Question on low light A580 vs A77 (resampled?)?

[bfitzgerald]

I think it's obvious to all that the 16mp CMOS is superior for low light work you can't ignore the real world shots and samples.
If that matters depends on what you shoot. Personally I do like low light and I could not say the 24mp results are acceptable for a modern camera.

Then there is the question of just how much resolution do you need. Well few need 24mp so that fairly easily seals the 16mp argument.
Losing half a stop (or not far off that) in low light is also not that funny which only enhances the problem further.

[agorabasta]

Barry, the a35 is the most recent 16Mp SLT, and it's just as noisy as a77.

And you'd better try proving things instead of declaring them 'obvious'.

[Greg Beetham]

For what it’s worth, DXO gives the 16MP sensor in the A35 a worse high ISO score than the A55 sensor and the latter has a better score than the A77. http://www.dxomark.com/index.php/Camera ... and3)/Sony
So where does this leave us with high ISO in the A77? Well for one thing you have 50% bigger files at high ISO for no material advantage over the 16MP sensor…but hey it gives that GPU something to do, dealing with 8million more bytes of noise, doesn’t it? Don’t want the dang thing dozing off now do we.
Greg

[agorabasta]

DxO measures the noise at 18% grey level. That level is dominated by the natural shot noise from the discreet nature of photon capture. So the DxO figures simply reflect the tiny differences in quantum efficiency between different modern sensors.

But the noise that really ruins the image is not the shot/photon noise. In fact, that shot/photon noise is exactly the thing that creates image.
The real problem is the read electronic noise that's completely foreign to image nature. And that electronic noise sits at levels much lower than 18% grey for all modern APS sensors. It really shows in the image at levels of 5% and lower. And that's the excessive noise that we see in the a77 and a35 images.

So you may forget DxO - it really tells nothing of use.

[bfitzgerald]

I was mostly referring to the 16mp CMOS that is used in the A580, D7000, K5 etc etc.
Those are the ones I've looked at v the 24mp one and I think the 16mp one it better.

But it would be interesting for someone to rip the SLT mirror out and see if that improves things, it probably does a little.

[agorabasta]

But it would be interesting for someone to rip the SLT mirror out and see if that improves things, it probably does a little.

I did as much, look here - http://www.photoclubalpha.com/forum/vie ... 0&start=15

[alphaomega]

I have followed this discussion with great interest, but I must admit to being confused over this noise issue. So potentially the A77 has more detail at low ISO and same noise as A580 or A35 when downsampled to 16Mp (at higher ISOs).
For a simple soul like me, who just wants to take good pictures and where my current top of the range A580 and NEX-5 satisfy my requirements, it seems to me that the obvious choice is a 16Mp sensor and that means NEX-5N as an improvement over my NEX-5 (Key points optional EVF and faster AF).
What is the point in having big files and downsampling to 16Mp if that size is adequate? It would seem to me that 24Mp APS-C is for those who have had to migrate to A900 to get the low ISO detail they crave for and now have the possibility of getting the same or better through an A77 or NEX-7. I like the idea that I can just bump up the ISO on my A580 and NEX-5 as required without worrying about noise. The more I read and think about it the more convinced I am that 16-18 Mp is the "right" sensor size for APS-C with current technology. I also think that perhaps a better distinction between A65 and A77 would have been to make one video orientated with lower heating and one still orientated with maximum sensor power supply and minimum noise. I would rather have a camera with max. still capability and fewer frames per charge. Good third party batteries are now so cheap and good.

[agorabasta]

OK, let me clarify how the 16Mp matters appear to me -

1. Noise:

a35 - noisy
NexC3 - noisy
a55 - OK, but not too great with noise
Nex5N - very good low-noise performance
a580 - the best due to cold sensor (not used for LV feed, normally)

2. Other image quality issues

a55; a35 - SLT mirror ghosting is an issue
a55; a580; a35 - excessive PF at wide apertures, likely due to reflections between IR filter and the sensor/microlenses surface, very high res due to lax LPF
NexC3 - very nicely balanced LPF, cleanest rendition of sharpest transitions, lowest moiree, lowest corner colour shifts with the wide rangefinder lenses
Nex5N - LPF too lax, colour moiree and other edge intermodulation distortions (though better than the former 14Mp Nexes)

Compared to the 24Mp sensor cams - the a35 is just as noisy as a77/65; the 5N noise is essentially the same as Nex7 at image level, but the resolution of the 7 is incomparably better due to near total absence of moiree and other edge distortions (same goes for the 77/65).