ISO and noise on 30D, 5D

[akhater]

One of the first things I learned in photography is that a lower ISO means less noise. And after 4 years I was proven wrong.
If you own a 30D/5D I'm think this might interest you

http://www.adidap.com/2006/09/09/canon-eos-30d-iso-vs-noise/

[KevC]

How interesting.

[dsze]

...yes, interesting. I'm not sure what this really means though. Noise is very relative. I just wonder how many confounding variables were controlled with this study. Surely, someone with more knowledge will chime in.

[akhater]

dsze it is "blackframe noise" so noise tested on a picture taken with caps on with short exposures at different iso

[dsze]

dsze it is "blackframe noise" so noise tested on a picture taken with caps on with short exposures at different iso

Yes, I realize this. ...but no other variables are given. Very few details of the study are given and I don't see any links to the full research. Additionally, I'm not so sure how "blackframe noise" really translates into 'real-world' noise ....unless you're shooting with your lens cap on all the time.

[farrukh]

It's proof that 30D is surely 20dN. I think 30d firmware can be hacked to use with the 20d.

[AeroSmith]

Very interesting. You'd think Canon might try to make 100, 200, 400, 800, etc. the lowest noise points on all of its cameras.

[papucla10]

wow

[Servo'd]

wow
nice

[akhater]

Yes, I realize this. ...but no other variables are given. Very few details of the study are given and I don't see any links to the full research. Additionally, I'm not so sure how "blackframe noise" really translates into 'real-world' noise ....unless you're shooting with your lens cap on all the time.
dsze, the link to where I picked up my news from is at the bottom of the post

[Nikolas]

means nothing
How does this translate in the real world?

[sandro9mm]

Nikolas
in real world it means that iso 160 is better than iso 100 on canon 30D :)

[akhater]

well as you can read in the conclusion

JPG shooters or RAW shooters that don't use EC but trust the camera's light meter are better off using the 160, 320, ... group than the 100,200 ...

if You shoot RAW and are used to expose to the right than you can get the less noise of the 160, 320 group on the 100, 200 group (that suffers less postorization) by overexposign by 1/3

[sandro9mm]

that's another point agains getting 30D and instead goin with 20D used

[gjl711]

Interesting article, though I would like to have seen a bit more on the test setup. Was only one pic taken? 10 pics? If more than one. Was the averaged, mean or median used? Following one of the links there is a great discussion about the primary IOSes and all in between being derived. For those that do not want to read the entire thread on dpreview, this link (http://forums.dpreview.com/forums/read.asp?forum=1019&message=19928234)does a great job explaining it.

[ijohnson]

Amazing to see that 400iso on the 5D is almost equal to 100. Scary revelation.

[sandro9mm]

ijohnson
I don;t think that's gonna be same in real world situation...

[ijohnson]

In the real world I have a heck of time telling the difference between 100 and 800 anyway. In fact, Without blowing up the images to 100% its difficult to see noise at 1600 and 3200.

Based on the article, why would you say that the real world would be any different than what the tests show?

[Tyreman]

I can remember years ago with film if you did a shot and wanted to blow it up you used 100 iso film.
Now I have found that over the last few years even 400 iso can be blown up and I've got great results.

[sandro9mm]

ijohnson'
I read that whole article, all posts by john... I understand what is he saying, and probably he is right.

but there is huge difference between shots with cap on and with real world photography were details are sometimes more important than noise. I'm getting 30D in couple of days myself, gonna test that theory myself.

[Jon]

I can't find a detailed description of his methodology there, and my 5D tests suggest that if you're using ANR it kicks in well before Canon says it does, as early as 1/125 sec. which will have an impact on your noise levels. Grain of salt time, folks.

[dsze]

dsze, the link to where I picked up my news from is at the bottom of the post


There is no information in that link you provide (0r very little at least) about this experimenter's methodology, variable control, details of the experiment, etc. MANY parts of this research are missing and such 'research' wouldn't even get read by any reputable peer-reviewed journal. As a researcher myself, numbers mean nothing to me without detailed descriptions of the experiment and conditions. There just isn't enough detail here for me to even give this a second thought while in the field shooting.

