50D v 350D noise test

[carregwen]

I thought I would see what difference there is between the two cameras (I own both). There has been a lot of general talk in photography mags etc on the high pixel rate of the 50D and whether it is too much for a crop sensor. The 350D is 8mp, and the 50D is 15mp - nearly twice the amount for the same size sensor.

Both shots taken on a tripod in low light at ISO400, f16 ( I was trying to simulate landscape shots). The 50D had 'noise reduction' set to Standard. The 350D does not have noise reduction. I used the same lens for both - a Sigma 105mm macro, and the same framing/FOV. Shot in RAW with no processing at all. These shots below are screen-prints from the RAW editor. The last two are at 200% zooming in on the phone keypad.

I think the 50D is a bit brighter, but there does seem to be slightly more noise/grain - look in the LCD screen of the phone.

What do you think?

The full shot as it was from both cameras


The 350D @ 200%


The 50D @ 200%

[bmpress]

I thought I would see what difference there is between the two cameras (I own both). There has been a lot of general talk in photography mags etc on the high pixel rate of the 50D and whether it is too much for a crop sensor. The 350D is 8mp, and the 50D is 15mp - nearly twice the amount for the same size sensor.

Both shots taken on a tripod in low light at ISO400, f16 ( I was trying to simulate landscape shots). The 50D had 'noise reduction' set to Standard. The 350D does not have noise reduction. I used the same lens for both - a Sigma 105mm macro, and the same framing/FOV. Shot in RAW with no processing at all. These shots below are screen-prints from the RAW editor. The last two are at 200% zooming in on the phone keypad.

I think the 50D is a bit brighter, but there does seem to be slightly more noise/grain - look in the LCD screen of the phone.

What do you think?

My interpretation of your two shots looks like you have higher resolution with the 50d and about equal noise performance. I have had a 50d for about six months and comparing my actual pictures to my old 20D which is 8 mpix, the 50d is infinitely better in all respects.

[carregwen]

My interpretation of your two shots looks like you have higher resolution with the 50d and about equal noise performance. I have had a 50d for about six months and comparing my actual pictures to my old 20D which is 8 mpix, the 50d is infinitely better in all respects.

I'm finding about the same. The noise is about the same, but the extra resolution detail and the fact you can print larger makes the 50D a much better camera. The added functionality in the 50D is obviously a big bonus over previous cameras. I do wish people in other fora and magazines would stop banging on about the 50D have 'noise problems'. It doesn't - although I would think it now very difficult to squeeze any more pixels into a crop-factor sensor. I see the new G11 from Canon has actually REDUCED the mega-pixel figure to address noise concerns in the G10. But then the G cameras have even smaller sensors than APS-C.

[Dave Humphries]

Rob, Barry,

My eyes are seeing more noise in the 50D shot, sorry - even on the pale background wall.

It's a higher frequency than that from the 350D (no surprise there), but I'd say it is no more than I get with my Nikon D5000 at 400ISO and easily dealt with by the NR in ACR, i.e. without needing to go to more extreme solutions like Neat Image.

[kevinbythebeach]

lol this is the second noise comparason Ive seen posted on the web with a 50D versus whatever, where the poster zooms in on a phone..

[carregwen]

lol this is the second noise comparason Ive seen posted on the web with a 50D versus whatever, where the poster zooms in on a phone..

Well, I'm still not giving you my phone number...kiss, kiss, kiss,

[Colin Southern]

I'm missing something here ...

If the two original shots have the same FOV (or are cropped to the same FOV as these are) then they should have different pixel dimensions, and yet they're the same (or close enough).

[carregwen]

I'm missing something here ...

If the two original shots have the same FOV (or are cropped to the same FOV as these are) then they should have different pixel dimensions, and yet they're the same (or close enough).

I put the shots into ACR. Selected the same area of the phone - which took them up to 200%. I then used the Vista screen-grab tool which created an image of what was in view, so they should be the same size. No?

The 1023x711 pixels is of the screen-grab shot.

[Colin Southern]

I put the shots into ACR. Selected the same area of the phone - which took them up to 200%. I then used the Vista screen-grab tool which created an image of what was in view, so they should be the same size. No?

The 1023x711 pixels is of the screen-grab shot.

No - at 100% (or 200%) "magnification" there should be more pixels in the 50D crop (or for the same number of pixels on the screen you'd appear to be "more zoomed in" on the 50D). If there's not then something must have been down-sampled somewhere - and down-sampling changes noise.

The screen grab may upset the applecart too (I'm wondering if the video drivers may introduce anti-aliasing or other "distortions").

Sean (McQ) is more of a wizz on this kind of thing that I am - he posted on a similar topic not too long ago ...

Which is better: low resolution mode (sRAW) or downsizing afterwards?

