Nikon D200 and Canon 350D/Rebel XT Noise Tests

Hi all,

I've placed on DIMi a new section and started it with noise tests on
the Nikon D200 and Canon 350D/Rebel XT. There are also uncompressed
TIFFs that you can download to judge noise and/or resolution issues for
yourself. The new page is at http://
www.dimagemaker.com/specials/cameras/camtests.php

BTW the D200 and 350D tests were done using the same lens, a Tamron
18-200mm XT lens. More tests will be going up this week.

Also don't forget the DIMi competitions, there are now three of them
running in April.

Cheers,

Wayne

Wayne J. Cosshall
Publisher, The Digital ImageMaker, http://www.dimagemaker.com/
Blog and Podcast http://www.digitalimagemakerworld.com/
Personal art site http://www.artinyourface.com/

Hi All,

My apologies but I have placed revised TIFF files for the Canon 350D
test up on DIMi and in the article. I discovered a mistake in the
creation of the original ones. The Nikon D200 tests were fine. The
articles are at http://www.dimagemaker.com/index.php
and the TIFF files are at
http://www.dimagemaker.com/specials/cameras/camtests.php

Again my apologies, but I corrected them as soon as I realised the
error.

Cheers,

Wayne

Wayne J. Cosshall
Publisher, The Digital ImageMaker, http://www.dimagemaker.com/
Blog and Podcast http://www.digitalimagemakerworld.com/
Personal art site http://www.artinyourface.com/

Paul Furman wrote:
wayne wrote:

Hi all,

I've placed on DIMi a new section and started it with noise tests on
the Nikon D200 and Canon 350D/Rebel XT. There are also uncompressed
TIFFs that you can download to judge noise and/or resolution issues for
yourself. The new page is at http://
www.dimagemaker.com/specials/cameras/camtests.php

The D200 samples are exposed a tad darker according to the histogram but
the noise looks about the same to me.


Eh? I thought that the noise was considerable more noticable on the 350D,
which is surprising to me considering reports elsewhere to the contrary.

--
Thomas T. Veldhouse
Key Fingerprint: 2DB9 813F F510 82C2 E1AE 34D0 D69D 1EDC D5EC AED1

wayne wrote:

Hi All,

My apologies but I have placed revised TIFF files for the Canon 350D
test up on DIMi and in the article. I discovered a mistake in the
creation of the original ones. The Nikon D200 tests were fine. The
articles are at http://www.dimagemaker.com/index.php
and the TIFF files are at
http://www.dimagemaker.com/specials/cameras/camtests.php

Again my apologies, but I corrected them as soon as I realised the
error.

I just checked the 800 ISO tiffs & the D200 is a lot darker than the
rebel sample. It looks like you matched the zoom this time.

Paul Furman wrote:
Again my apologies, but I corrected them as soon as I realised the
error.

I just checked the 800 ISO tiffs & the D200 is a lot darker than the
rebel sample. It looks like you matched the zoom this time.

I would like to know the details of the conversion. Was the original in RAW
format? How was the conversion handled? Were any settings applied in post
processing (i.e. automatic exposure adjustments, levels, etc)?

Details that DO matter before noise can be discussed.

--
Thomas T. Veldhouse
Key Fingerprint: 2DB9 813F F510 82C2 E1AE 34D0 D69D 1EDC D5EC AED1

In message ,
"Thomas T. Veldhouse" wrote:

Paul Furman wrote:
Again my apologies, but I corrected them as soon as I realised the
error.

I just checked the 800 ISO tiffs & the D200 is a lot darker than the
rebel sample. It looks like you matched the zoom this time.

I would like to know the details of the conversion. Was the original in RAW
format? How was the conversion handled? Were any settings applied in post
processing (i.e. automatic exposure adjustments, levels, etc)?

Details that DO matter before noise can be discussed.

