Nikon D4s: Bayer processing breakthru?

Started Jun 4, 2014 | Discussions thread
richard stone Senior Member • Posts: 2,645
Re: Thanks to all: Nikon D4s: Bayer processing breakthru?

Tom Schum wrote:

Many thanks to all for your input and opinions.

It seems to me that we all know how bayer has been continually optimized over the years and we are left with only about 70% of the pixel count in terms of resolution. A breakthrough would go beyond this, and nothing was said about a breakthrough in the Shutterbug review.

So for now I might call it a subjective reading of D4s resolution byt the Shutterbug staff, as suggested.

DPR lab tests carry a lot of weight, at least for me, so I'll be waiting for them.

Here are a few opinions of my own. I use a CFA camera (Fuji X-E1) as well as my Sigma DP3M. There is a different look to the images from each, and it might only be the style these days but CFA is a little more gentle in the images. Foveon will definitely wow you but the comfort level with CFA seems to be a little bit higher for me because it is not reaching out to grab your attention so much. There is a place for both in the photographic art.

I've seen posts about trying to make Bayer look like Foveon, but nary a one about making Foveon look like Bayer. The crisp softness of Bayer and other CFA images is not easily produced from Foveon images I think, because in post processing one might need the complementary function to sharpness enhancement. I don't know if such a function is available.

It seems to me that the Foveon mantra of "as good as a bayer with twice as many pixels" has not been quite on the mark lately. It was true several years ago, but we have seen some advancements in Bayer/CFA processing that seem to erode the clear advantages of Foveon. Now, with the D4s, we might see parity between the images from the Nikon full-frame sensor at low ISO and the Merrill at low ISO. One reason we might not have seen such comparison images in the recent past is that no D4s owner or reviewer is particularly interested in exploring it. DPR's method of doing test images seems to involve careful sharpening until the limit, for any given sensor, and this might show us something interesting when compared to Merrill. Or, maybe not. I'm guessing in a couple months we will have some DPR test shots from the D4s, and then this will all blow over (if it is any kind of storm at all).

The issue (as I see it) is that the Foveon compared to Bayer has better per pixel sharpness/resolution, so the edge roll-off is more defined. And dramatic. Combine that with the lack of an AA filter and the sense/sensation/appearance of detail is much greater in a Foveon image.

But is sense of detail the same thing as detail? My SD10 had a superb sense of detail in the image, but I doubt that it would match up with the actual detail from a Merrill sensor. Still, the SD10 images are superb and crisp.

A second aspect is that the Bayer sensors are getting better, as is the processing. But it cannot, at a pixel level, match the Foveon direct and complete measurement at every surface pixel, and with no interpolation. Clearly that does not matter for every image. And the energy required to get that information at every Foveon pixel has led directly to the high ISO noise issues. The information at the level of the small lower layer Merrill pixels gets vanishingly small and difficult to use.

My reading of the information from Sigma and other sources about the Foveon quattro leads me to think that the new Sigma sensor will do better than ever at detail and produce very crisp images with more detail than before. Colors and shadows should be better. With smaller file sizes.

But I think in the end many will still prefer the Bayer sensor images, and they will be improving too. One issue that to me is telling is the usual report that the Bayer sensor images need more sharpening, and that it is (unfortunately) easy to over-sharpen the Foveon images.

I am glad that Sigma is being properly compared with Nikon and Canon et al, and with Pentax MF. It's impressive, really. And it is entirely possible (probable, really...) that better Foveon "quattro" type sensors can be produced than this initial version.


 richard stone's gear list:richard stone's gear list
Sigma SD10 Sigma sd Quattro Sigma 17-50mm F2.8 EX DC OS HSM Sigma 30mm F1.4 DC HSM Art
Post (hide subjects) Posted by
(unknown member)
(unknown member)
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow