[Brightcolours]

Bayer sensors get detail (luminance) from all 3 CFA colours, so the difference is not that big. How did you come up with the 27-30mp figures?

[JJ_SO]

Now I tried to open the same files here in the office. "Fluently", I'd say. 25 sec to open the same file which needs 90 sec on an iMac. Editing is no superfast, but still snappy.

 

 don't get it.   What's the point in treating customers third class?

 

So, for Windows users, no problem. The software is not the quickest converter available, but acceptable. I just will not buy into Windows, just becasue Sigma is too ignorant.

[Klaus]

Bayer sensors get detail (luminance) from all 3 CFA colours, so the difference is not that big. How did you come up with the 27-30mp figures?

 

Well, there's a bit of debate about the effective resolution from a bayer sensor. 

It also depends on the specific sensor. Most work with 4x4 layouts, but Fuji has 6x6.

 

Honestly I don't dare to make a statement myself but I have seen figures around the 67-75% mark of the nominal value (ok, thus 26-29mp to match the 19mp Foveon).

 

29mp is also the figure from Sigma for their interpolated images (only god knows why they do that because the RAWs are 19.6mp on the APS-C model).

[Klaus]

Some samples. BTW:

http://dc.watch.impress.co.jp/docs/revie...09844.html

[Klaus]

I like this one:

http://dc.watch.impress.co.jp/img/dcw/do...0.jpg.html

[Klaus]

The problem with both Quattros is ... what standard zoom lenses do you use ?

 

The 18-35mm on APS-C is great but doesn't really cut it in the real life (2x ratio. heavy) so that leaves the 17-50mm or 17-70mm. The 17-50mm is dated.

 

The 24-70/105mm on APS-H would be a 31-91/136mm which doesn't thrill me at least.

[JJ_SO]

For the APS-C model Sigma claims this figures:


 

Effective Pixels: Approx. 29.5MP

<span style="color:rgb(70,70,70);text-align:center;">T(Top): 5,440×3,616 / M(Middle): 2,720×1,808 / B(Bottom): 2,720×1,808
<span style="color:rgb(70,70,70);text-align:center;">Total Pixels: Approx. 33.2MP</span></span>

 

So, by saying 19.2 MP, you're counting only the blue layer and ignoring the other layers? Anyway, who cares? It is the closest approach to films like Kodachrome, which also had 3 layers. The rendering is special, the sharpness usually better than the Bayer result on a print. That's what I see.


 

Btw. I don't think, this system is a typical "standard zoom"-application as they bundle it with 30/1.4 Art prime. But in this little reportage (
http://www.sigma-global.com/en/cameras/s...mpression/) the 17-70 is not the worst lens I've seen so far.

 

There are many many other Bayer sensor on the market, nobody has to complain about the Sigma way.

[Klaus]

19.6mp are the "true" pixels with one pixel having the RGB information.

Your screen has also 1920x1080 pixel (or whatever) - thus it doesn't count the RGB sub-pixels either.

 

I don't complain about Sigma - I just don't understand why they upsample to 29mp. Their 29mp image simply doesn't contain more information than the 19.6mp image. Thus while they have the advantage of NOT having the Bayer-interpolation in the first place. they still try to mimic purely it for marketing reasons.

[JJ_SO]

The resolution is not increasing by their marketing maths, that's true. I just keep in mind that the other two layers, while they are not adding more resolution, contribute a lot to color rendering.

 

And the guy who wrote the little article showed some very good examples, like the red of the motorbike and the turquoise / green / blue of the water surfaces. Also, "The new SFD function is designed to reduce noise and extend dynamic range by taking seven bracket shots." interests me. While usually HDR appear unreal to me, I can see the quality improvement in the sample.

[Brightcolours]

Well, there's a bit of debate about the effective resolution from a bayer sensor. 

It also depends on the specific sensor. Most work with 4x4 layouts, but Fuji has 6x6.

 

Honestly I don't dare to make a statement myself but I have seen figures around the 67-75% mark of the nominal value (ok, thus 26-29mp to match the 19mp Foveon).

 

29mp is also the figure from Sigma for their interpolated images (only god knows why they do that because the RAWs are 19.6mp on the APS-C model).

Bayer only loses you a few percentages, actually (around 10-12% if I remember correctly). It does matter which kind of demosaicing is used, that is why some RAW converters give more detailed images than others.

And Sigma just counting very pixel 3 times with the older foveon and now a bit less with the quattro nonsense is just childlike silly, to be honest. It really does have 19mp.

 

The reason Sigma foveon looks so sharp is mostly down to no AA-filter (which then just just mostly fake extra sharpness). And yes, with saturated to dark primary colours, they have a resolution advantage with blue. Not anymore with green or red.

 

They capture blue detail for every pixel. They capture less green detail (half of it) compared to a bayer CFA. They capture the same red detail as bayer CFA