All About Pixels

[dennismv]

A couple of questions:

1.  Does anyone know how large the pixels are on Hasselblad H3DII 22 and 39 ?

The sensor size is the same, yet megapixel count is different.  I wonder thus if 22MP sensor's pixels are larger.  If they are larger, common sense tells me that colors will be better defined and there will be less noise.

2.  Are Hasselblad's pixels any different from current Nikon/Canon dSLRs ?

Does anyone know if each pixel is the same -- receives all color wavelengths or if there are Green, Red, Blue pixels, like in the other cameras.  In video cameras for example there are 3CCD -- 3 sensors where each captures R G B colors.  In digital cameras there is usually one sensor, so I wonder how that sensor is set up on Hasselblad.  Hasselblad says that it gets 16 bit color samples, which also interests me in what's going on in the sensor.

[BernardLanguillier]

Yes, the pixels of the Kodak sensor used in the 39MP version are smaller than those of the 22MP version. This is, by the way, the same sensor used by PhaseOne in their backs.

Both are in fact pretty bad at anything higher than ISO 200, 400 being the max.

These are Bayer sensors just like those of the DSLRs, but they lack an anti-aliasing filter, which can result in some artifacts, but does also increase the sharpness per pixel to some extend.

Regards,
Bernard

[dennismv]

thanks.

I'm surprised that such large pixels are bad at higher ISOs.  Is there a technical reason why ?  To contrast with Canon/Nikon sensors that go to higher ISOs, it's interesting to see that larger sensors in MF are worse...

[John Sheehy]

thanks.

I'm surprised that such large pixels are bad at higher ISOs.  Is there a technical reason why ?  To contrast with Canon/Nikon sensors that go to higher ISOs, it's interesting to see that larger sensors in MF are worse...
[a href=\"index.php?act=findpost&pid=182311\"][{POST_SNAPBACK}][/a]

The inverse relationship between pixel size and noise is partly myth and illusion.

First, there is no direct relationship between pixel noise and image noise.  There is another factor, and that is displayed viewing angle of the image's pixels, and even the brightness of the display.  Lots of pixels can make up for inferior pixels in an image.  Also, sensors with large pixel pitches don't always have big pixels; there can be lots of dead space between pixels.  The more dead space, the more photon shot (counting) noise there will be at any given ISO.

The difference between Canon DSLR and Nikon D3 high ISO noise and high ISO noise in most MFDBs is that the latter have no optimization for high ISO; their high ISOs are nothing more than the base ISO, under-exposed.  With current technology, there is a certain flat amount of noise added to the signal when reading out the sensor, which exists regardless of how low the gain is, so the amount of absolute noise, relative to signal, added by readout is greater the less you amplify the signal.  Being that all ISOs are done with the same actual electronic gain at the photosites, the high ISOs have much more noise than if they were achieved by using more gain at the photosites (the way Canons and the D3 work).