24MP - where's the resolution?

[Jorgen Udvang]

I shouldn't do this stuff, but now at least I don't have to worry about upgrading to more megapixels anytime soon. I saw the SLT77 samples were up at dpr, so I thought I'd check how much more resolution I'd get if I bought that one instead of the GH2 that's on the urgent part of my shopping list at the moment. For comparison's sake (and because dpr's "comparometer" allows for four cameras), I included two other cameras that I've been considering.

To follow the trend for higher ISO at any cost, I checked the ISO 800 files. The below crops are all 100% of images shot in RAW and enlarged by me (Bicubic in PS CS4) to 24MP. No sharpening or other adjustments except that I increased exposure of the GH2 sample by 1/3 of a stop to get it in line with the others. I downloaded the jpegs from dpr, not the RAWs. The GH2 and E-5 files appear a little larger due to the different aspect ratio.

SLT77:

GH2:

D7000:

E-5:

Nah... might as well include a high-end camera as well, they measly 12MP D3s:

 

[Terry]

What I didn't see on the Sony is what lens they used for this. The Panny uses the super sharp 50 macro. Did you see what they shot it with?

 

[Jorgen Udvang]

What I didn't see on the Sony is what lens they used for this. The Panny uses the super sharp 50 macro. Did you see what they shot it with?

They used the 50mm f/1.4 at f/9. The Zuiko is superior, but lenses are part of the deal, and Sony doesn't have much to offer in that area, unless you want big and bulky. Saving a few hundred grams on the camera isn't much use if the lenses weigh twice as much as the body.

Another thing that surprises me is that the GH2 shows more or less the same level of noise as the Sony, even when blown up to 150% of its original pixels. It's interesting also that the D7000 shows much less noise when blown up to SLT77 size, but not much less detail.

 

[ustein]

>They used the 50mm f/1.4 at f/9.

I would think this is with this sensor already diffraction limited.

 

[Jorgen Udvang]

>They used the 50mm f/1.4 at f/9.

I would think this is with this sensor already diffraction limited.

In that case, they have a problem, and may have moved beyond a sensible resolution for APS-C sensor. Still, it's mysterious to me that the GH2, which has a similar pixel pitch, is doing so much better.

But wait... if diffraction is pixel pitch related, the sensor size has little or nothing to say. In that case, dpr's testing method is flawed, since they choose aperture size according to sensor size (f/6.3 for the Panasonic).

 

[jonoslack]

Interesting stuff Jorgen - but if you're comparing jpgs, then how relevant is it when we all shoot RAW? (especially with this 'angels on the heads of pins' type of comparison).

I'm not even going to think seriously about the A77's sensor performance until I've seen RAW conversions done by me in Aperture with up to date firmware.

Like Uwe I'd question the Sony 50 f1.4 at f9 - but I do agree with you about the lens issue with Sony. On the other hand, if they have the lenses you want (16-50, 70-300 and the 135 f1.8 in my case) then that's good enough. . . . as long as the 16-50 is good at 16 and f5.6 I'll be sorted.

 

[Paratom]

Why comparing at ISO 800?
We all know the d3s is the high ISO machine.
Is ISO 800 what you use most of the time?
So why not compare at ISO 100 or 200?

I still dont understand why high ISO seems to be so important to many (I see it important for some applications like indoor sports or maybe events) but IMO 80% of what most people do and as long as you have good glass, whats the reason to go up in ISO so often?
When I shot film I allways thought ISO 400 is high ISO.

 

[Terry]

Someone posted the higher ISO shots but using Capture 1. To my eye (with a very quick scan), there is a pretty big difference.

http://forums.dpreview.com/forums/read.asp?forum=1037&message=39453211

 

[Jorgen Udvang]

Why comparing at ISO 800?
We all know the d3s is the high ISO machine.
Is ISO 800 what you use most of the time?
So why not compare at ISO 100 or 200?

I still dont understand why high ISO seems to be so important to many (I see it important for some applications like indoor sports or maybe events) but IMO 80% of what most people do and as long as you have good glass, whats the reason to go up in ISO so often?
When I shot film I allways thought ISO 400 is high ISO.

Because, if I posted the ISO 100 images, many would have said "We all know that 4/3 sensors perform well at low ISO, but at higher ISO...". The difference was similar at low ISO btw., which is much more relevant to me, since I rarely go over 400.

 

[Jorgen Udvang]

Interesting stuff Jorgen - but if you're comparing jpgs, then how relevant is it when we all shoot RAW? (especially with this 'angels on the heads of pins' type of comparison).

I'm not even going to think seriously about the A77's sensor performance until I've seen RAW conversions done by me in Aperture with up to date firmware.

Like Uwe I'd question the Sony 50 f1.4 at f9 - but I do agree with you about the lens issue with Sony. On the other hand, if they have the lenses you want (16-50, 70-300 and the 135 f1.8 in my case) then that's good enough. . . . as long as the 16-50 is good at 16 and f5.6 I'll be sorted.

RAW files don't show up too well in the forums, so there was no way around the jpegs. Hopefully, other Sony lenses are more up to it than the 50mm, but don't take it for granted. The A77 has the same number of pixels as your A900, but in half the area. Another side of it is that we don't really know what is due to lens, what comes from the sensor and what can be improved in RAW conversion yet.

