MF CMOS Sensor and Technical Cameras

<p>At PDN Expo today, a seemingly very knowledgeable rep of one of the companies made an interesting point. Medium format CMOS sensors, he said, particularly full-format MF Sensors (if they are developed) will run into problems at the edges of images taken on a technical camera with wide angle large format lenses. The problem, he says, is the angle of light, which needs to be straighter on a CMOS sensor than a CCD sensor. Interesting point, and an important one, it seems to me, inasmuch as wide angle is, I imagine, a frequent use for MF digital (it is for me in 4x5 film, e.g.) and the progress of technology suggests that once CMOS ramps up it will replace CCD entirely. If so, does this mean a limitation on the use of technical cameras and large format lenses? Will technical cameras be replaced by perhaps better fixed body ILC cameras, maybe Phase One's oft teased new camera? (I say ILC and not DSLR because with CMOS the mirror too will become a thing of the past, I bet, at least once EVF technology advances far enough.) Just curious what those on this forum think. I'm speculating on less information than you all have. Thanks in advance.</p>

<p>If there is sufficient demand then new lenses will be developed which will solve this problem. This is what happened with smaller format digital cameras. </p>

<blockquote>

<p>If there is sufficient demand then new lenses will be developed which will solve this problem. This is what happened with smaller format digital cameras.</p>

</blockquote>

<p>Thanks. I can see how the sensor might be developed to solve the problem, but the lens? I wide angle lens inherently sends light in at wide angles, doesn't it?</p>

<p>LF wide angles are the same as RF lenses in the smaller formats construction wise. - Nobody bothered about rear elements almost kissing the dark slide or FP shutter and they deliver great results on film.<br>

Digital sensors seem to benefit from retro focus constructions like Zeiss Otus (retrofocus standard lens) or the widest Canon tilt shift lens. <br>

I can not understand / Imagine why there would be a huge issue with blowing up Leica's current concept for their RF CCDs & CMOSes with decentered micro lenses in the format corner to 6x7cm FMF size once such sensors can be made at all.<br>

I do see a future for technical cameras in the digital MF realm. Good WA lenses for them might not exist yet but if you want some movements why not use a wide made for a big MF SLR? what was made for 6x9 or 6x8 should grant a bit of movements on 6x7... Just avoid genuine straight wide angles like Jupiter 12 or Super Angulons.</p>

Lobalobo, which dealer or manufacturer? If you go back tomorrow ask at Capture Integration (and say hello to Dave

Gallagher for me.

<blockquote>

<p>Decentered micro lenses in the format corner to 6x7cm FMF size once such sensors can be made.</p>

</blockquote>

<p>Exactly the suggestion the rep made, though they'd need to do it, which they might not if they figure that few buyers would use the backs on technical cameras where, I take it, the physics alleviate the issue.</p>

<blockquote>

<p>"Medium format CMOS sensors, he said, particularly full-format MF Sensors (if they are developed) will run into problems at the edges of images taken on a technical camera with wide angle large format lenses."</p>

</blockquote>

<p>- Yeah, right. That was Leica's excuse for not putting a full-frame sensor in their M8 in 2007. Lo and behold, just two years later in 2009, up pops the M9 with a full-frame sensor and no mention or complaint about poor image quality in the corner of the frame.</p>

<p>Your "very knowledgeable" rep is talking pure BS. He's a rep, that just about says it all. Of course that's only my humble opinion.</p>

<p>In any case, the large format lens used would have to be exceedingly short to throw an incident angle acute enough to trouble a 56mm wide sensor. And both CMOS and CCD sensors need a similar filter and microlens array over the top of the sensor proper. It's in the geometry of this array where there's an issue with incident angle. Not to mention that many LF wide-angle lenses are of retrofocus design in order to allow a reasonable distance for the bellows and standards to collapse.</p>

<blockquote>

<p> I wide angle lens inherently sends light in at wide angles, doesn't it?</p>

</blockquote>

<p>Not necessarily. If that was strictly the case then a fisheye lens would be impossible. As I mentioned above, a retrofocus design allows for a lens to have a longer back focus (and therefore narrower projection angle) than its focal length would suggest.</p>

<blockquote>

<p>And both CMOS and CCD sensors need a similar filter and microlens array over the top of the sensor proper.</p>

</blockquote>

<p>But that "need" is not absolute. Most of the medium format CCDs actually don't have microlenses, especially all the ones in the 48 x 36 mm class, and that is precisely why they have worked well with technical-camera lenses.</p>

<p>Without microlenses, you get a back which is even more disabled than normal at reaching higher ISOs, but the tech-cam shooters are not too perturbed by that, as long as they get really sharp and colour-faithful images out to the corners.</p>

<blockquote>

<p>Your "very knowledgeable" rep is talking pure BS.<br /> In any case, the large format lens used would have to be exceedingly short to throw an incident angle acute enough to trouble a 56mm wide sensor.</p>

