Anyway as I said as ISO100, sameish to the 7D for random read noise. 5.88 read noise for masked area of 7D2 at ISO100.
However, it does seem to me that there is vastly less banding.
So the usable DR is increased a bit at low ISO compared to the 7D.
Here are my center black frame numbers from the 7D (sometimes the masked areas findings give a little bit less read noise than the main sensor areas, rarely a little more, so the masked areas stuff might be around +/- 1/4 stop or so) compared to these and accounting for the 18 vs 20 it looks like it will have a trace worse (to an entirely insignificant degree) lower ISO DR measured (but again, with basically no banding on the 7D2, effectively it will feel better) and anywhere from 1/4 to 3/4 stops better DR at high ISO measured.
Assuming the ISOs were matched to the same gain amount (that has varied a good +/- 1/3 stop at times, at least).
Seomwhere I had plots that normalized the older cams for the ISO gain differences.
You can compare his numbers above to this old stuff:
Here is my old 7D measurement chart (DR is 100% view, DRn is 8MP normalized as per DxO Print Screen plot results):
and for some other cams:
and all-together now:
and this one does show them adjusted for various gain per ISO number (it shows three 7D plots since I had not ISO per gain value when I did the plot back then and I gave three likely possibilities):
Anyway I hadn\'t gotten around to the high ISO DR yet, but using your numbers it seems as I said 1/4 to 3/4 stops better than 7D depending upon masking vs main sensor difference and how they turn out. So better tech than the 7D and getting close to the 5D3 style tech at high ISO for read noise but still not as good (and not even quite a match for 5D2 read noise there either, although with less ugly banding and blotching than the 5D2 at high ISO shadows, so for scenes for lots of very near black it might look better at high ISO than the 5D2, even the 7D already did, although certainly not for SNR) as that and clearly worse than the 6D tech for high ISO read noise.
Other than fixing up the banding even a little more than the 6D, the sensor seems to perform older than the 6D tech and newer than the 7D tech.
Anyway as I said as ISO100, sameish to the 7D for random read noise. 5.88 read noise for masked area of 7D2 at ISO100.
However, it does seem to me that there is vastly less banding.
So the usable DR is increased a bit at low ISO compared to the 7D.
Here are my center black frame numbers from the 7D (sometimes the masked areas findings give a little bit less read noise than the main sensor areas, rarely a little more, so the masked areas stuff might be around +/- 1/4 stop or so) compared to these and accounting for the 18 vs 20 it looks like it will have a trace worse (to an entirely insignificant degree) lower ISO DR measured (but again, with basically no banding on the 7D2, effectively it will feel better) and anywhere from 1/4 to 3/4 stops better DR at high ISO measured.
Assuming the ISOs were matched to the same gain amount (that has varied a good +/- 1/3 stop at times, at least).
Seomwhere I had plots that normalized the older cams for the ISO gain differences.
Here is my old 7D measurement chart:
and for some other cams:
and all-together now:
and this one does show them adjusted for various gain per ISO number (it shows three 7D plots since I had not ISO per gain value when I did the plot back then and I gave three likely possibilities):
Anyway I hadn\'t gotten around to the high ISO DR yet, but using your numbers it seems as I said 1/4 to 3/4 stops better than 7D depending upon masking vs main sensor difference and how they turn out. So better tech than the 7D and getting close to the 5D3 style tech at high ISO for read noise but still not as good (and not even quite a match for 5D2 read noise there either, although with less ugly banding and blotching than the 5D2 at high ISO shadows, so for scenes for lots of very near black it might look better at high ISO than the 5D2, even the 7D already did, although certainly not for SNR) as that and clearly worse than the 6D tech for high ISO read noise.
Other than fixing up the banding even a little more than the 6D, the sensor seems to perform older than the 6D tech and newer than the 7D tech.
Anyway as I said as ISO100, sameish to the 7D for random read noise. 5.88 read noise for masked area of 7D2 at ISO100.
However, it does seem to me that there is vastly less banding.
So the usable DR is increased a bit at low ISO compared to the 7D.
Here are my center black frame numbers from the 7D (sometimes the masked areas findings give a little bit less read noise than the main sensor areas, rarely a little more, so the masked areas stuff might be around +/- 1/4 stop or so) compared to these and accounting for the 18 vs 20 it looks like it will have a trace worse (to an entirely insignificant degree) lower ISO DR measured (but again, with basically no banding on the 7D2, effectively it will feel better) and anywhere from 1/4 to 3/4 stops better DR at high ISO measured.
Assuming the ISOs were matched to the same gain amount (that has varied a good +/- 1/3 stop at times, at least).
Seomwhere I had plots that normalized the older cams for the ISO gain differences.
Here is my old 7D measurement chart:
and for some other cams:
and all-together now:
and this one does show them adjusted for various gain per ISO number (it shows three 7D plots since I had not ISO per gain value when I did the plot back then and I gave three likely possibilities):
Anyway I hadn\'t gotten around to the high ISO DR yet, but using your numbers it seems as I said 1/4 to 3/4 stops better than 7D depending upon masking vs main sensor difference and how they turn out. So better tech than the 7D and getting close to the 5D3 style tech at high ISO for read but worse than the 6D tech for high ISO read noise.
Other than fixing up the banding even a little more than the 6D, the sensor seems to perform older than the 6D tech and newer than the 7D tech.
Sep 18, 2014 at 10:07 PM
Previous versions of skibum5's message #12586976 « 7D Mark II dynamic range analyzed »
