Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

My posts · My subscriptions
  

  Previous versions of skibum5's message #10767114 « Got the 1DX: Shadow recovery/DR sample »

  

skibum5
Offline
Upload & Sell: Off
Re: Got the 1DX: Shadow recovery/DR sample


OK, I have examined a 1DX black frame. It's not really very encouraging news, at least not entirely, although it definitely is better than the 5D3, mostly because it has less nasty banding and it is nice to see the control that a bit better again.

Overall it's probably, usably, near a tie for best from Canon, quite close to 1Ds3 probably. So it's nothing so much bad as just living in half a decade ago glory levels and not progressing any.

The DR measures about par for the course for Canon since 5-6 years ago.

After seeing their astonishing break-through at light collection and record high iso performance I had gotten excited they would maybe at least get the 1DX to D4-like dynamic range at low ISO, even if hardly D800 level, but nope, probably 1.2-1.4 stops worse than D4.

I'd need to go check the other files, but off-hand, I'd guess it's one the better Canon efforts in terms of banding, at least. It has, for sure, less low ISO shadow banding than 5D2/5D3/7D/50D. I vaguely recall that the 1D4 was kinda bad for banding in low ISO shadows, so the 1DX is probably better than that as well. It's probably somewhere in the mix with the 1Ds3/1D3/40D as the best in terms of low iso banding from Canon. The 1DX actually has more horizontal banding than the 5D3, which has none of that at all, but it's quite low levels of horizontal all the same, and while it still has some noticeable vertical banding, it has quite a lot less of that than the 5D3 or 5D2 or 7D, etc., considerably less. So the banding on the 1DX is probably getting to be more like 1Ds3, only this time in the other direction (you'd think if they can stop it in one direction one cam and in another direction on another cam then they could stop it in both directions though but the 1Ds3 seems to be the onyl cam they ever solved vertical banding with, they are close for horiz with the 1DX and the 5D3 is the only one they have ever totally solved horizontal banding at low ISO with).

Believe it or not, the 1Ds3 remains the best performing (granted, by a perhaps meaningless degree compared to the 1DX) ISO100 camera canon has made overall. The 1Ds3 has 3 more MP (ok not a huge deal, but technically it's better). The 1Ds3 measures about 0.25 or so stops better, not a big deal by any means and maybe just too small to actually notice the difference being realistic, but technically better. Off-hand, form what I vaguely recall the 1Ds3 has banding performance of roughly the same order (I need to go check though, but call it a tie for now, although it could swing either way). I'd guess the 1Ds3 is less color-blind, so probably a tiny advantage there. The 1Ds3 does somewhat worse SNR at ISO100, maybe a stop?, that sounds like the biggest difference so far, but at mid-gray SNR isn't really THAT huge of a deal at ISO100 since all the top cameras are relatively clean there. Whatever the case the 1DX clearly does not move things forward from the 1Ds3 days and appears to be about on par or maybe just a touch off.

ISO 100 back frame read noise NOT normalized this is at 100% view level (lower is better):

5D3 6.4-7.0 ADU [6.4(beta copies); 6.6 (many release copies); 7.0 (my copy )] = the worst per photosite read noise, by a tiny margin, since the 30D-era.

5D2 6.1 ADU

1DX 5.8 ADU (3.86 normalized to 8MP requires 0.666x conversion factor )

1Ds3 5.7 ADU

7D 5.4 ADU

1D4 5.2 ADU

It seems Canon reach the lowest level of per photosite read noise with their 1D4-era cameras.

OTOH they reached the lowest banding levels with the 1Ds3/1D3/40D and now also 1DX.

Per total image the highest measured engineering DR they have produced is in the 1Ds3, 1D4, 5D2 and perhaps 1DX. But the 1Ds3 and 1DX appear to have much less banding than the others so they should be more usable than say the 5D2, 5D3, 7D, 1D4 and such even if the others appear to measure as well.

So 1Ds3 and 1DX are arguably Canon's best for DR, but the 1DX is nearing 1/2 stop worse than numerous Nikon bodies and solidly over a stop worse than Nikon's new D4 and of course miles behind anything having an Exmor sensor.



ISO 100 engineering dynamic range (normalized to 8MP as per DxO):
All of the Exmor stunners:
DxO's D800: 14.3
DxO's D7000: 13.9
DxO's D3x: 13.65
DxO's D5100: 13.6

And the new Nikon stunner, considering it does not make use of any fancy Exmor patents:
DxO's D4: 13.1 (Note Nikon was able to get this WITHOUT using Exmor patents so anyone else, theoretically should be able to do so without patent violation as well.)

