Panasonic Panasonic µ4/3 noise & noise reduction

[arpoador]

So I was reading the noise page of a review of the Sony SLT-A55 on DigitalCameraInfo.com, and came across a pair of interesting charts.

The second shows how badly the DMC-G2 (my camera) fares against the competition in terms of image noise (with in-camera noise reduction applied):

It's pretty much the noisiest camera across the board, at all ISOs. That's kind of disappointing, but not too surprising. Even though it is a new(er) camera, it uses the last-generation's chip set. But to be fair, the Nikon is actually 1.5 generations old, and it's still the best of the lot. I suspect the new D7000 would make it an even more open-and-shut case.

What gets interesting is when they show the first chart, which shows the raw noise with no noise reduction applied:

With no noise reduction, the Panasonic is kinda average. It's noisier than the others at low ISOs, but beats the average as you go up. In fact, between 800 and 3200 (which is the top of its range) it beats the older CaNikons handily, and is in the running with the much-praised NEX sensor and the brand new pellicle camera.

Even more interesting, perhaps, is to track the curves in the two charts together. The G2 with full noise reduction applied is barely different from the unreduced one. Not a very effective noise reduction system, if looked at in isolation. The others all improve dramatically, especially the CaNikons. That's visibly what gives the D5000 such a good high-ISO performance.

On the one hand, it might be easy to just fault the effectiveness of the Panasonic noise reduction system. But I wonder. They seem to be taking Leica as their mentors in lens design... perhaps they're also applying a Leica-minimalist approach to noise reduction.

In my own experience, I was at first quite disappointed in the G2's noise level at the lowest ISO, not only compared with the E-P2 I had owned earlier, but even with my little Canon S90. But after a while I found that the noise was fairly easy to kill with Lightroom3. And then I discovered that I often preferred the image with the noise, particularly if I'm reducing the resolution.

Here's an image (G2/PL45/f2.8/ISO100) first with noise reduction:

And then without:

This image is just cropped to square and converted to jpg for uploading, with no other post.

(you can compare the two with arrow keys here, or zoom into the no-noise-reduction image at full size here.)

The noise is clearly visible in the continuous surface areas, particularly if you zoom into the full size image. To my (admittedly aging) eyes, though, the noisy image is nicer. The noise actually becomes a kind of "grain", that doesn't detract significantly from the feeling of smoothness in the surfaces.

I posted an earlier thread, describing my transition from an Olympus E-P2 to a Lumix G2. In that thread, I had originally planned to complain harshly about the image quality of the G2 (particularly in low-ISO RAW images) compared to that of the E-P2, but held my fire after reading Amin's Mirrorless camera shootout, where he measured very similar results between the two cameras.

Now I'm glad I did. I think my biggest concern, really, was the sudden appearance of so much low ISO noise. Since it's so easy to knock out with Lightroom3 - when I want to - I've grown to live with it. And I kinda like having the control.

 

[kathyh]

I agree with you. I like the shot without the noise reduction better. I read somewhere, sometime that ISO 200 is the best for low noise on the G1. I assume it would be the same for the G2.

 

[BBW]

Definitely prefer the second for it's more crisp feel - with regard to the edge of that orange bowl. It just seems to be more "real" and with me when I look at it, José .

I feel that for me to really know what I think about any of the photos I take, I need to have them printed. That's my goal - to pick out some of my own photos that I really like and finally get them printed. I won't print too large because for my own use or to give to someone, I don't really like large 11 x 14 type shots. Perhaps there are some that I might like in larger sizes but many times I prefer a 5 x 7 or an 8 x 10 - perhaps that's because I live in a smallish house, I don't know. I'd like to see what "noise" translates to in print. This whole digital photography world can be so caught up in noise and pixels.... When I was happy and oblivious using iPhoto, I got some great shots with my unimpressive cameras and had them printed - they looked great. I'm pretty picky when it comes to prints but they looked good to me. So that's my plan - get some of my current photos printed....and see what I see. When I use the zoom in Lightroom or used to use the magnifying glass in Aperture, I used to worry that I had "noise" but I think sometimes that it's serious overkill to use these techniques to judge one's own photos. I realize that for professional photographers, they have to hew to a higher standard but for some of us who enjoy photography in a "serious" way and like to see our photos printed for ourselves or to give to others or even to sell on a lower key level, I do wonder about the noise issue. All that said, I am still wet behind the ears in all of this.

 

[Djarum]

BBW,

In my own experiences with my PEN, if you are printing 8x10, you won't see any noise until ISO 800. In some poorly exposed shots I was able to see some slight noise at ISO 400, but I had to be looking for it. I've also printed some ISO 200 shots at 11x14 and I didn't see any noise either. When it comes to print, only poster size prints will show the real difference between some of the larger sensors and mFT. Of course, YMMV.

 

[Djarum]

In regards to the photos, I like the first one the best. It looks smoother to me, and I don't see any benefit of the bowl without noise reduction. The soft grain in the brown table doesn't add to the picture, so the NR helps here I think.

 

[Boyzo]

Definitely prefer the second for it's more crisp feel - with regard to the edge of that orange bowl. It just seems to be more "real" and with me when I look at it, José .

I feel that for me to really know what I think about any of the photos I take, I need to have them printed. That's my goal - to pick out some of my own photos that I really like and finally get them printed. I won't print too large because for my own use or to give to someone, I don't really like large 11 x 14 type shots. Perhaps there are some that I might like in larger sizes but many times I prefer a 5 x 7 or an 8 x 10 - perhaps that's because I live in a smallish house, I don't know. I'd like to see what "noise" translates to in print. This whole digital photography world can be so caught up in noise and pixels.... When I was happy and oblivious using iPhoto, I got some great shots with my unimpressive cameras and had them printed - they looked great. I'm pretty picky when it comes to prints but they looked good to me. So that's my plan - get some of my current photos printed....and see what I see. When I use the zoom in Lightroom or used to use the magnifying glass in Aperture, I used to worry that I had "noise" but I think sometimes that it's serious overkill to use these techniques to judge one's own photos. I realize that for professional photographers, they have to hew to a higher standard but for some of us who enjoy photography in a "serious" way and like to see our photos printed for ourselves or to give to others or even to sell on a lower key level, I do wonder about the noise issue. All that said, I am still wet behind the ears in all of this.

While the thread is quite enlightening and the examples informative, I'm with BB noise in m43 at low ISO does not concern me.

The IQ of m43 is already very high for average use.

It is a small format but then so too was 35mm film I once used Film I would never go back to for obvious reasons.

With film you had grain even at ISO 50 or 100 so with 35mm the old trick was to "move in" frame the subject tight, this applies to m43 also.

The interesting grain effect with 35mm film was that with the right subject lighting (and the old Leica shooters new this) was to move in use natural contrast light so then the very grainy out of focus background did not matter so too with m43 the apparent noise in the background is insignificant relative to the main subject.