[Rondec]

My first argument was the Nokton 25/0.95 which allows me to shoot at ISO 200 and not worry about sensor size.

Cosina seems to be making my wishes come true since they just announced a 17/0.95 lens.

One reason was enough for me to get the E-PL2. Now I got two to love this system. Three if you count the OM-D body. Good enough for me. YMMV, of course.

The biggest improvement in sensor technology has been improved dynamic range. This is definitely something where cameras from the Kx through the K5 (and I suppose the K-01) really overwhelm earlier cameras. At base iso, dynamic range in the K5 is 14 Evs.

This is the area where four thirds sensors really have lagged behind on the technology curve. They aren't too bad from a noise standpoint, but from a dynamic range standpoint at low isos, they just can't measure up.

[abacus07]

My first argument was the Nokton 25/0.95 which allows me to shoot at ISO 200 and not worry about sensor size.

Cosina seems to be making my wishes come true since they just announced a 17/0.95 lens.

One reason was enough for me to get the E-PL2. Now I got two to love this system. Three if you count the OM-D body. Good enough for me. YMMV, of course.

the iso improvements in digital cameras have started to break the dependence on aperture for shutter speed. You no longer need a big aperture to have a fast shutter speed.

iso = shutter speed

aperture = depth of field

There are plenty of times that I want a fast shutter speed but also want depth of field in low light. A K-5 is much better than M43 for this. M43 is a nice little system but it's not perfect for everyone.

[ElJamoquio]

M4/3, obviously, because I already have all I need for this scenario. My E-PL2 is my favorite camera for low light shooting these days.

BTW, why exactly do you think that FF sensors are better than MF ones in low light? If your logic would be true, why waste money on anything less than a MF sensor?

Not that I'm ever going to buy a 645D, but if I did, low light ISO would be one of the lower-ish priorities of mine. That said those scores are normalized to pixel size.

[gazonk]

There are plenty of times that I want a fast shutter speed but also want depth of field in low light. A K-5 is much better than M43 for this. M43 is a nice little system but it's not perfect for everyone.

I just compared with dxomark today and discovered that the GX1, which has the 16mp sensor that the OM-D probably also uses, would be a step down from my K10D. Set to the minimum ISO 160, DR and noise is at a similar level to ISO 200 on my K10D. And in fact, if I want maximum DR and cleanness, I avoid ISO 200 on my K10D - whenever I can, I use ISO 100. Although the high ISO of the GX1 is just very slightly better than the K10D, I conclude that state-of-the-art m4/3 still can't quite match a more than 5 year old model.

[johnmflores]

I just compared with dxomark today and discovered that the GX1, which has the 16mp sensor that the OM-D probably also uses, would be a step down from my K10D. Set to the minimum ISO 160, DR and noise is at a similar level to ISO 200 on my K10D. And in fact, if I want maximum DR and cleanness, I avoid ISO 200 on my K10D - whenever I can, I use ISO 100. Although the high ISO of the GX1 is just very slightly better than the K10D, I conclude that state-of-the-art m4/3 still can't quite match a more than 5 year old model.

I take tests for what they are - tests. In my personal experience, having taken thousands of photos with the GH2 (similar sensor to the GX1 and OM-D) and the Pentax K20d, the cameras are very nearly equal in image quality. Close enough that if you put two unlabeled photos side by side people would not be able to tell the difference. M43 may lag a bit in IQ, but is actually sharper in many instances.

[abacus07]

I take tests for what they are - tests. In my personal experience, having taken thousands of photos with the GH2 (similar sensor to the GX1 and OM-D) and the Pentax K20d, the cameras are very nearly equal in image quality. Close enough that if you put two unlabeled photos side by side people would not be able to tell the difference. M43 may lag a bit in IQ, but is actually sharper in many instances.

This is actually what we're saying. 4 yr old APSC Pentax's are matching current (or close to) M43's. If it's important to a person the K-5 is a step up in performance. That's not a knock on the M43's. They are very good and compact.

[Unsinkable II]

My first argument was the Nokton 25/0.95 which allows me to shoot at ISO 200 and not worry about sensor size.

Cosina seems to be making my wishes come true since they just announced a 17/0.95 lens.

One reason was enough for me to get the E-PL2. Now I got two to love this system. Three if you count the OM-D body. Good enough for me. YMMV, of course.

My mileage varies very greatly.

The "25/0.95"... a US$1,200 manual lens which provides the equivalent DOF and inferior image quality compared to a US$100 autofocus lens on full frame?

The "17/0.95" is really a 35/1.8 for US$1,500 instead of US$200. Sending its image to a poorer sensor.

[wjjstu]

Heh the OM-D with EVF and tiltable screen only has 1 extra cm added to the viewfinder hump but is otherwise similar or smaller than K-01 in the other dimensions. I wonder if Pentax will consider an actual mirrorless system in the future rather than one-off hacked K-mount SLR's like we got with this.

[ogl]

My mileage varies very greatly.

The "25/0.95"... a US$1,200 manual lens which provides the equivalent DOF and inferior image quality compared to a US$100 autofocus lens on full frame?

The "17/0.95" is really a 35/1.8 for US$1,500 instead of US$200. Sending its image to a poorer sensor.

+1

[johnmflores]

This is actually what we're saying. 4 yr old APSC Pentax's are matching current (or close to) M43's. If it's important to a person the K-5 is a step up in performance. That's not a knock on the M43's. They are very good and compact.

And their closer to current APS-C cameras than many think or want to believe:

[K1D27H]

And their closer to current APS-C cameras than many think or want to believe:

Uh, the one from Panasonic....I can see luminance noise even when viewing from distance. The one from K-5? not that obvious.

