Here's something to try - I don't have the right version of Photoshop to open RAW 5D2 files yet, so I had to ad-lib a bit. I shot with the cap on at ISO 100, 125, 160, and 200. Increased exposure of all by +2 in DPP - no visible pattern or noise (still pitch black). I then converted the 100, 125, and 160 to JPG and opened in PSE-5. Raised Shadows by "20" on the slider and found that the ISO 100 had considerably more pattern noise than the 125 or 160.
I have to run off to work now, but I'll look into this a bit more tomorrow afternoon.
Well on my 5D2 using the full stops would be a mistake, probably. This corroborates what is said elsewhere in web tests. Even DXOmark's graph of s/n ratio suggests these settings.
p.15 #11 · Are you seeing 5D2 Criss-Cross Pattern?
michael49 wrote:
I just don't get why ISO 100 is so much worse than ISO 50 - and why the interval ISO's are so much better - very strange.
ISO 50 is an overexposed ISO 100 that is pulled back 1 stop digitally. You get less noise, but at the cost of less dynamic headroom. Remember the absolute noise is not affected by exposure.
The same may be the case with 160, 320, 640...
They are the best settings for SNR, but not necessarily for DR.
It could possibly be exactly opposite, that ISO 100 and 125 are in fact pushed ISO 80 (and ISO 50 pulled 80), this has to be verified.
-------------------------------------
Edited, added some more.
p.15 #12 · Are you seeing 5D2 Criss-Cross Pattern?
Interesting to see if the ISO performance of Brainiac's 5D2 is exemplary for all 5D2's out there... And if not (meaning other 5D2's show different performance - let's say ISO100 cleaner than ISO160), what could be the explanation?
p.15 #13 · Are you seeing 5D2 Criss-Cross Pattern?
alundeb wrote:
ISO 50 is an overexposed ISO 100 that is pulled back 1 stop digitally. You get less noise, but at the cost of less dynamic headroom. Remember the absolute noise is not affected by exposure.
The same may be the case with 160, 320, 640...
They are the best settings for SNR, but not necessarily for DR.
It could possibly be exactly opposite, that ISO 100 and 125 are in fact pushed ISO 80 (and ISO 50 pulled 80), this has to be verified.
-------------------------------------
Edited, added some more.
50 is just overexposing 100 by 1 full stop and leaving it as is, very clipped in highlights (but a full stop of extra light) but you can do it the same in iso100 if you just set exposue properly and then add 1 stop more light
in read noise test ISO160 had the lowest, 320 was pretty good too
the 1/3 up stops from were just pushed 1/3 up and seem pointless
the 2/3 up ones get better read noise than the next lowest full stop, i heard they do some sort of odd range value thing so it is not quite like just over-exposing by 1/3 apparently but it is some sort of amped mode or something i need to look into it more but probably no free lunch snce they probably do some sort of highlight clip but maybe there is enough freedom to still cap much of it anyway
once you get to ISO1600 from then on it appears that every 1/3 stop up is just ISO1600 pushed up another 1/3 stops, while they are claimed to be real until ISO6400 the read noise and all doesn't seem to give you anything useful and you just seem to loose another 1/3 highlights so for RAW using above ISO1600 is probably pointless actually
if you can give the shadows lots of light the higher iso do them better, given same amount of exposure time. of course you can only do that if there were not enough highlight etc anyway
i believe the 1dsmkiii has a different amp system adn the intermediate iso work totally differently
possibly, it's late so who knows what i just typed
p.15 #15 · Are you seeing 5D2 Criss-Cross Pattern?
Holy crap, my 5d2's repair is already completed according to their website! They only got it yesterday. If they overnight it like usual and I get it tomorrow I'll be astonished!!
*edit* - It's on the truck this morning for delivery, damn that was quick!
p.15 #16 · Are you seeing 5D2 Criss-Cross Pattern?
Got the camera, i will post up new examples tonight.
Here is their typical cryptic notes on the paper I got back from them:
"We have examined the product according to your request, and, it was found that the adjustment of the C-MOS assembly was incorrect causing noise to occur from time to time. Electrical adjustments were carried out on the C-MOS assembly."
God, I wish I could just talk to these techs sometimes.
