[MJSfoto1956]

With the K-3 you would end up with 60 megapixel images...............who needs 645Z then?

There is still a need for a true, medium format, 50mp camera. But that is an interesting point. I can see the product manager for the 645z nixing any idea of a K-3 delivering "more" pixels than their camera!

M

[rawr]

It would be really SWEET if Pentax would offer this in its cameras

I agree. It's another creative way to use sensor shift shake-reduction, just as Pentax have done already for diffraction correction, astro-tracing and composition adjustment. I'm sure Pentax could make it work.

The idea has been discussed in these forums before, by the way. There are several threads about it, as well as the associated concept of super-resolution.

[hjoseph7]

I wonder how long it's going to take from the moment you press the shutter to the moment the image displays on the playback ? I'm assuming this is going to take some time(and battery life) considering in-camera HDR. In any case Olympus was always a leader when it came to innovation.

[Digitalis]

It is done with a precision shift of only micrometers to allow each physical sensor site to have multiple inputs to the final image. This would be filling in the blanks, so to speak. Image circle should not be an issue. What is intriguing to me is the part of about aligning the R, G, and B sites at the same point to essentially emulate a Foveon.

That wouldn't increase linear resolution, it would just increase the amount of colour information at each pixel. Olympus is basically making the same stupid claim that Sigma has been making for years with their foveon sensors: where they multiply the base resolution three times in order to advertize that their cameras provide 40Mp of resolution, when in linear metric it is only 14mp. Shifting the microlenses would also introduce colour artefacting with moving subjects.so this is a very niche technology - like hasselblad multishot digital backs, I only know three photographers here in Australia that use these.

[stevebrot]

That wouldn't increase linear resolution, it would just increase the amount of colour information at each pixel.

Except that they are reportedly doing both "filling in the space between" and aligning the photo sites. The combination should allow for very high S/N ratio as well as superior resolution and color fidelity.

Shifting the microlenses would also introduce colour artefacting with moving subjects.so this is a very niche technology - like hasselblad multishot digital backs

Exactly. The linked article at PetaPixel made that exact comparison and the comments above (mine included) also mentioned that it would not work well with moving subjects or IBIS.


Steve

[Digitalis]

The combination should allow for very high S/N ratio as well as superior resolution and color fidelity.

Foveon made the same promises too...look how that turned out: Delta-E goes out the window as soon as ISOs start climbing. I can imagine that using an optically stabilzed lens could make this technology useful for hand held use in the future.

[maltfalc]

it can't be done in-camera with pentax, but if you shoot a quick series of shots handheld you'll get enough random shifting that you can combine them in photoshop or other software to get the same effect. using a tripod and composition adjust rather than shooting handheld should be even more effective, though it's a bit more work. google "super resolution".

[Rondec]

It does seem like something Pentax could do in-camera. I wouldn't use it a whole lot, but it certainly could bump resolution/quality of landscapes on still days.

[MJSfoto1956]

That wouldn't increase linear resolution, it would just increase the amount of colour information at each pixel

I don't think you understand how a Bayer array works. The fact is, all derived "pixels" from the "raw" information supplied by a Bayer array are interpolations -- this technology does in fact increase measurable resolution (MTF). We were experimenting with this back in the 1990s and it works.

Michael

[Nicolas06]

I don't think you understand how a Bayer array works. The fact is, all derived "pixels" from the "raw" information supplied by a Bayer array are interpolations -- this technology does in fact increase measurable resolution (MTF). We were experimenting with this back in the 1990s and it works.

Michael

Reality is our 16MP sensor are just 4MP full color sensor. Interpolation give the 16MP. But this is half a marketing claim. Technically your full HD monitor is 2MP full pixels but is able to display 6MP, 1/3 red, 1/3 blue, 1/3 green. This is used by computers to increase actual resolution of text for example. The computer know your model of screen, how the red/green/blue sub pixel are spread and draw the shape of fonts more accurately by only using the sub pixel that really matches the global shape.

So you sensor is using interpolation to try to simulate 16MP full color matrix. But this is by far not of same quality as real full 16MP image. You can get a sense of that by getting say a 645Z or D800 shoot, resize it to 16MP and look at it at 100%. Your results will be better with 645Z or D800 even through you kept final size to 16MP.

So our 16MP is say something is between 4MP and 16MP... Maybe 8MP or something like that. Difficult to say because it also depend of the patterns it need to reproduce. For some use case the interpolation fail short (like when you have moire) in other cases it work quite well.

So to go back to getting more resolution, I'am affraid by moving full pixel on a 16MP sensor that you'll just get full color 16MP shoot. That will be already quite better but not the max you can achieve.

If you move not full pixel size but something different like 1/2 pixel size and 4 shoots, you can somehow get an interpolated 64MP shoot. They may not be worth really 96MP but for sure they give you more than you current 16MP sensor... Maybe get you at same level as current 645Z or something. Not that bad.

And as for 645Z, nothing prevent it to interpolate to 200MP this way. It has already been done by other camera maker.


Biggest issue for me: it really work only if you use a tripod.

[zekewhipper]

From what I read about how the system would capture all the colors, I would say just install a Foveon sensor and be done with it.

[Nicolas06]

From what I read about how the system would capture all the colors, I would say just install a Foveon sensor and be done with it.

This work pretty well at iso 100... Foveon sensor tend to give you the better quality than sensor using bayer by a huge margin at iso 100. The problem apparently is when for when you need more than iso 400... The foveon loose its adventage.

Second problem is the whole industry is working with bayer. It mean all derawtisers algorithms, optimsations effort are put into that. It mean that most of the R&D effort is put into making better bayer sensor... not foveon.

Even if you design is really interresting, you typically need money to refine it and money to full use it. Foveon is alone and there no way to leverage others work while for bayer one can just buy latest Sony sensor on one side and use latest DxO/Lightroom derawtisers algorithms to get the best results. Work is splitted among many.

[Rondec]

The thing about Foveon sensors is that people develop them with Sigma's own software and from what I can see, its default settings have sharpness/clarity and saturation bumped way up -- looks unnatural to me. Not sure if you can drop those levels down or not.

[akshay869]

If higher resolution is all one is after, they can try median stacking in photoshop. It yields really good results for photos taken at high iso. Also, there is a software that does the work.

Take better photos, reduce noise, increase resolution of digital photos. Leading superresolution software.

[glasbak]

Old technology, Ricoh already made a camera with sensor shift RICOH RDC-7.