disaster: cfv-50c + 24 Apo Digitar lens + cambo 1200wrs

[marcgoodwin]

I've recently purchased a Hasselblad cfv-50c digital back to use with a Schneider 24 Apo Digitar lens mounted on a Cambo WRS 1200 pancake camera. The first test is a total disaster. Apart from coverage issues which I expected to some extent with a wide angle lens like this, there seems to be some really weird colour fringing / blooming that I did not expect (see screen shots). Many questions arise.

Is this combination of back + camera + lens simply not compatible?
Could this be a problem with the sensor in which case I can perhaps still return the back?
Is this a problem with the lens, as the dealer suspects?
What solutions are there?

[Doug Peterson]

Your dealer can familiarize themselves with the issue of small micron microlensed sensors and symmetrical wide angle lenses here: https://digitaltransitions.com/phase-one-iq250-tech-cam-testing/

A lot of work has gone into Phase One's Capture One software to handle extreme color cast and falloff as well as possible. So the results of our Phase One testing may not be fully relevant to a Hassy back using Phocus.

[Christoph B.]

I think the 4th image is very telling; the color shifts correspond to the movement of the lens, so I'm pretty sure it's the lens and not the back that causes the problem.
Try projecting an image with the lens on a transparent piece of paper, if you can see the same strange colour shifts in the same places I'd ask the dealer for a replacement.

Not sure whether it could be caused by microlens issues on the back (24mm is a rather wide angle and will cause the light to fall on the sensor on a very hard angle on the edges of the image circle) but it's a possbility...

@Doug; wouldn't that cause the lens to produce those strange color shifts all over the edge of its image circle and not just in those strange hotspots?

[torger]

The CFV-50c is not compatible with the Apo Digitar 24XL, every dealer should know this. If they don't they should not be selling tech camera gear. This is really upsetting me, for real. Selling gear at this level and not knowing s**t about how it performs is an insult to the customers who spend huge amount of cash on these type of systems.

It's not like it's a secret how these lenses are designed, you can even get full lens data and software from Schneider for the full Digitar series. It's a double gauss symmetric lens, very low angle of incoming light, even tougher on the sensors than the more well-known 28XL.

There's nothing strange with the blob spotty pattern, this is what the Sony sensor does when crosstalk happens. As a rule of thumb if you care about image quality is that you should never have the yellow blob visible in your LCC shot, that indicates that crosstalk levels are simply too high.

As far as I know Capture One is no better than Phocus at correcting the crosstalk, that is they do nothing to correct it. Capture One is better in hiding Dalsa CCD artifacts like tiling and microlens ripple, but the Sony sensor does not have those issues, but instead have huge issues with crosstalk when faced with a short symmetric lens, and this cannot really be canceled out in full. I'm sure Doug wouldn't sell an IQ350 to be used with an SK24XL, Capture One or not, because he's one of the dealers that actually have this knowledge.

Dealers love to say that "it's a problem with the lens", but there's nothing wrong with the lens. The problem is in the sensor design that they don't have light shields between pixels like the Kodak CCDs had, which existed at the time when this lens was made. Without light shields and those tiny deep pixels you get light leaks like crazy between pixels (crosstalk) which no LCC algorithm can cancel out. The reason that it doesn't have light shields is that it compromises other aspects of pixel design and well, the Sony sensor was not made for tech camera users, it was made for SLR cameras.

It doesn't make much difference to us users though. Kodak is dead, Schneider Digitar have ceased manufacturing and Sony is the new standard so you need lenses adapted for that sensor. There are no tech camera lenses on the market today that have been designed to work with the Sony sensors, but you can still get good results with Rodenstock Digaron lenses which are a little bit retrofocus, as long as you keep within certain limits. Those where designed with the 6um Dalsa in mind, and the Sony is even worse than those when it comes to crosstalk, so you just have to live with that you can't use the full shift range with retained quality.

This has been known for since the IQ250 was new, so that dealers still today don't know this is a scandal. How hard can it be? The expensive dealer structure with those huge margins are supposed to help the customer makes the right choices, but obviously it doesn't work very well...

The manufacturers have chosen to not talk about crosstalk, I don't know why, as it's a standard term used in the sensor manufacturing business, sensors are designed to handle a particular max angle of incoming light, and lenses are designed to deliver a particular max angle of light, if those are not compatible they should not be used together. It's as simple as that. However, in the will to sell more gear(?) these rules are nowadays ignored and they push to the dealers (which in some cases like here push it further to the customers obviously) to decide how much you can break the rules and live with reduced and unpredictable color fidelity. And, oh, without sharing the technical facts. It's not serious, but that's the way the medium format companies like to roll.

