Nex7 noise performance is simply fabulous

[agorabasta]

Below is a 100% crop of a ISO3200 raw sample from IR.

I think no other APS sensor ever delivered an image as clean and detailed.

[[SiC]]

Looks great indeed

[Lonnie Utah]

Looks like iso800 on my a900.

[redsim74]

Below is a 100% crop of a ISO3200 raw sample from IR.

Impressive.

[InTheSky]

I waiting for my Nex 7 to arrive in a few weeks, but I know it is not good as the Nex 5N or other APS-C camera when talk about noise at high ISO. On this picture, I think we are kind of distrurb for having look too much at those sample picute in the past years . What I see on this picture is that the NEX 7 deliver great detail due to the 24mp, and probably reduce to 16mp or less, the picture will be good as other APS-C sensor.

I will not say it looks like the ISO800 on A900, the sensor look a little bit better in noise, but not better in keeping detail at high iso. But I will do my own test probably. But even if the camera is better or not It will become for me my killer portable camera for stock photo and commercial use. When good control of the light condition is manage, I have been able to get great picture at ISO 6400 with my "old" A900.

Regards,

Frank

[agorabasta]

... but I know it is not good as the Nex 5N or other APS-C camera when talk about noise at high ISO.

Sorry, but no, you don't.

To me it's quite clear that the 7 is much better at noise/detail than the 5N. Having naturally stronger per-pixel photon noise, the 7 still has lower per-image-area read noise coupled with an effectively laxer LPF.

Means, if you need no more detail than that from the 5N, you just add more NR filtering and get the same detail at lower noise; or you get more detail at the same noise. And so you get a chance of choice.

[Greg Beetham]

Aborabasta the exif states that the exposure was 1/640sec at f8, that doesn’t seem to be a very low ambient light value even if it was taken at ISO3200; fooling around with my light meter I come up with around EV10.5 at ISO100, any thoughts?
Greg

[agorabasta]

Greg, it's not the average luminosity of the scene that matters here. It's the values in the darkest parts of the scene that are most affected by the noise, and their test scene contains some really dark areas, and those areas are clean.

[alphaomega]

I have been wondering for a while what to purchase (A65, A77, NEX-5N or NEX-7) or not to purchase at all. My objectives remain lesser weight/bulk for same or better image quality. Having read what's available I am sure all new Sony cameras are excellent in their own ways. My main concern about another NEX has been lack of SSS in body until a sufficient number of dedicated E mount lenses beome available.
I just uploaded early 21 Century Provia 100 images taken on my XD7 with Sigma 24-70 F2.8 scanned on my Dimage Elite 5400 II and they passed Alamy today. These will have been taken at F8/1/250s or thereabouts. I have more or less resolved that I will now purchase the NEX-7 next spring with the new Alpha to NEX adapter so I can use my Alpha lenses as well until a full complement of E mount lenses with inbuilt SSS are available. I have no doubt that the A77 is a fabulous camara, but for my purposes the NEX-7 would seem ideal and versatile with the new adapter.
I think the NEX-7 scores over NEX-5N. I am sold on the idea that the NEX-7 "extends" the reach of lenses due to the higher pixel count in addition to providing more room when correcting perspective distortion.
Hopefully the price range for NEX-7 will be more acceptable come the spring.

[artington]

Bit late in the day here but the opinion expressed by agorabasta in this post is at odds with one he made in a post I started a couple of months ago about a77 vs a65, where he compared IQ between the a77 and a700 in the following terms:

"Resolution is very different, for sure, but that's partly from focus diff. But there's a difference of at least 0.7EV in the noise levels; so even the V1 a700 was much better with noise than the early a77.

Such a bad noise performance is either intro'ed into pre-release a77 on purpose, or it's simply an abysmally bad sensor that Sony somehow managed to create after all their successes at 14-16Mp."

Now please be assured that I am making no criticism here. It may well be that the early a77 samples were flawed in some way. Certainly the colour chart comparison that agorabasta posted comparing the a700 and the a77 was conclusive and suggests that the new 24Mp APS-C sensor is simply too densly populated. But if tbat's the case why does it apparently do so well on the NEX-7?

