High ISO test - 5d2, 7d, 1d2, 40d

I took some time to test all the cameras I have access to and did a little ISO testing.

This is supposed to simulate shooting indoors with bad light, no flash and a reasonably fast lens. Lens in use is a 85mm f/1.8. The different sensor sizes have been flattened out by different distance to the subject. FF is 5 feet, 1d2 is 6.5 feet, 1.6 crops are 8 feet away from the face of the subject. All pictures have been taken three times and the sharpest selected. I have placed a labeled shopping bag in the background to show depth of field.

There are dozens of pictures here and I won't load up the thread with them but you can see everything here:
http://darmok.dyndns.o​rg …oor-port/full/index4.html
http://darmok.dyndns.o​rg …ort/middlecut/i​ndex4.html
http://darmok.dyndns.o​rg …awtests/plain/i​ndex4.html
http://darmok.dyndns.o​rg …/colorcorrect/i​ndex4.html
http://darmok.dyndns.o​rg …/rawtests/cut/i​ndex4.html

Please feel free to quote or link sets you find interesting but don't wipe out the thread please

Raws are available. I have a second test for a second posting that has raws converted just for WB and nothing else (no noise reduction, sharpening etc).

Full picture, all at iso 3200:

1d2:
http://darmok.dyndns.o​rg ….0mm-200-f1.8-3200iso.jpg

Very high ISO results from 5d2 and 7d:

5d2 6400 iso:

This is a iso 1600 test just using the 7d and the 5d2.

Lens in question is 70-200 f/4 non-IS and 1.4x Mk II extender.

Shot jpeg and raw. Raws converted with camera WB only, no noise reduction, no sharpening. Then color corrected by simply bracketing entirely unused regions out.

The 7d pictures have been taken twice: once at full reach, then at full reach/1.6 to make the picture more comparable.

5d2 jpeg:

IMAGE LINK: http://darmok.dyndns.o​rg …mm-1000-f11.0-1600iso.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …f11.0-1600iso_fromraw.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …mm-1250-f11.0-1600iso.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …f11.0-1600iso_fromraw.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …mm-1250-f11.0-1600iso.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …f11.0-1600iso_fromraw.jpg

5d2 jpeg:

IMAGE LINK: http://darmok.dyndns.o​rg …mm-1000-f11.0-1600iso.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …f11.0-1600iso_fromraw.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …mm-1250-f11.0-1600iso.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …f11.0-1600iso_fromraw.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …mm-1250-f11.0-1600iso.jpg

IMAGE LINK: http://darmok.dyndns.o​rg …f11.0-1600iso_fromraw.jpg

What were your conclusions? Mine were 5DII>1DII>7D>40D, with the difference between the last 3 being rather small.

Sp1207 wrote in post #12891382
What were your conclusions? Mine were 5DII>1DII>7D>40D, with the difference between the last 3 being rather small.

I am not sure. Don't just look at the 3200 iso shots, my impression changes every time I look at a different set. Feel free to follow the links at look at iso 800 or whatever.

I have to make a decision whether to send the 7d back to CLP, but the situation is complex. There are some situations even in this limited test where the 7d does very little damage, some other situations where it has an f-stop or maybe a little less of a disadvantage compared to the 5d2, and then there are shots in there where the 7d just screws up the whole picture.

One thing is for sure: I am shocked how much noise there is (see the raw tests) and how much noise deduction is done. Now, Canon seems to have found a very well working NR algorithm so it's not too bad. But alas to my eyes the 7d has the same noise as the 40d, just higher limits (and of course much higher resolution).

Too me the 5d appears to produce better shots at high ISO.

To make viewing a little less painful (for me and for you ) I created subdirectories by ISO speed:

This is for the 1:1 cutout of the low light indoor situation:

http://darmok.dyndns.o​rg …or-port/middlecut/iso400/
http://darmok.dyndns.o​rg …or-port/middlecut/iso640/
http://darmok.dyndns.o​rg …or-port/middlecut/iso800/
http://darmok.dyndns.o​rg …r-port/middlecut/iso1250​/
http://darmok.dyndns.o​rg …r-port/middlecut/iso1600​/
http://darmok.dyndns.o​rg …r-port/middlecut/iso2000​/
http://darmok.dyndns.o​rg …r-port/middlecut/iso2500​/
http://darmok.dyndns.o​rg …r-port/middlecut/iso3200​/
http://darmok.dyndns.o​rg …r-port/middlecut/iso4000​/
http://darmok.dyndns.o​rg …r-port/middlecut/iso5000​/
http://darmok.dyndns.o​rg …r-port/middlecut/iso6400​/
http://darmok.dyndns.o​rg …-port/middlecut/iso1280​0/
http://darmok.dyndns.o​rg …-port/middlecut/iso2560​0/

In my screen the best images (the doll) are 5D>7D>1D>>40D, the noise differences between 5D and 7D seems to be 1 stop, slight difference 7D and 1D.

...not sure why you'd use a 1.4 extender on a 7D to do an ISO test...

CameraMan wrote in post #12891623
...not sure why you'd use a 1.4 extender on a 7D to do an ISO test...

It's on the 5d2, too, so the lens setup is identical.

I know but why? Isn't it degrading the picture a bit? I haven't used one since my 35mm days and I did not like extenders then. They were very degrading to the image.

CameraMan wrote in post #12891679
I know but why? Isn't it degrading the picture a bit? I haven't used one since my 35mm days and I did not like extenders then. They were very degrading to the image.

The 70-200 + 1.4x is what I use right now for reach (birds, ships) so it's what I tested. I don't think it has much in common with the degradation from noise.

If the weather clears up I can repeat the "cereal box" shoot without the extender and with all cameras. But it's tricky since it's outdoors natural light and the light can change faster than I might be able to go through 4 cameras.

Another interesting and fair test would be 5d2+200mm+extender versus 7d1+200mm-no-extender. After all you get the pixel density for the reach and pay in noise for it. So removing the xt might be fair.

Heck, you could do a test with the extender. Take a shot of the cereal box at the full 200mm and put the extender on and pull it back a bit so that it's comparable to the 200mm non extender shot and see what you get. Don't use the High ISO either. Take the best shot you can (ISO 100 if you can) and then with the extender take another at ISO 100 and see what you get using just one camera.

...this may have already been done here somewhere.

https://photography-on-the.net …ghlight=1.4+ext​ender+test

A few replies down (#12 & #13) he does a test and there is quite a difference...