Jon, is absolutely right... take it with a grain of salt.

[akhater]

Sandro I have the 30D since 6 months, it is a GREAT camera... the spot meter is sweet and the big LCD just wonderful... I'm sure you'll enjoy it, just try to stay away from the 125, 250, ISO groups

[Tom W]

Interesting article, though I would like to have seen a bit more on the test setup. Was only one pic taken? 10 pics? If more than one. Was the averaged, mean or median used? Following one of the links there is a great discussion about the primary IOSes and all in between being derived. For those that do not want to read the entire thread on dpreview, this link (http://forums.dpreview.com/forums/read.asp?forum=1019&message=19928234)does a great job explaining it.

Good link, gjl711. That's a good explanation of what's being tested.

Here's another, simpler ISO test that appears elsewhere in that thread (done with a 5D) that shows the visible noise at various ISO settings. It appears that the overall noise level (which presumably includes more than just the measured "dark frame noise") varies directly with the ISO setting:

http://forums.dpreview.com/forums/read.asp?forum=1019&message=19957275

I think that Eric's test illustrates what you'll see when all forms of noise are included. Though it's not a numerical measurement, it is a visible measurement. I believe that he has stated that he intends to repeat that type of test with an image that includes deeper shadows to see how the results compare to the measured "dark frame" noise mentioned elsewhere in the thread.

[AeroSmith]

ijohnson'
I read that whole article, all posts by john... I understand what is he saying, and probably he is right.

but there is huge difference between shots with cap on and with real world photography were details are sometimes more important than noise. I'm getting 30D in couple of days myself, gonna test that theory myself.

Well, I hate to say it but my real world results with the 5D indicate that you can really crank up the ISO with virtually no increase in noise. I wish the 1D MkIIn was as good. The fact is, in the real world, no camera can rival the 5D's high ISO performance. And I think this article just reconfirms what those of us who own 5Ds already knew.

EDIT: And in fact I might go so far as to say that the 5D produces sharper colors at 400 than it does at 100. Though I only have my anecdotal real world experience on which to base this theory. So take it for what it's worth (2 cents :D ).

[sandro9mm]

AeroSmith
5D is a great camera I had a fortune of holding it couple of times :)

what about 30D?

[Keiffer]

You know this had me going for a bit, to the point where I started shooting in the 160/320/640 range yesterday on my 30D. I didn't notice any major difference. So last night I set the cam up on a tripod and shot all the different ISOs and can honestly say on my 30D Real world does not show his findings. At 100% crops I started to notice noise kicking in at 400 but no big deal if your not sitting there pixel peeping. Then I took my Iso 1600 shot and ran it through Noise ninja and wahlaa, as good as is 200. Moral of this story, Shoot in whatever ISO you need to achieve the shutter speed you need and if forced to use high iso, run it through noise ninja or neat image if you feel it needs it. BTW I shot them jpg, what I normally shoot.

[dsze]

You know this had me going for a bit, to the point where I started shooting in the 160/320/640 range yesterday on my 30D. I didn't notice any major difference. So last night I set the cam up on a tripod and shot all the different ISOs and can honestly say on my 30D Real world does not show his findings. At 100% crops I started to notice noise kicking in at 400 but no big deal if your not sitting there pixel peeping. Then I took my Iso 1600 shot and ran it through Noise ninja and wahlaa, as good as is 200. Moral of this story, Shoot in whatever ISO you need to achieve the shutter speed you need and if forced to use high iso, run it through noise ninja or neat image if you feel it needs it. BTW I shot them jpg, what I normally shoot.


...Amen. Well said. Yesterday I snapped about 100 shots of my own kids at the park (something I haven't had time to do lately) and for a brief second I considered this thread as I upped my ISO to shoot in a tunnel, but quickly disregarded as it really didn't matter in real-world shooting; at least not as much as getting my shutter speed and aperture correct.