Now I imagine there will be some benefit for the NEX-7 in not having a partially transparent mirror to contend with but I doubt it will make such a difference to its noise quotient.

So, where are we? Right now, on the basis of IQ observations such as these I am veering away from the Nex7 to the Nex5N with its smaller sensor, which is perfectly big enough for my printing needs, even if in truth I woud prefer to have a built-in EVF.

[agorabasta]

SLT mirror is absolutely no problem, look here - http://www.photoclubalpha.com/forum/vie ... 0&start=15 - it doesn't add any discernible noise.

The sensor is obviously dumbed down so that a77 never overheats. The sensor-based stabilisation prevents effective heatsink, so they had little choice. But they could allow a special high-quality mode for a rare low-noise shot now and then.

[artington]

SLT mirror is absolutely no problem, look here - http://www.photoclubalpha.com/forum/vie ... 0&start=15 - it doesn't add any discernible noise.

The sensor is obviously dumbed down so that a77 never overheats. The sensor-based stabilisation prevents effective heatsink, so they had little choice. But they could allow a special high-quality mode for a rare low-noise shot now and then.

Interesting. So you are suggesting that the sensor on the NEX-7 is a significantly cleaner variant of that on the A77/A65? My understanding from the marketing info / reviews suggest that the same sensor is used and, not being up with the technology, my reaction has been that a sensor is a sensor. I was not aware tbey coukd be tweaked. If you are right it's a big selling point for the NEX7, contrast focus and lack of in-camera IS notwithstanding.

DK - do you have any thoughts about this?

[agorabasta]

So you are suggesting that the sensor on the NEX-7 is a significantly cleaner variant of that on the A77/A65?

I suggest exactly the opposite. It's the very same sensor. It is just operated at different timings/voltages so as to lower its power consumption and heat generation.

After all, a35 is noisier than a55, they both are much noisier than a580 and much more so than just a mirror deals - all those cams share the same sensor. Then the C3 is much noisier than 5N, and they supposedly share the same updated version of that sensor.

[David Kilpatrick]

The NEX-7 has significantly less emphasis on speed and I would not be surprised to find the noise levels much better.

Anyway, I can now get into my NEX-7 raw files thanks to Adobe ACR 6.6 despite the fact they have no proper camera ID - they show up as MODEL-NAME where the data should have the NEX-7 info.

But I can at least process the raws from the prototype. I do not have a full range of ISO settings, or matching subjects, because I was using four different cameras and several lenses. I did manage to get one view on both the A77 prototype and the NEX-7 prototype, the A77 using the 16-50mm at 50mm (f/11, 1/1000th) and the NEX-7 using the 18-55mm at 55mm (1/1250th, f/10) which are clearly identical light levels and comparable settings.

First the A77, processed using ACR 6.6, Strong Tone Curve, black level 3 (default for all exposure apart from this). Sharpness 25, Radius 0.5, Detail 25, Masking 0; NR Luminance 25, Detail 50, Contrast 0, Colour 25, Colour Detail 50), 100% clip from one area of the frame left of centre:

a77-1650-50-1600-acr6p6.jpg (452.38 KiB) Viewed 6264 times

Second, the NEX-7 with everything identical to the above:

nex7-1855-55-1600-acr6p6.jpg (438.51 KiB) Viewed 6264 times

In both cases I also got JPEGs which are really quite horrible compared to the raw processed results. I did not go above 1600 on these camera for this test, I did push the NEX-5N all the way to 25600 and concluded that in everyday use I would also limit that to 1600. ACR does not perform miracleswith 12800 and 25600 NEX-5N files, if anything the in-camera JPEGs take the correct approach (a simplified clean-looking image with minimal textural detail).

David

[agorabasta]

David, the 'MODEL-NAME' tag means a very early firmware version. Even the ACR6.5 shows the Nex7 hame correctly for the more recent firmwares. And the 6.5 version could even process the Nex7 raw if the SonyModelID EXIF tag were modified to the a77's value of 287.