Ian, can you check with which camera you can use higher iso with no or acceptable noise. (sorry for my english, I am dutch)
regards Burry

Burry - your English is very good - put it this way, you don't want to even hear my Dutch !

I'll prepare some noise comparisons for tomorrow morning and sone lab noise measurements using DXO Analyzer later in the week.

Ian

(Updated - see Pentax *istDL exposure corrections)
The following are 100% (1 image pixel to 1 screen pixel) crops from the exact centre of the frame. All the cameras were tripod mounted. Adobe Photoshop CS2 Smart Sharpen @50% radisu 1 pixel has been applied to each image.

Canon EOS-350D (Digital Rebel XT)
ISO 800

ISO 1600

Konica Minolta Dynax 5D (Maxxum 5D)
ISO 800

ISO 1600

Nikon D50
ISO 800

ISO 1600

Olympus E-500 (EVolt)
ISO 800

ISO 1600

Pentax *istDL
ISO 800

ISO 1600

The above are clearly underexposed, so I have re-done them with +0.5EV correction:
ISO 800

ISO 1600

The Nikon is clearly the cleanest and most neutral result closely followed by the Canon. I would not be so happy with the others, both colour shift and/or higher levels of noise are apparent, the Oly predictably so. Is this fair would you say?

The problem with my test is that there is such a wide range of exposure variations from camera to camera.

I've printed central crops of the shots to an equivalent size of about 10 inches high (for the full frame height), so a 3:2 frame would print to about 10x15 inches.

I've not fully examined them yet, but the real test is not the detail in ceramic, but some surrounding areas the are cropped out in the first set of samples.

I'll scan the prints and post them here a bit later.

Ian

> The Nikon is clearly the cleanest and most neutral result
> closely followed by the Canon. I would not be so happy
> with the others, both colour shift and/or higher levels of
> noise are apparent, the Oly predictably so. Is this fair
> would you say?

Hi Ian,
The 5D and E500 are so far out it's hard to believe they were shot under the same conditions.
Did you use RAW or Jpegs for these tests and as the WB a custom one or on auto?
Your comment regarding exposue variations also intregued me, what metering did you use and was the exposure manual or auto?
Tony

I would agree with you Tony. Was the cameras metering used at all Ian, or did you use exactly the same exposure for both on manual. Are you using continuous lighting or flash. Surely the test was to show noise at high ISO's, not whether the cameras metering are all exactly the same in terms of accuracy.

It would seem reasonable to use the same colour temp setting at the very least whether it be a custom setting or altered in the raw converter.

I'm aftaid I don't understand the reasoning behind scanning a print in this instance

All shots were taken using auto (aperture priority) exposure at f/8, with auto white balance, diffused daylight from a window through net curtains and the meter mode was full frame evaluative. The images were recorded to best quality setting JPEGs.

I'm going to repeat the experiment using fixed white balance and exposure settings.

Ian

As promised, I have repeated the test with the white balance and exposure settings constant.

As it's night time now, I have used incandescent tungsten lighting in the room and set the cameras to their tungsten white balance presets. The exposure was metered using the Canon EOS-350D. (1/10th second f/8 for ISO 800, 1/20th second f/8 for ISO 1600).

These are quite slow shutter speeds, but digital sensors don't really suffer from reciprocity failure like film and the conditions are representative of real conditions a photographer will encounter.

Well, despite this, wide ranging variability in the results from camera to camera remains!

The Canon seems to have a lot more sensitivity than the others and produced the brightest images. The choice of tungsten preset seems also to be a lottery

See for yourself:

Canon EOS-350D (Digital Rebel XT)
ISO 800

ISO 1600

Konica Minolta Dynax 5D (Maxxum 5D)
ISO 800

ISO 1600

Nikon D50
ISO 800

ISO 1600

Olympus E-500 (EVolt)
ISO 800

ISO 1600

Pentax *istDL
ISO 800

ISO 1600

I wanted to see what happened when using the cameras as most of the potential owners would use them - on auto. The results speak for themselves!

As for printing - the logic is very clear to me; that evaluating an image on-screen is not the same as evaluating a print. And the prints confirm this because the notable on-screen noise differences in the ceramic areas are practically non-existent in the prints. Where there *are* noticeable differences are in the deeper shadows that aren't shown in the crops.

