[ogl]

Some of DA and DA*lenses can't cover 23.5*23.5 mm sensor.
For example, DA14 can cover only 23.5*18.6 mm, DA*16-50 just a bit more.
Some zooms can cover APS-H, some can not.

Square sensor is only for FF lenses and several DA.

---

Last edited by ogl; 02-15-2009 at 12:33 PM.

[jeffkrol]

Nikon is over to the right of Panasonic...
Square sensor...hmmmmm

[GaryM]

The DA 70mm f/2.4 Limited, DA 40mm f/2.8 Limited, DA*200mm f/2.8, DA*300mm f/4.0 and DA 35mm f/2.8 Macro are full-frame lenses. However, some of the other DA lenses (mostly the shorter focal lengths) are known to produce image circles that are too small to be used with full-frame cameras.

Pentax Lens/Camera Compatibility - Mark Roberts Photography

I would need to see edge resolution performance tests before I'm ready to think of the DA40 or DA35 as full-frame compatable. And on this I'm skeptical, especially about the DA35.

[spentax]

I like the idea. I've more than once thought about buying a robot royale for just such a format.

[jay]

I would need to see edge resolution performance tests before I'm ready to think of the DA40 or DA35 as full-frame compatable.

I'm fairly certain the DA40 is the same optical design as the old M40, though I could be wrong?

Also, although it is quite clever to think Pentax would introduce a square sensor to be able to use the lens to its "fullest potential," on closer inspection, the rumor wasn't thought out very well.

Yes, it is true that lenses are completely symmetric along the optical axis; theoretically, every lens casts a perfect circle.

But, as some have pointed out, if you keep the current width the same, and increase the height, you will require a slightly larger image circle to cover the square frame. Mathematically, this is because the diagonal length of the APS-C frame is NOT the same as the diagonal length of a square frame of APS-C width.

If Pentax *were* to introduce a square frame of APS-C width, many lenses would show extreme vignetting at the four corners of the frame. Of course, these would mostly be the wide-angle zooms, where coverage is most difficult.

Compounding the problem for the wide-angle lenses is the petal-style lens hoods with which most are equipped. These are NOT symmetric devices; they reduce flare from the top of the lens at the expense of vignetting the top and bottom of the frame, and most would consider them fairly necessary for outdoors shooting.

Most people reading this board would have no problem remembering to take the lens hood off before they shoot square/verticals, but from a marketing / relations standpoint, I would be pulling my hair out if I have to prepare customer service representatives to answer the huge call volume of people wondering why their $2000 D-SLR makes all their lenses vignette.

That's why I highly doubt Pentax would mess around with that. They're bold -- I'll agree. But they're not stupid. And although the person who started the rumor was able to draw up several impressive-looking diagrams and offer details that add plausibility (haha, my favorite was the "Fresnel lens adapter" for the 540FGZ...), I don't think it was thought through enough.

[Clarkey]

For my money, I don't think that something so radical as square format will be the go for Pentax (unless of course we are talking the 645 replacement?).

I pick another recangular APS-C (although if it weren't for the cost, the APS-H would be great) with CMOS, and better use of AF (more cross type sensors, buffer? - better screw type/ SDM drive?) and metering area. Oh, and maybe 18mp.

If there is anything announced at all.

I'm just hoping for the K20d sensor in the K300d, along with battery and a lighter size.

And I'll be happy.

[cousinsane]

I am pretty sure the DA35/2.8 macro is not a full-frame lens. I've seen test shots with the lens mounted on a film body.

The DA 70mm f/2.4 Limited, DA 40mm f/2.8 Limited, DA*200mm f/2.8, DA*300mm f/4.0 and DA 35mm f/2.8 Macro are full-frame lenses. However, some of the other DA lenses (mostly the shorter focal lengths) are known to produce image circles that are too small to be used with full-frame cameras.

Pentax Lens/Camera Compatibility - Mark Roberts Photography

[jay]

I would love the irony if they didn't announce anything.

At all.

[Clarkey]

I would love the irony if they didn't announce anything.

At all.

Very possible. But maybe one lens just to keep us interested.

I wonder if prices are going to keep going down if they don't announce anything?

[jsherman999]

I would love the irony if they didn't announce anything.

At all.

I don't know if this quite matches the definition of irony, but it would be stupid.

I think we'll see some minor body upgrades and gadgetry. An announcement of limited scope but maximum hype.

Sadly, it's more likely that we'd see something like a white K20D with GPS than a FF or square-sensor camera.




.