This is one of the reasons I call for RAW comparisons (RAW, not
conversions). The D200 image looks darker overall, but it also has more
contrast. There is no difference in contrast in the RAW files; RAW
files are records of photons captured, plus or minus noise. "Properly"
converting RAW files distorts the capture in many ways; some colors have
saturation boosted, which can boost noise; different transfer curves are
used by default, noise reduction is applied; etc, etc. I am interested
in the RAW capture, only. JPEGs are for people who like to trade
quality for speed and efficient storage, but you can never see what the
camera captured in the RAW data from its JPEG, or an arbitrary RAW
conversion in software.
--


John P Sheehy

wrote:

This is one of the reasons I call for RAW comparisons (RAW, not
conversions). The D200 image looks darker overall, but it also has more
contrast. There is no difference in contrast in the RAW files; RAW
files are records of photons captured, plus or minus noise. "Properly"
converting RAW files distorts the capture in many ways; some colors have
saturation boosted, which can boost noise; different transfer curves are
used by default, noise reduction is applied; etc, etc.

Would the *same* raw conversion be valid? (honest question, as in, "both
images converted with ACR with these settings"?) The reason I ask is that
most of us have no way to actually look at the raw data in the raw, so to
speak.

--
Jeremy |

In message ,
Jeremy Nixon wrote:

Would the *same* raw conversion be valid? (honest question, as in, "both
images converted with ACR with these settings"?) The reason I ask is that
most of us have no way to actually look at the raw data in the raw, so to
speak.

I don't think such a beast exists. Every RAW format has information
about how the information should be processed, and/or the developer
deals with different cameras in different ways.

You just load the RAW into IRIS, subtract the blackpoint (which you may
have to determine from a blackframe at the same ISO), select "convert a
CFA image" (after choosing the camera from preferences, or one with the
same CFA pattern) from the menu, and then white-balance it. Do the same
with another camera at the same f-stop and aperture, scale so both are
equally bright, and you have as equal a comparison as you're going to
get. You need a detailed subject, of course, if you want to see the
visibility of detail through the noise, as cameras differ in AA filter
strength.

For blackframe noise, you just load the file and scale both the same in
the output. It is convenient to add or subtract an offset to get the
average value of noise the same. For example, most Nikons have black as
0; most Canons have black at 126 to 256. The Canon format has a slight
benefit for binning, as binning a blackfframe with both positive and
negative noise brings black closer to true black than a mix of half
zeros and half positive numbers do (it also helps for removing banding).

If it sounds complicated, it's only because it's unfamiliar.
--


John P Sheehy

wrote:

I don't think such a beast exists. Every RAW format has information
about how the information should be processed, and/or the developer
deals with different cameras in different ways.

Yes, I think you must be right.

You just load the RAW into IRIS,

Unfortunately, IRIS is Windows-only, so not terribly relevant.

--
Jeremy |

Jeremy Nixon wrote:
wrote:

This is one of the reasons I call for RAW comparisons (RAW, not
conversions). The D200 image looks darker overall, but it also has more
contrast. There is no difference in contrast in the RAW files; RAW
files are records of photons captured, plus or minus noise. "Properly"
converting RAW files distorts the capture in many ways; some colors have
saturation boosted, which can boost noise; different transfer curves are
used by default, noise reduction is applied; etc, etc.

Would the *same* raw conversion be valid? (honest question, as in, "both
images converted with ACR with these settings"?) The reason I ask is that
most of us have no way to actually look at the raw data in the raw, so to
speak.


There is more to that. If you pick the same ISO, shutter speed and apeture
for both cameras and shut off all automatic adjustments in ACR, you still
don't have the same thing. You also need to find an equivalent ISO.
Unfortunately, brands of cameras are not consistant in this area. Some shoot
ISO200 at ISO180 for instance. Thus a close match would be one where exposure
compenstated for this ... so, judgement by the histogram is required.

--
Thomas T. Veldhouse
Key Fingerprint: 2DB9 813F F510 82C2 E1AE 34D0 D69D 1EDC D5EC AED1