D7000 High ISO NOT THAT GOOD

[kirbinster]

I was very surprised to see in side-by-side tests the D7000 high ISO does not even look as good as the D5000. Check this out: Digital Cameras: Digital Photography Review, News, Reviews, Forums, FAQ

I had it compare the D7000, D5000 and D3000. Unfortunately they don't have the D700 as a comparison choice.

[TheMoons678]

thats pretty cool, good find.

[vandergus]

I don't know, looks pretty good to me. Kind of a toss up to 1600 ISO, but the D7000 looked better at 3200. Still pretty subjective, though, and maybe not as great a difference as some people have suggested. Also remember that with that widget, the D7000 image is magnified to a greater extent than the D5000 image. Comparing them at like magnifications would make the D7000 look a little better.

[kirbinster]

Why do you think the D7000 is magnified more that the D5000, they are both DX bodies. Also I don't consider 1600 really high ISO. Put the frame on the white cross with the black background, the D7000 seems to have a lot of color noise in the black areas. Also be sure to compare raw not jpeg.

[prabbit237]

Originally Posted by kirbinster

Why do you think the D7000 is magnified more that the D5000, they are both DX bodies.

Look what's shown on the screen. If you move it so the cross is just barely off-screen at top and bottom on the 7000, it shows some stuff above and below the cross on the 5000. It may be due to the **********ing that's showing the pictures on the page but the 7000 is definitely magnified by maybe 10% or so SOMEWHERE along the line.

[inkista]

Welcome to the old Canon 50D vs. 40D high iso noise performance controversy. :^)

In your mind, resize the crosses to be the same size. Then realize that the noise will resize, too, if you're judging noise across the whole image.

When an image is bigger in a 100% crop, that means the sensor has a higher resolution with more megapixels, and will look worse, because the image is at higher magnification. Here's a good explanation of how resolution can affect the appearance of noise in 100% crops: The Real Megapixel Myth gives an example of how the dpreview test of judging noise by 100% crop can be misleading across sensors with significant resolution differences if you don't know what you're looking at:

first, here's how dpreview's crops for the D3x (left) and D3 (right) look:


But note the size difference of the images. At equal magnification, where the image is the same size (the D3x was downrezzed):



Megapixels still count.

[Sterling]

Why do you think the D7000 is magnified more that the D5000, they are both DX bodies.

DX just means they have the same sensor size, not that they produce the same image size.

[dcclark]

Agreed on the size problem -- the D7000 has a boatload of pixels, and displaying the images at the same size would really change things.

In those examples, I think that the D7000 shows a lot more detail at the same ISOs, compared to the D5000.

[inkista]

Ok, quick and dirty, but I downloaded the JPEGs for the RAW test [added: at iso 3200] (I had no idea how to open a D7000 RAW). The D7000's was 4928x3264, the D5000's was 4288x2848. I split the difference and resized both to 4608 wide (so not biasing towards uprezzing or downrezzing), with proportional sizing. Then cropped. Results. D5000 on the left, D7000 on the right.

D5000

D7000

So, which one has the better high ISO noise performance? The one that's better at the per-pixel level? Or the one that's better at the overall image level? Dpreview thinks that the per-pixel performance is more important on the assumption that this translates to real world usage, so they show you all the pixels at the same size. And when all the sensor resolutions were equable, it probably worked well. But now, many of us think they are wrong to judge noise performance this way, because a lot of folks cannot correctly interpret what those 100% crops actually mean.

[vandergus]

Wow...

Which ISO value did you pick?