[zoltan1983snapper]

I would need to see edge resolution performance tests before I'm ready to think of the DA40 or DA35 as full-frame compatable. And on this I'm skeptical, especially about the DA35.

The DA35 works on film without serious vignetting only at some apertures, and even then there's light fall-off in the corners. It does work, though.

[jeffkrol]

Yes if the square sensor was designed to fit into the APS-C image circle, you would lose image width with a square sensor.

But if the sensor was 23.5X23.5, then there would be no change to how the DA lenses would function if the sensor was digitally cropped to act like an APS-C sensor. If the camera was switched to portrait or landscape, the 1.5 crop factor would still apply.

If the camera was used in full square mode, then it would have a crop factor of 1.3 (43mm FF image circle / 33mm DFX image circle)

Maybe this is where the APS-H rumor came from. Both a square sensor and the APS-H sensor have a 1.3 crop factor.

It would be a unique move for Pentax. Certainly define them in the APS-C market. It would also require a new mirror box, that would be a step toward a FF mirror box.

Personally, I would rather see an APS-H sensor used. I like the 3:2 ratio better then the 1:1.

Even if it is a BS rumor, I still think it is an interesting idea. Although, in a way it smells a little like the PZ-1P, and its panorama mode.

Sensor would have to be 21X21mm or per wafer yields would drop too low and be too expensive.....
VLSI chip yield demonstration
APS-H is worse.....

[jay]

Sensor would have to be 21X21mm or per wafer yields would drop too low and be too expensive.....
VLSI chip yield demonstration
APS-H is worse.....

Even more reason why this rumor wasn't thought out very thoroughly.

Not only because the chips don't fit very well on the wafer, but also, the larger the chip, the higher number of bad chips (remember that fabrication defects are obviously randomized over the entire area of the chip -- so the larger the area, the higher probability there will be a defect in a certain chip)

Great post!

EDIT: However, it should be noted that Samsung can currently fab on 300mm wafers, and has developed 450mm wafer technology -- so these numbers may not be meaningful, depending what their fab process is. That applet is based off 150mm wafer size, I believe

If you run the numbers for a 23.5x23.5 chip on a 300mm process, you can produce 108 chips per board, with a 90+ percent yield. Not bad. APS-C is 164 chips with a 95% yield.

It would definitely be more costly, though.

[falconeye]

EDIT: However, it should be noted that Samsung can currently fab on 300mm wafers, and has developed 450mm wafer technology -- so these numbers may not be meaningful, depending what their fab process is. That applet is based off 150mm wafer size, I believe

jay, you stated to study semiconductor engineering.
The applet is ok, as we can scale all input values such that a 300mm wafer rather than a 150mm wafer is matched.
Also, we know and already have discussed the yield formulas which must be applied.
Still, we are left with one factor of uncertainty. Maybe, you have a guess for us here...

What defect density would you expect, given the process and given the fact that most defects kill a pixel, not the sensor? Or equivalently, what is the current APS-C yield rate in the industry?

[jay]

jay, you stated to study semiconductor engineering.
The applet is ok, as we can scale all input values such that a 300mm wafer rather than a 150mm wafer is matched.
Also, we know and already have discussed the yield formulas which must be applied.
Still, we are left with one factor of uncertainty. Maybe, you have a guess for us here...

What defect density would you expect, given the process and given the fact that most defects kill a pixel, not the sensor? Or equivalently, what is the current APS-C yield rate in the industry?

0.08 - 0.2 defects / cm^2 is a rough estimate for common chip manufactures, and what we usually run calculations with in my classes -- but as an undergrad, we haven't focused a lot on specific processes and variations in it, and I'm not sure how much that number has changed since our course text was written.

If it were 0.08 defects / cm^2, we're looking at roughly a 32% chance a particular chip is going to have "a defect" (whatever defect that may be)

Some defects may cause hot/dead pixels -- some may cause other problems. And I'm not familiar with their policies on hot/dead pixels. I'm assuming only a small percentage of those defects would cause more than a dead pixel, so yield is probably fairly high compared to logic circuitry (where you simply can't have defects).

I'll look into this and get back to you with some numbers comparing the various chip sizes -- including their yield per wafer (after factoring in defects). I'll try to run it with a few different tolerances on dead pixels.

Haha, if Pentax won't tell us what they're going to do, we'll just do the research ourselves.

We should all get together and design a 50 MP full-frame sensor with back-illumination and fluorite microlenses and set up a 450mm wafer size fab process and just make our own damn cameras :-)

We could pull it off. Right?

RIGHT?