DP Quattro: Sigma official announcement

Started 5 months ago | Discussions thread
Raist3d
Forum ProPosts: 32,715Gear list
Like?
Re: Not exactly gone...
In reply to DMillier, 5 months ago

DMillier wrote:

Sorry, not sure I follow this. Let me speak aloud...

A simple monochrome sensor with say 20 million photosites arranged in a single layer will sample luminance only at 20 million locations and colour not at all. Result = true 20 MP mono image

A 20 million pixel Foveon X3 sensor would have 60 million photosites arrange in three layers such that it measures luminance and wavelength at 20 million spatial locations. Result = true 20MP full colour image.

A 20 million pixel Bayer sensor has 20 million photosites arranged in a single layer with filters over the photosites such that 10 million measure green channel, 5 million red channel and 5 million blue channel. None of these photosites are arranged at the same spatial location, so it is not possible to directly measure colour at a particular location. Instead the information about the wavelength and brightness of the light at each photosite has to be used to algorithmically compute the colour at 20 million spatial locations from information gathered from nearby locations. Effectively, spatial resolution is given up in order to work out colour. Result = < 20MP of full colour resolution.

The quattro chip has 20 million photosites arrange unequally in three layers 20 + 5 + 5. How are these photosites used and what does this mean for spatial and colour resolution compared to old Foveon X3 and Bayer CFA approaches?

The "blue layer" is full 20 million photo sites measuring full spatial location of a 20MP sensor.  Picture a monochrome sensor but with a blue filter (note: the layer is not really exactly blue but that's quipping in this discussion).

So on the blue layer ONLY, the new Foveon = older Foveon.

Now picture the previous foveon but now combine each group of 4 green and red photo sites into one.  So you have one bigger photo site in those two layers effectively reducing the spatial resolution.

So vs the Bayer sensor, the Bayer sensor never has all spatial photo site location's data in any color at all (green is the best case and for a 20 MP sensor it's 10 million as you said, while this FOvoen has true 20 MP blue).  The Bayer sensor has 5 million red and blue but intermixed along with the others, and "small" - i..e. as small as each other photo site while the new Foveon Quattro has a 4x the size area of each spatial location for each green and red.

So it's effectively a resolution 1/4th-ening but then it doesn't have to interpolate color like Bayer guessing both spatial location and color in between different data of color readings.

I think the best way to look at this is to the supplied Sigma diagram and the Bayer sensor diagram.

-- hide signature --
-- hide signature --

Raist3d/Ricardo (Photographer, software dev.)- I photograph black cats in coal mines at night...
“The further a society drifts from truth the more it will hate those who speak it.” - George Orwell

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow