There are some reasons to be concerned with the dynamic range. Here are a couple of points :
First, a lot of people noted the following highlight on DPR hands-on preview : "When shadow areas are lifted by a couple of stops, there's no obvious banding, but noise becomes prominent". However, the previous sentence is much more revealing : "According to Canon representatives, the 6D Mark II should outperform the original 6D (which it very evidently does) but may not offer the same kind of dynamic range and absolute resolution of the EOS 5D Mark IV.".
Then, there are the features : when the no-4K information leaked, many readers here and elsewhere complained that even to generate stable 1080P you might want to shoot 4K and stabilize in post. Then we learned that the camera would have electronic stabilization. So while it might not have 4K, it should be able to record smooth 1080P without having to sacrifice more resolution. I think most people were surprised to see that the 6D2 can generate in-camera HDR jpeg (manual, p253). I am assuming that this is perhaps to patch the actual lack of DR of the camera.
Finally there are the measurements made from the leaked CR2 : all we need to compute is the mean and the standard deviation of the values in the masked portions of the sensor (top and left bands, and avoiding the few rows storing the extra metadata there). That value is not dependent of the scene shot, only the ISO setting, and tells you the dynamic range as follow :
DR = log2((maxQ-mean(mask))/std(mask))
and where maxQ is the maximum value of the ADC (16383 for 14bits). With this expression, the two leaked 100ISO images achieve 11.04Ev DR (512 mean floor and 7.54 std.dev/noise); the 1600ISO image achieves 10.08Ev (2048 mean floor and 13.25 std.dev/noise) and the 40000ISO image achieves 6.95Ev (2048 mean floor and 115.6 std.dev/noise).
For comparison, a 5D4 achieves 12.55Ev at 100ISO (512 mean floor and 2.65 std.dev/noise), 10.62Ev at 1600ISO (2048 mean floor, 9.09 std.dev/noise), 8.16Ev at 12800ISO (2048 mean value, 49.98 std.dev/noise) and 7.01Ev at 32000ISO (2048 mean floor, 111.10 std.dev/noise). An 80D achieves 12.5Ev at 100ISO (512 mean floor, 2.73 std.dev/noise) and 6.92Ev at 12800ISO (2048 mean floor, 118.38 std.dev/iso).
If we get more 6D2 CR2s we will be able to get a much better idea of that DR curve. But for now, it looks as if the low ISO DR is even lower than the 80D (-1.5Ev), while the high ISO DR might be slightly over the 5D4 after catching up probably around 3200ISO...
First, a lot of people noted the following highlight on DPR hands-on preview : "When shadow areas are lifted by a couple of stops, there's no obvious banding, but noise becomes prominent". However, the previous sentence is much more revealing : "According to Canon representatives, the 6D Mark II should outperform the original 6D (which it very evidently does) but may not offer the same kind of dynamic range and absolute resolution of the EOS 5D Mark IV.".
Then, there are the features : when the no-4K information leaked, many readers here and elsewhere complained that even to generate stable 1080P you might want to shoot 4K and stabilize in post. Then we learned that the camera would have electronic stabilization. So while it might not have 4K, it should be able to record smooth 1080P without having to sacrifice more resolution. I think most people were surprised to see that the 6D2 can generate in-camera HDR jpeg (manual, p253). I am assuming that this is perhaps to patch the actual lack of DR of the camera.
Finally there are the measurements made from the leaked CR2 : all we need to compute is the mean and the standard deviation of the values in the masked portions of the sensor (top and left bands, and avoiding the few rows storing the extra metadata there). That value is not dependent of the scene shot, only the ISO setting, and tells you the dynamic range as follow :
DR = log2((maxQ-mean(mask))/std(mask))
and where maxQ is the maximum value of the ADC (16383 for 14bits). With this expression, the two leaked 100ISO images achieve 11.04Ev DR (512 mean floor and 7.54 std.dev/noise); the 1600ISO image achieves 10.08Ev (2048 mean floor and 13.25 std.dev/noise) and the 40000ISO image achieves 6.95Ev (2048 mean floor and 115.6 std.dev/noise).
For comparison, a 5D4 achieves 12.55Ev at 100ISO (512 mean floor and 2.65 std.dev/noise), 10.62Ev at 1600ISO (2048 mean floor, 9.09 std.dev/noise), 8.16Ev at 12800ISO (2048 mean value, 49.98 std.dev/noise) and 7.01Ev at 32000ISO (2048 mean floor, 111.10 std.dev/noise). An 80D achieves 12.5Ev at 100ISO (512 mean floor, 2.73 std.dev/noise) and 6.92Ev at 12800ISO (2048 mean floor, 118.38 std.dev/iso).
If we get more 6D2 CR2s we will be able to get a much better idea of that DR curve. But for now, it looks as if the low ISO DR is even lower than the 80D (-1.5Ev), while the high ISO DR might be slightly over the 5D4 after catching up probably around 3200ISO...
Upvote
0