Nine 10MP DSLRs - Noise at ISO1600 Comparison

The cameras in the comparison:

Canon EOS 40D
Canon EOS XTi
Canon EOS 1D Mark III
Nikon D40x
Nikon D80
Nikon D200
Olympus E410
Olympus E510
Olympus E3
Pentax K10D
Sony A100
Sony A200

All it`s clear - let your eyes speak ;)



The reference image:
http://67.19.183.244/PRODS/D40X/THUM...0XhSLI0100.JPG

For the comparative image are used 100% crops of the originals !

http://img180.imagevenue.com/loc450/..._122_450lo.jpg

Bottom image doesn't work. Don't know if it's me...

OK - this is mostly as I would expect - except for the D200 and the XTi
something seems to have gone wrong there!!
Overall - I would say standards are getting better (- some better than others!!)

Use the α700 or the α200 over the α100 ... the α100 is not even in production these days. Its ISO-1600 performance was dubious. ... depending on what version software you were running. :cool:

The Nikons seem to have the edge...

Quote:

---

Originally Posted by sparkerjc

The Nikons seem to have the edge...

---

Don't get fooled by this flawed test though. For a little more accurate idea bout noise performance read the dpreview reviews of these cameras (and see the samples given there).

The Nikons all do chroma noise filtering, and they lose details as a result.
You can not fully turn that off, even in RAW the Nikons filter some noise.
The Olympus uses a HUGE amount of filtering, losing even more detail in the process.

If you filter that noise in the photos taken with a Canon and Pentax, things level. The 40D has the edge over all these cameras, the XTi does not do bad either.
The Pentax K10D, Nikon D200/D80/D40X and Sony A100/200 all use the same sensor family. The Nikon D40X uses the most processing of all three Nikon models, which you can see in high ISO photos (lots of detail being eaten away, strange resulting artifacts).

The best advice for high ISO shooting: make sure you expose well.

Quote:

---

Originally Posted by coldrain

The Nikons all do chroma noise filtering, and they lose details as a result.
You can not fully turn that off, even in RAW the Nikons filter some noise.

---

I believe the new Nikons (D300 and D3) do not filter any noise at all if you turn the "noise reduction" settings to off. There may be the slightest chroma noise reduction still in Jpeg, but in RAW it looks to be... well... RAW. Here are some photos of the Rush Hour "3-Pack" DVD set (it was the first thing I grabbed).

(Below) Straight Jpeg (Large/Fine) from the camera, noise reduction turned "off", purposely overexposed by 2/3 to show more detail in the blacks. Only post process was 100% crop as shown:

http://farm3.static.flickr.com/2012/...ecef9250_o.jpg

(Below) Shot in 12 bit RAW and converted to Jpeg in Capture NX. Noise reduction turned "off", porposely overexposed 2/3 to show more detail in blacks. No additional post process (except 100% crop).

http://farm3.static.flickr.com/2367/...d533c98a_o.jpg

As you can see the RAW image shows a bit more noise. I cant tell if this is the difference between in-camera Jpeg compression vs. post process jpeg conversion or if it is in fact some small amount of noise reduction. My file size for the RAW to Jpeg is 8.89 MB, and the Straight Jpeg is 7.58 MB, so there may be something going on in the camera that compressed the Jpeg image more than what Capture NX does. This could result in a loss of detail and possibly "clean up" some of the noise in the image.

My tests are confirmed in dpreview.com's tests of the Sony A-700, where they do compare it to the D300. Here is a quote from that review:

Below we have a comparison of in-camera JPEG to RAW converted using Adobe Camera RAW with its Noise Reduction options set to zero (effectively 'no noise reduction'); RAW NONR. As you can see the Nikon D300's RAW images are just that, warts and all (but at least this means you can run your own noise reduction on them if you wish and you are getting the 'digital negative').

So, I think you will have to amend your noise reduction theory to include the D300 as one camera that does not do any noise reduction in its RAW images. Unfortunately, I can not confirm if the Jpeg is in fact free of any in-camera noise reduction.

I added more three 10MP cameras:

Canon EOS 1D Mark III
Olympus E3
Sony A200


http://img180.imagevenue.com/loc450/..._122_450lo.jpg

Quote:

---

Originally Posted by VTEC_EATER

I believe the new Nikons (D300 and D3) do not filter any noise at all if you turn the "noise reduction" settings to off.

---

Both the D300 and D3 also reduce noise before RAW is saved (and probably do some other processing too).
They just do it less than other models before them.

Do you see the "light" spots in the dark areas? You do not see that in other cameras. This is a Nikon chroma noise reduction artifact. It first appeared like this in the Nikon D200.

When I use the chroma noise filter in my Canon DPP RAW converter you get the same kind of light noise dots.

Another tell tale sign: Sony uses a same processor design in its A700 as Nikon uses in the D300. Look at how different they are in this respect.
What makes Sony worse is that they also use a very smearing noise reduction in the sensor itself (according to Sony), which shamefully you can not turn off.
Just ask yourself... how come the Sony A700 with a Sony CMOS has very Canon like chroma noise, but the Nikon D300 a Sony CMOS of the same generation has a lack of chroma noise?
Yes... the chroma noise has been filtered, and replaced by "surrounding colour" but with the smae light intensity the chroma noise pixel had. This results in the "dithering" white speckled Nikon images we know of the D200, D300 and D3.
And as you can see in your own image, there is quite a loss of detail already in the processed RAW image you posted. Just try reading the small text on the back of the DVDs!

Does that make the Nikon noise reduction something bad? No of course not. But it does flatter straight out of camera results if one chooses to ignore noise reduction has been applied (both in JPEG and RAW).
It does not say anything about which camera is capable of the best results (unless you are anti post processing of course).

It does make Nikon cameras less usable for certain applications though.
Here is an interesting article for you (about the D3 and D300):
http://www.astrosurf.com/~buil/nikon_test/test.htm

Older article about the D70:
http://www.astrosurf.com/buil/d70v10d/eval.htm

Quote:

---

Originally Posted by dolphin

I added more three 10MP cameras:

Canon EOS 1D Mark III
Olympus E3
Sony A200


http://img180.imagevenue.com/loc450/..._122_450lo.jpg

---

Why make the α200, if you couldn't improve it?

Well, just doing the mere comparison between the two SONY Alphas ... I think Coldrain has some serious reconsideration this week, on this low-cost beauty. :eek: LOL

Good job! :D