Ian

In another attempt to get things more consistent, I have used customer white balance calibration on each camera and attempted to make the exposure results more consistent, but it's not easy.

Canon EOS-350D (Digital Rebel XT)
ISO 800

ISO 1600 - this looks rather soft, probably because of a focus error. The 350 D does seem more temperamental in low light with the 18-55 kit lens than the others.

Konica Minolta Dynax 5D (Maxxum 5D)
ISO 800

ISO 1600

Nikon D50
ISO 800

ISO 1600

Olympus E-500 (EVolt)
ISO 800

ISO 1600

Pentax *istDL
ISO 800

ISO 1600

The Nikon still seems to produce the cleaner image to me, at least there are less artefacts in the 800 image. Frankly I would have expected the Canon to be better in this respect.

I wonder Ian if the sharpening you are applying to the files may not be helping with this. It could be for example that the amount of sharpening applied to the file in camera is different in the cameras, and therefore the standard Smart Sharpen you are applying may not be appropriate. For this reason, I still think the use of a Raw file would be better and allow you to put ina fixed colour temp to each file.

The slightly oof Canon 1600 shot you have shown seems somewhat unfair in terms of a direct comparison, and may suggest some inferiority in that camera. The ones shown previously certainly seemed OK.

Some fair comments there Stephen. I think RAW is a red herring here as not all the cameras are supported by the conversion tools I have here. JPEG is what a lot of the users of these entry-level cameras will be using by default.

It's still interesting to note the imperfections in exposure and white balance that have been unearthed.

I'm travelling today and tomorrow but I will try to keep in touch where I can.

Ian

Again the Nikon comes out best which, as a Nikon D70 user, is not a surprise. I've always been quite pleased with it's performance at ISO1600.
I'm rather surprised Ian you've used AF, I'd have thought manual focus would have been prefereable for these closeups. Again is the logic that this is how the average user of one of these cameras would work.
On this evidence KM, Oly and Pentax have some serious WB problems.

Ian,
Is it possible to test also with 400 iso. I don't know if I will use 800 or 1600, it looks terrible with the E500. I tried to remove the noise with neat image but after that the canon 350 gave better result at 1600 than the E500 at 800.
Burry

Yes, 400 ISO testing - no problem - I will do this on Thursday evening after I return from Munich.

I think it is taken for granted that the E-500 high ISO noise characteristic means it's around a stop behind most other DSLRs, if not a little more with some.

Ian

I'm in Germany on a press trip until tomorrow evening. I was rather in a hurry to get the latest results online before I left for the airport yesterday. I'll re-do the Canon tests when I get back.

In this instance AF should have been fine as the focus target is not particularly amorphous, so the Canon AF inconsistency is interesting. None of the other cameras appeared to have a problem.

Regarding white balance, even when calibrating the cameras with a white card, there are notable differences.

Ian

Hi everybody!

I bought today november edition of british Practical Photography magazine. They tested KM 5D and Nikon D50 side by side and the result is something nobody else has reached:

"there is much more noise on all of thy Nikon's files comapred to the respective ISO setting on the Konica Minolta..."

Hi Olavia and welcome to the DPNow discussion forum!

I can't agree with the Practical Photography findings. My tests show that the D50 managed to combine low noise and detail retention while the Konica Minolta 5D appears to be sacrificing a lot of detail in order to suppress noise.

See my latest test results:

http://dpnow.com/bb/cheaper-dslrs.pl?frames=n;read=39

Ian

> Hi everybody!

> I bought today november edition of british Practical
> Photography magazine. They tested KM 5D and Nikon D50 side
> by side and the result is something nobody else has
> reached:

> "there is much more noise on all of thy Nikon's files
> comapred to the respective ISO setting on the Konica
> Minolta..."

I have re-done all the tests and included ISO 400:

http://dpnow.com/bb/cheaper-dslrs.pl?frames=n;read=39

Ian

> Ian,
> Is it possible to test also with 400 iso. I don't know if
> I will use 800 or 1600, it looks terrible with the E500. I
> tried to remove the noise with neat image but after that
> the canon 350 gave better result at 1600 than the E500 at
> 800.
> Burry