In the last post, I showed that, at least in the default general-purpose settings, Lumiriver Profile Designer did not do a particularly good job of producing the same color from raw files from three different cameras, even though with the reproduction defaults (except for 2.5 dimension LUTs) it did pretty well with at least one pair of cameras.
Now I’ll look at a canned profile that most of you probably know. It’s Adobe Color Profile (ACP), which was introduced a few months ago as the new default Adobe profile for most cameras. ACP replaces Adobe Standard Profile (ASP) in its default status, but ASP is still around. ACP tends to be a bit more chromatic than ASP. I’ve tested both ASP and ACP across cameras before, and have not found them to do a good job of calibrating out camera differences, but I thought it would be useful to rerun those tests with the same raw files that I’ve used for the last few posts, so that we can make comparisons among ACP and the two Lumariver defaults, reproduction and general-purpose. In addition, I’ll look at camera-to-camera consistency — or, in this case, lack of it — which I hadn’t done in the past.
The cameras were, as before, the Fujifilm GFX 100, the Nikon Z7, and the Sony a7RIV. Lighting was from a pair of Godox AD600Pro strobes. Here are the pairwise chromaticity comparisons.
As with the Lumarive general-purpose default plots, there is seemingly no pattern. Look at the green patch in the graph immediately above. The Z7 is very close to the right value, with about the right chroma, and just a little hue angle shift. The GFX 100 is far less chromatic. How much of that is the result of the intrinsic differences among the color filter arrays (CFAs) of the cameras? Let’s take a look at the reproduction results for the Z7 and the GFX 100:
That tells us that it’s possible to get those cameras to get fairly close when looking at the green patch.
OK, let’s look at the stats for the cameras vs the reference (measured with an i1Pro 1D spectrophotometer):
I’ll put a description of the row titles at the end of this post. Let’s look at the first yellow row, the average Delta-E 2000. The GFX 100, which was the worst in the Lumariver reproduction case, is now in the middle. It was also in the middle with the Lumariver general-purpose case.
Comparing the cameras to each other, and ignoring the reference:
In general, the cameras are not materially closer to each other than they were to the reference. Basically, there is no discernable “look” that transcends the cameras that generated the raw files.
This is mildly depressing. The fact that the reproduction profiles very fairly tightly bunched — the worst aggregate camera-camera spread was better than the best one here — indicated that calibration of the camera was reasonably effective. You would think that applying a look to already tight spreads would widen tham, but it appears to do so.