Normally I suggest that people don't get too carried away with noise in modern DSLR cameras - for the most part you can only see it at high magnifications on screen; in a realworld print it's too small to see.

[carregwen]

No - at 100% (or 200%) "magnification" there should be more pixels in the 50D crop (or for the same number of pixels on the screen you'd appear to be "more zoomed in" on the 50D). If there's not then something must have been down-sampled somewhere - and down-sampling changes noise.

I don't think that sounds right. Are you saying that if you took two shots with the same lens but different mega-pixel rated cameras the view would look different when you zoomed in with Photoshop to the same zoom level? I don't think that's right.

I took another two shots just now of the back of a book. Here is one of them just to show what I shot


And here is the EXIF for both shots. You can see the differences in physical size and pixel dimension etc.



I then zoomed in to the full stop at the end of the word 'fame'. You can see that the number of pixels to represent the stop are different (as they should be) The 50D has MORE pixels, but if you looked at both shots in full the full stop would take up the same percentage area of the whole shot (as they should). The extra pixels in 50D help give more definition, but the don't affect the field of view of the shot. When you put a RAW into ACR it may have a different overall physical size compared to the same shot with another camera, but what you see in the viewr is exactly the same. And when you zoom it also looks the same.

350D shot


50D shot

[Colin Southern]

At 100% view you should have a 1:1 correlation between pixels in the image and pixels on the monitor, so if the full stop taken with the 350D takes up 35 x 35 pixels and the full stop taken with the 50D takes up 50 x 50 pixels then - at the same zoom level - the full stop taken with the 50D should be wider on the screen.

If they're the same size (on the screen) then I can only deduce that one of them isn't at 100% (ie not a 1:1 correlation).

[carregwen]

At 100% view you should have a 1:1 correlation between pixels in the image and pixels on the monitor, so if the full stop taken with the 350D takes up 35 x 35 pixels and the full stop taken with the 50D takes up 50 x 50 pixels then - at the same zoom level - the full stop taken with the 50D should be wider on the screen.

If they're the same size (on the screen) then I can only deduce that one of them isn't at 100% (ie not a 1:1 correlation).

I think I have worked out what's going on here, but I'm baffled as to why. If you load two RAW files into ACR - one shot with a 50D and the other with 350D, both of the same FOV scene, they do NOT change relative FOV when you zoom in by the same amount for each shot. At least they don't on my CS4 version (ACR 5.4). So, the phone shots above in post #1 which are both zoomed to 200% are as they were on screen in ACR, and they show the same field of view, even though one is 8mp and the other is 15mp and they have different physical sizes.

However... if you put the JPEG versions into CS4 and zoom both by the same zoom factor they are different. The 50D appears to zoom in more.

Can anyone help on this? I hadn't noticed this before, but then you wouldn't unless you were comparing two different versions of the same shot, as I was here. There are no settings in ACR that controls zoom.

[Dave Humphries]

Hmmm, I don't know the answer, but here are a couple of things to consider;

Were both files open in ACR at the same time?
If so, which one was opened first (I'm thinking this may set the zoom ratio for all pictures opened subsequently, which would normally be from same camera, so again you wouldn't notice)

If you open them one at a time in ACR (open one, screen grab, close, open other, screen grab, close), do you get the same result?
If you open them in the opposite order to last time, what happens?

Whereas CS4 has a button called 1:1, which I am guessing it does with reference to the current image only.

Just some thoughts

[kevinbythebeach]

reading this thread has made my head hurt....

Cant we just agree that the 50D is better than a 350D in every aspect???

[carregwen]

Were both files open in ACR at the same time?

Yes. It's the same no matter how they are loaded. Perhaps I should ask Adobe?

[carregwen]

Cant we just agree that the 50D is better than a 350D in every aspect???

Well, I think it's a lot better. I just can't understand all of the critics 'out there' who keep banging on about 50D noise. My head hurts too, but that's probably down to the Californian wine I just drank.

[Colin Southern]

I just can't understand all of the critics 'out there' who keep banging on about 50D noise.

I suspect that a large part of the problem is that people are comparing "apples" with "oranges". Having a higher pixel density (talking 50D here) may (in the absense of other technologies) give rise to more noise than, say, a 350D (with a lower pixel density) but - what they're failing to appreciate - is that the noisy pixels are physically smaller in a real-world image (but will appear the same size when viewed at 100%). I'm guessing that people working at 100% zoom are forgetting that they're "zoomed in" more with higher MP cameras.

I remember once when I had to shoot a production with a 20D set to an uncomfortably high ISO ... at 100% magnification they looked pretty noisy, but when viewed at postcard size, they all looked surprisingly clean.

I'm tempted to suggest that the ideal answer would be to disable any Photoshop views greater than about 33% (and all of a sudden all of these issues would just melt away!).