Actually, I would be more worried for the NEX 7. There's not a great selection of high quality native lenses available for that camera, and many buyers, at least those pixel peeping, might become disappointed with the results they get.

 

[Terry]

It will be interesting to see how e-mount lenses do. I have the Contax G 35mm and 45mm which should be up to the task.

 

[kit laughlin]

Jorgen, the point you make about sensor–lens combination is crucial, I feel. I am amazed by how close these images are, frankly. By this I mean that any of these tools can make nice images.

And this is forgetting Raw, for now.

The GH2 seems to punch well above its weight, and its lenses are decent. Glad I have one! Thanks, KL

 

[Jan Brittenson]

Diffraction is a function of the opening size, or 50/9mm. This is more than 16/5.6, so will have less diffraction. In addition, in a photographic lens the aperture is completely out of focus and any effect is spread uniformly across the image. There will be no diffraction rings or any other discernible pattern, all you get is a loss of contrast. And if you have 12 stops of dynamic range you could lose A LOT of contrast with a digital capture device. As a result you will NEVER just see detail 'disappear'. What you will see is veiling flare, because light at any point in the image will have passed within one half wavelength of any point along the aperture.

The 'diffraction limit' is only relevant when discussing entry/exit pupil size. I seriously doubt the Sony 50/1.4 comes anywhere near its theoretical diffraction limit at any aperture. More likely it's simply unable to resolve the pixel pitch at a contrast high enough to overcome the AA filter...

 

[jonoslack]

There are many different ways of testing lenses and cameras.
Shooting the A77 with the 50 f1.4 at f9 and no Adobe support is an interesting variant . . . . possibly not one to write of the camera though!

The DxO report is pretty favourable

E-P3: 51
E5: 56
A77: 78
Panasonic GH2: 60

DxO comparison


These are not small differences. You may not consider DXO to be the be-all and end-all of camera IQ (I don't either). . . but I don't consider the dpreview studio test to be so either! (even less so). . . apologies if you've read this in another thread.

It's not really supposed to be in support of the A77 (until I see the Aperture RAW conversion I won't make up my mind). Just intended as a reality check to those who are writing it off prematurely . . . .

 

[Jorgen Udvang]

But how can dpr use ACR for conversion if there's no support for the camera? I also wonder if Sony has been pulling the trigger to fast here. If the lacking resolution is due to a lens and/or conversion issue, it would have been smart of them to address those issues before releasing the camera, no? But if the lens is the culprit, they may simply have assumed that a 50mm lens with sufficient quality would be to expensive for the prospective buyers of the camera.

Are lenses involved in the DxO testing?

Edit:
Those DxO scores are largely irrelevant (as the dpr sample photos may also be, but at least those are photos):

- The differences between the A77 and the GH2 below ISO1600 are so small that they are irrelevant (a third of a stop higher ISO on the A77 makes more or less the same difference).

- For ISO sensitivity, the GH2 actually scores higher, but the Sony can shoot at ISO 50, which I assume gives it the higher total score.

- There's no measurement for resolution.

 

[Terry]

But how can dpr use ACR for conversion if there's no support for the camera? I also wonder if Sony has been pulling the trigger to fast here. If the lacking resolution is due to a lens and/or conversion issue, it would have been smart of them to address those issues before releasing the camera, no? But if the lens is the culprit, they may simply have assumed that a 50mm lens with sufficient quality would be to expensive for the prospective buyers of the camera.

Are lenses involved in the DxO testing?

DPR gets early pre-release beta copies of ACR.

 

[ustein]

>Shooting the A77 with the 50 f1.4 at f9

Don't you think that f/9 is high for a 24MP APS-C sensor?

 

[jonoslack]

DPR gets early pre-release beta copies of ACR.

Perhaps that, in itself, should ring a warning bell.

Look - I'm not saying that the 24mp sensor IS good - I'm saying that it's too early to tell whether it's good or not, and that the tests at dpr throw up as many questions as answers.

 

[jonoslack]

Point of focus in the dpreview shots:

It looks like the A77 shots have been focused further back than those for the other cameras - this means that you get dramatically different results depending on where you look on the image.

Here are a couple of screen shots looking at the playing card at the back.

100 ISO

800 ISO

3200 ISO

Suddenly the A77 looks much better.



all the best

 

[jonoslack]

Re: 24MP - where's the resolution? - Where's the point of focus!

Hi Jorgen
Okay Lets do this properly - I've also checked the SAME ISO 800 jpg files - my upsizing was not as careful as yours, but I used PS and made the height of each up to 4000 pixels.
(same as the A77).

All I've done is to pick a different part of the test image

Anyway, here are the results
SLT77:

GH2:

D7000:

E-5:

Nah... might as well include a high-end camera as well, they measly 12MP D3s:

I rest my case - these sort of results shouldn't allow us to jump to conclusions too quickly - especially when it leads to a general conclusion that a sensor is not okay (which is certainly the general conclusion around here)

I'm not suggesting that it's wonderful - just that jumping to conclusions is dangerous. . . .