And then it's the best of the distant rest:
DxO's D3: 12.2
DxO's D300s: 12.2
DxO's D700: 12.15
DxO's D300: 12.0
DxO's CANON 1Ds3: 12.0 (still Canon's king 5 years later, almost 1/2 stop DR and with much less banding than my 5D3)
DxO's 1D4: 11.95
DxO's D3s: 11.94
My 5D2: 11.93 (but it has a lot of banding so the usable isn't at all as high as for the others in the near 12.0 range)
DxO's 5D2: 11.86 (" ")
My 7D: 11.8 (also suffers from a lot of banding so the usable is not as good as some of the others in this range)
Two beta 5D3s: 11.76 (zero horizontal banding, but still a lot of vertical so not as high usable DR as some of the others in the range)

1DX: guessing 11.75 (not quite sure what the well depth is though, but that number should alter results too much, probably not by more than +/- 0.1 or +/-0.2 at the worst) (the banding is definitely less than for 5D3 and 7D and 50D and 5D2 and all the nasty ones, it's probably closer to the 1Ds3 banding, I have to go find my 1Ds3 and 1D4 samples, I don't remember exactly off-hand.)

DxO's 5D3: 11.74 (" ")
DxO's 7D: 11.7 (" ")
Another FM'ers 5D3: 11.7 (" ")
DxO's 1D3: 11.7
My 5D3: 11.6 (" ")

And then the rest of the back pack (although a few with very low banding may have slightly more usable DR than a few of the cameras in the set above):
DxO's 60D: 11.5
DxO's D200: 11.5
DxO's Rebel 600D: 11.45
My 40D: 11.4 (very little banding, so despite 1/2 worse measured than 5D2 in the real world it might do you at least as well)
My 50D: 11.4 (a ton of banding though)
DxO's 50D: 11.4 (" ")
DxO's 40D: 11.2


And below here are the old school bottom of the heap:
DxO's 20D: 11.0
DxO's D40: 11.0
DxO's D2Xs: 10.9
DxO's D2X: 10.9
DxO's 30D: 10.8
My second 20D a very late production run copy (I always felt it wasn't as good as the first copy, looking at DxO's scores, maybe so, it also had slightly different construction than my earlier copy 20D): 10.6
DxO's D70: 10.3




Jun 30, 2012 at 11:45 PM
skibum5
Offline
Upload & Sell: Off
Re: Got the 1DX: Shadow recovery/DR sample


OK, I have examined a 1DX black frame. It's not really very encouraging news, although it definitely is better than the 5D3, mostly because it has less nasty banding. And overall it's probably, usably, near a tie for best from Canon, quite close to 1Ds3 probably. So it's nothing so much bad as just living in half a decade ago glory levels and not progressing any.

The DR measures about par for the course for Canon since 5-6 years ago.

After seeing their astonishing break-through at light collection and record high iso performance I had gotten excited they would maybe at least get the 1DX to D4-like dynamic range at low ISO, even if hardly D800 level, but nope, probably 1.2-1.4 stops worse than D4.

I'd need to go check the other files, but off-hand, I'd guess it's one the better Canon efforts in terms of banding, at least. It has, for sure, less low ISO shadow banding than 5D2/5D3/7D/50D. I vaguely recall that the 1D4 was kinda bad for banding in low ISO shadows, so the 1DX is probably better than that as well. It's probably somewhere in the mix with the 1Ds3/1D3/40D as the best in terms of low iso banding from Canon. The 1DX actually has more horizontal banding than the 5D3, which has none of that at all, but it's quite low levels of horizontal all the same, and while it still has some noticeable vertical banding, it has quite a lot less of that than the 5D3 or 5D2 or 7D, etc., considerably less. So the banding on the 1DX is probably getting to be more like 1Ds3, only this time in the other direction (you'd think if they can stop it in one direction one cam and in another direction on another cam then they could stop it in both directions though but the 1Ds3 seems to be the onyl cam they ever solved vertical banding with, they are close for horiz with the 1DX and the 5D3 is the only one they have ever totally solved horizontal banding at low ISO with).

