matched the 5D's dynamic range at
100, but at ISO 100 it was almost a dead knockoff to the E-1. The grayscale
portion of the IT8 target looked like this: (no highlight recovery performed
at all on these 3-stop overexposed images--that's coming next):
Last easily identifiable Gray patch on the scale:
E1 - 12
5D - 9
7D (100) - 11
7D (3200) - 10
But where things get interesting is in the colors. I'll list out the number
of visible boxes stacked in each color:
E-1 - C 4, M 3, Y 2, K 6, R 5, G 5, B 5
5D - C 5, M 4, Y 5, K 7, R 6, G 6, B 7
7D (100) - C 4, M 3, Y 3, K 6, R 5, G 4, B 6
7D (3200) - C 5, M 4, Y 4, K 7, R 6, G 6, B 7
The bottom end is a little inconclusive, as 2-stops underexposed was easily
handled by all three cameras, but there appears to be some interesting
trends. The Canons have a tendency to increase saturation a bit in the low
levels--especially the blues as they turn more purple. But the biggest
thing I noticed was how the E-1 tended to go darker in the mid-lows but then
leveled out more near the end. A classic "toe" response curve. All of them
maintained visible steps all the way to the final square.
So, that meant I needed to look at something else. In the lower-right corner
of the IT8 target are six squares of similar colors which really really
tough to get a good response on. None of the cameras correctly captures J20
or J21. J30 was too dark on the E-1 and the 7D at 3200. But all did a pretty
comparable job.
So, basically here is my initial conclusion in this first-draft rough
and dirty comparison: The 5D and E-1 are in different leagues when it comes
to the top end of the dynamic range. The 5D is around a full stop better off
there, and most importantly, right near the top (when the sensor starts to
clip), the 5D does maintain a lot more color accuracy. Yellows are
particularily a disaster zone with the E-1. Even the gray scale starts to
change colors near clipping. The middle-range of tones is where the E-1
really shines and you can see how the bottom end and a curved response.
The real suprise to me is the 7D's performance. At ISO 100, the dynamic
range is hardly better than the E-1's. In fact, it experiences similar color
shifts as the E-1. But the ISO 3200 setting of the 7D revealed curves very
much like the 5D's at 100.
The premise behind this specific test and raw conversion was to first of all
generate some HDR profiles for PWP. As such, the conversions were kept
perfectly straight with no shenanigans performed. Also, they were done with
the converter engine in PWP, which is DCRAW based, I believe. I'll do some
additional conversions using ACR and Olympus Studio, which I know is better
behaved on the Olympus files. I will also do some highlight and shadow
recovery on these files to see what is really lurking in the extremes.
I'm quite impressed with the 7D (as I've always been) and find the high-ISO
performance to be quite admirable. But it is nice to know that the E-1 isn't
quite the dog that the world has declared it to be. Just do not overexpose
it.
AG
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|