[Laurentiu Cristofor]

The OM-D with that lens will weigh very close to 1 kg, it's not very light-weight anymore...

With my EPL2, the resulting combo is 705g.

Compare that to the K-5 which is 661g just body only. The Samyang 35/1.4 is 417g. Even the K-r is 544g.

Also, lightweight and small are not a concern for me - I didn't pick the E-PL2 combo because it was lighter. And if I would want an even smaller and more lightweight combo, there are other MFT lenses that provide that.

MF sensors are behind on the technology curve compared to FF sensors.

Yes, but my suggestion was to think about why that is the case.

If you wish to believe FF mirrorless isn't coming, that's fine.

Oh, I certainly believe they are coming. I am just not waiting for them.

This is the area where four thirds sensors really have lagged behind on the technology curve. They aren't too bad from a noise standpoint, but from a dynamic range standpoint at low isos, they just can't measure up.

Well, large DR is great, but how often do you actually take full advantage of it? As long as I can take my shots without losing information, I don't care if my DR is 10Ev, 12Ev, or 14Ev. The dxomark bar for determining high ISO performance is set at 9Ev - that is considered good DR in an image. I rarely needed more DR than the 10 Ev that my E-PL2 offers - incidentally, it's the same performance I get from my K-7, so it's not like I feel I am moving backwards. In fact the E-PL2 scores higher in high ISO score as well.

the iso improvements in digital cameras have started to break the dependence on aperture for shutter speed.

You can compensate with ISO, but why not use both low ISO and fast apertures? ISO improvements are also not huge - barely a bit over one stop between K10D and K-5.

That said those scores are normalized to pixel size.

Not sure what you mean by that and what you are trying to imply.

I just compared with dxomark today and discovered that the GX1, which has the 16mp sensor that the OM-D probably also uses, would be a step down from my K10D. Set to the minimum ISO 160, DR and noise is at a similar level to ISO 200 on my K10D. And in fact, if I want maximum DR and cleanness, I avoid ISO 200 on my K10D - whenever I can, I use ISO 100. Although the high ISO of the GX1 is just very slightly better than the K10D, I conclude that state-of-the-art m4/3 still can't quite match a more than 5 year old model.

The GX1 in high ISO scores higher than the K20D by about as much as the K20D does over the K-7. You might remember how much the K-7 was bashed for such difference.

Also, keep in mind that the high ISO score tells you how much more you can push the ISO while keeping the IQ above a certain bar. The GX1 can do this up to 703, the K20D up to 639, the K-7 up to 536, and the K10D up to 522. The GX1 holds the IQ bar better than any of those cameras. And MFT sensors will get better in time as they'll get more money for R&D. Keep in mind that this is a different technology from CMOS or CCD, so it improves at a different rate.

The "25/0.95"... a US$1,200 manual lens which provides the equivalent DOF and inferior image quality compared to a US$100 autofocus lens on full frame?

Inferior IQ by what rule? Theoretically, it offers the same IQ. Practically, a better lens will offer better IQ than an inexpensive one. Build is another advantage. In the end, the combos would cost the same, so whether you purchase an expensive body to go with inexpensive lenses or an inexpensive body to go with expensive lenses is up to you. I decided to invest in lenses because bodies will always get better in time.

Personally, I don't care about FF - I compared with APS-C - and on Pentax, the only 24/1.4 and 35/1.4 lenses are provided by Samyang. And they are much larger and heavier than the Noktons. Funnily, combined with a K-5 they cost about the same, so no price savings either.

Talk is cheap though, so here are some images from my MFT camera, working hard to overcome the limitations of its small sensor through the use of superior glass.















And for the last demo, here's a shot with a 100% crop from the border - just in case you are wondering whether these only look good when scaled down for web sharing:



100% crop from the above:



I am happy with these results, as far as IQ goes. I am hitting my artistic limits more often than my equipment limits.

[ElJamoquio]

Not sure what you mean by that and what you are trying to imply.

'What I mean' is that the SNR for that test is really only 'valid' for a given megapixel size, and comparing across various sensor and pixel sizes during 'actual use' with post-processing is not as straightforward as 'one number' might suggest.

For example I'd much rather have a low SNR with more pixels than a 'medium' SNR with low pixels. For a perhaps bad example but one I'm familiar with - if you're trying to take pictures of fuel distribution in an engine illuminated by a laser, you'll have a lot of noise, but you can just downsample/etc. to get rid of the noise.

I'm not implying anything; if you like the 4/3rds system or the Nikon system or the whatever, I'm happy that you're happy.

[johnmflores]

Uh, the one from Panasonic....I can see luminance noise even when viewing from distance. The one from K-5? not that obvious.

Uh, I didn't say identical. I said closer than some might think. Consider for a moment that those are ISO6400. That's rare air that M43 is not supposed to be in. But it puts up a pretty good fight.

Again, I'm not suggesting the M43 is better or even the equal - when the lights go down I'm reaching for my K-5. If I happen to have my GH2 instead, I can often get the shot, but the colors are often wonky and the DNG files brittle in post. I submit these samples becauset I get the sense that there are some here who have not spent any real time with M43 and derive their opinions solely from DxO.

[Laurentiu Cristofor]

I'm not implying anything;

By implying, I just meant that I didn't understand the implication of your pixel-size normalization statement over the decision of whether to buy a 645D or not.

'What I mean' is that the SNR for that test is really only 'valid' for a given megapixel size, and comparing across various sensor and pixel sizes during 'actual use' with post-processing is not as straightforward as 'one number' might suggest.

I don't think that the dxomark low light ISO score is dependent on megapixels. Denser sensors may produce different "grain", but I don't think such differences impact the SNR, because it's a ratio to begin with - you cannot eliminate noise without eliminating signal as well.