[falconeye]

I was in my local photostore and the salesman told me a Pentax rep had visitid them recently. He had heard rumors about a new 6x7 from Pentax. The sales man also mentioned full frame but probably used wrong term for 6x7. Sounds all wrong, who would make a 6x7 sensor?

I think it is pretty clear what is behind this.

When launching the 645D, Pentax said straight away they plan for less cropped MF sensors down the road and lenses would remain MF. And this is what they still say and lenses confirm.

Digital 6x7 doesn't make sense for Pentax. And we would have heard about such a project. The problem is that unlike Pentax 645 lenses which have really outstanding absolute resolution (i.e. if measured in lp/mm), 6x7 lenses have good relative resolution but not so good absolute one. So, a digital 6x7 would probably require a new lens line which makes no sense for Pentax to do.

So yes, I think it means Pentax considers the 645D a success story and plans to continue on it. And that's about it.

[Cambo]

Mmmmh, somehow I don't think stitching sensors is possible, as there would be surely a small gap between sensors... It would have been done ages ago, otherwise... And FF bodies would be a lot cheaper!!!

I didn't mean stitch the SENSORS together, I meant stitch the resultant 4-part image together...



Could EASILY be done in the software with a processor powerful enough. Or even in software in the host computer; leave the images unstitched in the camera, or with a small proxy for viewing, then have the RAW software in the computer stitch the fullsize image together...

Hey that's brilliant, if I do say so myself. Above paragraph © 2011 by Cameron R. Hood.

And hey, Pentax, I'm available for consulting work; I have one good idea every few years...I'm not fast, but I am expensive!

I think it is pretty clear what is behind this.

When launching the 645D, Pentax said straight away they plan for less cropped MF sensors down the road and lenses would remain MF. And this is what they still say and lenses confirm.

Digital 6x7 doesn't make sense for Pentax. And we would have heard about such a project. The problem is that unlike Pentax 645 lenses which have really outstanding absolute resolution (i.e. if measured in lp/mm), 6x7 lenses have good relative resolution but not so good absolute one. So, a digital 6x7 would probably require a new lens line which makes no sense for Pentax to do.

So yes, I think it means Pentax considers the 645D a success story and plans to continue on it. And that's about it.

I think having great lenses becomes less of an issue now with lens correction software, where as long as the resolution is good, chromatic aberration, distortion and the like almost don't matter. So I wonder why all those landscape and fashion photographers made the 67 a standard for all those years if the lenses were only resolving 'relatively' rather than 'absolutely'...so their images were only relatively good...



Cheers,
Cameron

[Aristophanes]

Look to Phase One, Mamiya, Hasselblad, and/or Alpa for possible larger sensors up to 6x7.

Eventually.

[Caat]

Yes, I could see Pentax with two different 645s - an 'APS-C' and 'FF' version.

[thibs]

Agreed

645D Will become 'entry level' so to say.
Mk2 or mk3 will be full 645 frame or About (like 1.1 crop).

[falconeye]

I didn't mean stitch the SENSORS together, I meant stitch the resultant 4-part image together...
[...]
Hey that's brilliant, if I do say so myself.

Hi Cameron,

I must have missed an important part of the idea along the road.
Where do you get the missing pixels obscured by the inevitable sensor borders from?

This isn't pano stitching with overlapping images.

if the lenses were only resolving 'relatively' rather than 'absolutely'...so their images were only relatively good...

LW/PH is relative and lp/mm is absolute. A larger sensor typically boosts LW/PH but not necessarily lp/mm as the lenses can be hard to make for this.

The great detail about the 645D is that it has almost the lp/mm resolution of a K-5 (80% of it to be precise).

[ghelary]

Hi Cameron,

I must have missed an important part of the idea along the road.
Where do you get the missing pixels obscured by the inevitable sensor borders from?

This isn't pano stitching with overlapping images.


LW/PH is relative and lp/mm is absolute. A larger sensor typically boosts LW/PH but not necessarily lp/mm as the lenses can be hard to make for this.

The great detail about the 645D is that it has almost the lp/mm resolution of a K-5 (80% of it to be precise).

Well, such a camera exist, I can't recall exactly who did it, but a good friend of mine, described me that one on his relation had developped a digital large format based on 5DmkII sensors. I guess, to make the final image, you need a bit of content aware filling in photoshop, but I doubt that large format lenses can outresolve an array of 5DmkII sensors. So the missing pixels shouldn't be too difficult to fill.

However, such a custom made camera would be ridiculously expensive (I do not know how he got his sensors) and I'm not sure how this would compare with a high end MF or another custom made camera (let say based on a scanner technology, like the first digital cameras were made)

[wjjstu]

Mmmmh, somehow I don't think stitching sensors is possible, as there would be surely a small gap between sensors... It would have been done ages ago, otherwise... And FF bodies would be a lot cheaper!!!

I think it has in fact been done before.

[eurostar]

Well, such a camera exist, I can't recall exactly who did it, but a good friend of mine, described me that one on his relation had developped a digital large format based on 5DmkII sensors. I guess, to make the final image, you need a bit of content aware filling in photoshop, but I doubt that large format lenses can outresolve an array of 5DmkII sensors. So the missing pixels shouldn't be too difficult to fill.

How? The part of the image that fall over the gap between the image catching part of the sensors how is rebuilt?

[ghelary]

How? The part of the image that fall over the gap between the image catching part of the sensors how is rebuilt?

Well, I let you think about, I think I said photoshop.

Please imagine how thick is the junction zone compared to the overall size of the image

[falconeye]

I think it has in fact been done before.

Yes, stitching two dies together is normal procedure for some of the FF sensor makers. You need large enough reticles to avoid it and not everybody has them.

However, it isn't stitching after the fact. It is optical stitching where two different reticles are used to illuminate different portions of one die. The resulting die is one piece of silicon. It actually increases the cost.

[falconeye]

Well, such a camera exist, I can't recall exactly who did it, but a good friend of mine, described me that one on his relation had developped a digital large format based on 5DmkII sensors.

You can probably produce optics like a complex prism to bring up to, say four sensors into the focal plane without any gap.

This is why I asked you for the missing part of your idea.

However, filling the gaps in siftware isn't feasible.
Even if only very few pixels are inserted (or missing), the human's gorgeous pattern matching capability would see them.
(And a packaged sensor has a huge border)

[Smeggypants]

And make the camera powerful enough to stitch them together quickly. Can probably be done with a multi-processor camera.

Cheers,
Cameron

You'd need some nifty optics to split the image into 4 quadrants to account of the gap between sensors.

[eurostar]

Well, I let you think about, I think I said photoshop.

Please imagine how thick is the junction zone compared to the overall size of the image

Even if it a single pixel wide, there would be a row of missing pixel. And I am sure is multiple pixel wide.

[Tom S.]

I'm holding out for a 8x10 with a 140,382GP sensor.