</blockquote>

<p>The BS is from your side, I'm afraid. There are <em>already</em> problems with large format-style lenses on the microlensed 44 x 33 mm sensors (both CCD and CMOS) when shifted, and on the microlensed 54 x 41 mm sensors when straight-on. Check the threads on getdpi.com, especially posts by a chap called Torger who has investigated the pixel crosstalk (colour desaturation) and microlens ripple problems in depth.</p>

<blockquote>

<p>- Yeah, right. That was Leica's excuse for not putting a full-frame sensor in their M8 in 2007. Lo and behold, just two years later in 2009, up pops the M9 with a full-frame sensor and no mention or complaint about poor image quality in the corner of the frame.</p>

</blockquote>

<p>Leica solved this problem by some clever engineering, which obviously took some time to develop. The M8 used standard microlensed CCDs, whereas the M9 (and larger S2) use radially offset microlenses that had to be developed in conjunction with Kodak.<br /> The problem is though, the same approach <em>won't work</em> for medium format tech-cams, because once you shift/tilt the lens, you centre its axis over the wrong microlens geometry - possibly making matters even worse than if you had normal microlenses!<br /> It's by no means going to be an easy problem to solve, and I think the rep that Lobalobo conversed with was on the money in this regard.</p>

<blockquote>

<p>I can not understand / Imagine why there would be a huge issue with blowing up Leica's current concept for their RF CCDs & CMOSes with decentered micro lenses in the format corner to 6x7cm FMF size once such sensors can be made at all.</p>

</blockquote>

<p>As I pointed out to Rodeo, the issue is that such a design is not conducive to the tech-cam style of shooting where shifts and tilts are the norm. It suits Leicas because they are nearly always shot straight-on.</p>

<blockquote>

<p><br /> I do see a future for technical cameras in the digital MF realm. Good WA lenses for them might not exist yet.</p>

</blockquote>

<p>Oh, they do exist! There are some staggeringly good "medium format tech digital" designs in the 23mm - 60mm focal length range from Rodenstock and Schneider (the Rodies are generally preferred in the very short focal lengths). The limitation is not the lenses; it's the way the sensors interact with the images they project. </p>

<p>So, Ray, is it your view that for technical cameras, for tilt and shift, it will continue to be the case that a sensor with no micro-lenses will be required? Would this also mean that tilt-shift will not work well with (at least large) CMOS sensors, which, as I understand, have a problem with angled light even in the absence of micro-lenses? If this is right, then I wonder what the progress of the MF back industry will be. Based on the Phase One, Hasselblad, and Pentax ads for the new Sony sensor, it seems that CCD sensors are destined for the dust bin of history. The market will determine this, I suppose, but I hope there are enough technical camera users to induce some solution to the tilt-shift problem. I'm not a pro--just a hobbyist, currently shooting 4x5 film on a Crown Graphic but with fantasies of buying a technical camera and a medium format back instead of a sports car as a mid-life crisis purchase; hate for the fantasy to die in the name of progress.</p>

<blockquote>

<p>Lobalobo, which dealer or manufacturer? If you go back tomorrow ask at Capture Integration (and say hello to Dave Gallagher for me.</p>

</blockquote>

<p>Not sure I should identify the rep, who may have been off message (from his principal's perspective). Sorry I didn't make it back the next day.</p>

<p>And yet strangely, there are Tilt/Shift lenses as short as 17mm working perfectly acceptably on Full-frame DSLRs with CMOS sensors. How far do you want to shift? Even on film an extreme shift wasn't sharp.</p>

<blockquote>

<p>And yet strangely, there are Tilt/Shift lenses as short as 17mm working perfectly acceptably on Full-frame DSLRs with CMOS sensors. How far do you want to shift? Even on film an extreme shift wasn't sharp.</p>

</blockquote>

<p>Maybe that's the difference between Ray and Joe, a difference in degree, not kind.</p>

<p>I haven't read all the responses, but one thing that is forefront here to get all of these technical issues inline, like the alignment of edge light, is a tremendous amount of $$$. A tremendous amount! How many years away? 5?</p>

<blockquote>

<p>And yet strangely, there are Tilt/Shift lenses as short as 17mm working perfectly acceptably on Full-frame DSLRs with CMOS sensors. </p>

</blockquote>

<p>But that's an apples/oranges comparison. We're talking about "large format-style lenses" here, specifically the modern "digital" wideangles which are far better in edge performance (on the right sensor) than medium format SLR retrofocus designs. The full-frame DSLR tilt/shifts, OTOH, like Canon's 17mm and 24mm, are as strongly dependent on retrofocus design as medium format SLR wideangles. Just look at how far from the focal plane their rearmost optical surfaces are - in the 65 mm ballpark (44mm for the camera body flange + another ~2cm forward into the lens barrel, to give clearance for the tilt/shift mechanisms). I'd expect such a DSLR retrofocus wideangle to behave fine on a CMOS sensor, just as I'd expect a medium format retrofocus wideangle to behave fine on the new larger CMOS sensors. </p>

 

<blockquote>

<p>How far do you want to shift? Even on film an extreme shift wasn't sharp.</p>

</blockquote>

<p>I'm no practitioner of shifting, but I gather that those who do, expect to be able to perform a shift of about half the height of the sensor. </p>