Believe it or not, the 1Ds3 remains the best performing (granted, by a perhaps meaningless degree compared to the 1DX) ISO100 camera canon has made overall. The 1Ds3 has 3 more MP (ok not a huge deal, but technically it's better). The 1Ds3 measures about 0.25 or so stops better, not a big deal by any means and maybe just too small to actually notice the difference being realistic, but technically better. Off-hand, form what I vaguely recall the 1Ds3 has banding performance of roughly the same order (I need to go check though, but call it a tie for now, although it could swing either way). I'd guess the 1Ds3 is less color-blind, so probably a tiny advantage there. The 1Ds3 does somewhat worse SNR at ISO100, maybe a stop?, that sounds like the biggest difference so far, but at mid-gray SNR isn't really THAT huge of a deal at ISO100 since all the top cameras are relatively clean there. Whatever the case the 1DX clearly does not move things forward from the 1Ds3 days and appears to be about on par or maybe just a touch off.

ISO 100 back frame read noise NOT normalized this is at 100% view level (lower is better):

5D3 6.4-7.0 ADU [6.4(beta copies); 6.6 (many release copies); 7.0 (my copy )] = the worst per photosite read noise, by a tiny margin, since the 30D-era.

5D2 6.1 ADU

1DX 5.8 ADU (3.86 normalized to 8MP requires 0.666x conversion factor )

1Ds3 5.7 ADU

7D 5.4 ADU

1D4 5.2 ADU

It seems Canon reach the lowest level of per photosite read noise with their 1D4-era cameras.

OTOH they reached the lowest banding levels with the 1Ds3/1D3/40D and now also 1DX.

Per total image the highest measured engineering DR they have produced is in the 1Ds3, 1D4, 5D2 and perhaps 1DX. But the 1Ds3 and 1DX appear to have much less banding than the others so they should be more usable than say the 5D2, 5D3, 7D, 1D4 and such even if the others appear to measure as well.

So 1Ds3 and 1DX are arguably Canon's best for DR, but the 1DX is nearing 1/2 stop worse than numerous Nikon bodies and solidly over a stop worse than Nikon's new D4 and of course miles behind anything having an Exmor sensor.



ISO 100 engineering dynamic range (normalized to 8MP as per DxO):
All of the Exmor stunners:
DxO's D800: 14.3
DxO's D7000: 13.9
DxO's D3x: 13.65
DxO's D5100: 13.6

And the new Nikon stunner, considering it does not make use of any fancy Exmor patents:
DxO's D4: 13.1 (Note Nikon was able to get this WITHOUT using Exmor patents so anyone else, theoretically should be able to do so without patent violation as well.)

And then it's the best of the distant rest:
DxO's D3: 12.2
DxO's D300s: 12.2
DxO's D700: 12.15
DxO's D300: 12.0
DxO's CANON 1Ds3: 12.0 (still Canon's king 5 years later, almost 1/2 stop DR and with much less banding than my 5D3)
DxO's 1D4: 11.95
DxO's D3s: 11.94
My 5D2: 11.93 (but it has a lot of banding so the usable isn't at all as high as for the others in the near 12.0 range)
DxO's 5D2: 11.86 (" ")
My 7D: 11.8 (also suffers from a lot of banding so the usable is not as good as some of the others in this range)
Two beta 5D3s: 11.76 (zero horizontal banding, but still a lot of vertical so not as high usable DR as some of the others in the range)

1DX: guessing 11.75 (not quite sure what the well depth is though, but that number should alter results too much, probably not by more than +/- 0.1 or +/-0.2 at the worst) (the banding is definitely less than for 5D3 and 7D and 50D and 5D2 and all the nasty ones, it's probably closer to the 1Ds3 banding, I have to go find my 1Ds3 and 1D4 samples, I don't remember exactly off-hand.)

DxO's 5D3: 11.74 (" ")
DxO's 7D: 11.7 (" ")
Another FM'ers 5D3: 11.7 (" ")
DxO's 1D3: 11.7
My 5D3: 11.6 (" ")

And then the rest of the back pack (although a few with very low banding may have slightly more usable DR than a few of the cameras in the set above):
DxO's 60D: 11.5
DxO's D200: 11.5
DxO's Rebel 600D: 11.45
My 40D: 11.4 (very little banding, so despite 1/2 worse measured than 5D2 in the real world it might do you at least as well)
My 50D: 11.4 (a ton of banding though)
DxO's 50D: 11.4 (" ")
DxO's 40D: 11.2


And below here are the old school bottom of the heap:
DxO's 20D: 11.0
DxO's D40: 11.0
DxO's D2Xs: 10.9
DxO's D2X: 10.9
DxO's 30D: 10.8
My second 20D a very late production run copy (I always felt it wasn't as good as the first copy, looking at DxO's scores, maybe so, it also had slightly different construction than my earlier copy 20D): 10.6
DxO's D70: 10.3




Jun 30, 2012 at 11:28 PM



  Previous versions of skibum5's message #10767114 « Got the 1DX: Shadow recovery/DR sample »