70D vs 7D - preliminary review for bird photography

neuroanatomist

I post too Much on Here!!
CR Pro
Jul 21, 2010
26,101
4,679
Marsu42 said:
neuroanatomist said:
I'm not sure if this represents a difference specific to ML, a difference in testing methodologies, or a difference in video vs. still processing (if the ML information is based on video, but maybe it's not).

It's certainly based on stills raw shooting, ML doesn't only do video :) ... to be sure what's it all about I crossposted the issue on their forum, I hope their devs will have a look at your DxO data: http://www.magiclantern.fm/forum/index.php?topic=9867.msg94974#msg94974

Thanks! Interested to see their response...
 

ejenner

EOS RP
Nov 28, 2011
205
7
Well it's easy enough to test. Maybe not to tell exactly what the base ISO is (it almost certainly isn't exactly multiple of 100 or even 160).

Done the test, know what I want. Clearly the OP knows what he wants from experience of what he shoots.

Just take some controlled shots of a scene with a very large dynamic range and see what gets clipped and when and how noisy the shadows are.

But then I've read of people swearing blind that ISO320 was cleaner than ISO100 or 200 on a 5DII. Not sure what they were shooting (possibly a lens cap).

I don't understand this discussion from people with the cameras in question in their possession.
 

Marsu42

Canon Pride.
Feb 7, 2012
6,314
0
Berlin
der-tierfotograf.de
neuroanatomist said:
Thanks! Interested to see their response...

Well, the current discussion ended in the thesis that there are "fundamental mistakes in the data presented in these graphs, this makes all graphs invalid" and there isn't any dr gained by pulled isos ever, period, make my day, punk :)

You can read up on the discussion yourself, the guy in question is an adamant believer in the "100s multiple" theory, and the discussion over there unfortunately often is way less nice then over here I'm afraid to say. I hope the real ml devs will say something on this, too, as their wiki states otherwise.

However it's interesting that there is a tiny bit less read noise in pulled isos up to iso800, see here: http://home.comcast.net/~nikond70/Charts/RN_e.htm
 

neuroanatomist

I post too Much on Here!!
CR Pro
Jul 21, 2010
26,101
4,679
Marsu42 said:
neuroanatomist said:
Thanks! Interested to see their response...

Well, the current discussion ended in the thesis that there are "fundamental mistakes in the data presented in these graphs, this makes all graphs invalid" and there isn't any dr gained by pulled isos ever, period, make my day, punk :)

You can read up on the discussion yourself, the guy in question is an adamant believer in the "100s multiple" theory

Even if 'there isn't any DR gained by pulled ISOs', the guy's argument is clearly biased by his being 'an adamant believer in the "100s multiple" theory'. If you leave that unproven assumption out, you're left with the fact that an ISO other than ISO 100 (or multiples thereof) might have more DR, because ISO 100 itself has digital gain applied.