[eronald]

I suspect the colored spots might be re-reflections of light that is mirrored back towards the lens from the sensor or its cover glass.
This issue plagued some Canon lenses too, leading to a redesign.
Of course, Torger has a different opinion, and he is probably right. Or both of us

Edmund

The CFV-50c is not compatible with the Apo Digitar 24XL, every dealer should know this. If they don't they should not be selling tech camera gear. This is really upsetting me, for real. Selling gear at this level and not knowing s**t about how it performs is an insult to the customers who spend huge amount of cash on these type of systems.

It's not like it's a secret how these lenses are designed, you can even get full lens data and software from Schneider for the full Digitar series. It's a double gauss symmetric lens, very low angle of incoming light, even tougher on the sensors than the more well-known 28XL.

There's nothing strange with the blob spotty pattern, this is what the Sony sensor does when crosstalk happens. As a rule of thumb if you care about image quality is that you should never have the yellow blob visible in your LCC shot, that indicates that crosstalk levels are simply too high.

As far as I know Capture One is no better than Phocus at correcting the crosstalk, that is they do nothing to correct it. Capture One is better in hiding Dalsa CCD artifacts like tiling and microlens ripple, but the Sony sensor does not have those issues, but instead have huge issues with crosstalk when faced with a short symmetric lens, and this cannot really be canceled out in full.

Dealers love to say that "it's a problem with the lens", but there's nothing wrong with the lens. The problem is in the sensor design that they don't have light shields between pixels like the Kodak CCDs had, which existed at the time when this lens was made. Without light shields and those tiny deep pixels you get light leaks like crazy between pixels (crosstalk) which no LCC algorithm can cancel out. The reason that it doesn't have light shields is that it compromises other aspects of pixel design and well, the Sony sensor was not made for tech camera users, it was made for SLR cameras.

It doesn't make much difference to us users though. Kodak is dead, Schneider Digitar have ceased manufacturing and Sony is the new standard so you need lenses adapted for that sensor. There are no tech camera lenses on the market today that have been designed to work with the Sony sensors, but you can still get good results with Rodenstock Digaron lenses which are a little bit retrofocus, as long as you keep within certain limits. Those where designed with the 6um Dalsa in mind, and the Sony is even worse than those when it comes to crosstalk, so you just have to live with that you can't use the full shift range with retained quality.

This has been known for since the IQ250 was new, so that dealers still today don't know this is a scandal. How hard can it be? The expensive dealer structure with those huge margins are supposed to help the customer makes the right choices, but obviously it doesn't work very well...

The manufacturers have chosen to not talk about crosstalk, I don't know why, as it's a standard term used in the sensor manufacturing business, sensors are designed to handle a particular max angle of incoming light, and lenses are designed to deliver a particular max angle of light, if those are not compatible they should not be used together. It's as simple as that. However, in the will to sell more gear(?) these rules are nowadays ignored and they push to the dealers (which in some cases like here push it further to the customers obviously) to decide how much you can break the rules and live with reduced and unpredictable color fidelity. And, oh, without sharing the technical facts. It's not serious, but that's the way the medium format companies like to roll.

[eronald]

I think the 4th image is very telling; the color shifts correspond to the movement of the lens, so I'm pretty sure it's the lens and not the back that causes the problem.
Try projecting an image with the lens on a transparent piece of paper, if you can see the same strange colour shifts in the same places I'd ask the dealer for a replacement.

Not sure whether it could be caused by microlens issues on the back (24mm is a rather wide angle and will cause the light to fall on the sensor on a very hard angle on the edges of the image circle) but it's a possbility...

@Doug; wouldn't that cause the lens to produce those strange color shifts all over the edge of its image circle and not just in those strange hotspots?

Whether it is a lens or sensor issue -my explanation or Torger's- , I would expect it to shift if you shift.

Edmund

[torger]

I suspect the colored spots might be re-reflections of light that is mirrored back towards the lens from the sensor or its cover glass.
This issue plagued some Canon lenses too, leading to a redesign.
Of course, Torger has a different opinion, and he is probably right. Or both of us

Let me be humble, this is the first time I see an 24XL LCC shot. It's the worst I've seen so far, which is exactly what we would expect as it's probably the most extreme lens you can get in terms of low angle of incoming light. I think I have a 28XL LCC shot from the IQ250 (same sensor) laying around somewhere which I looked at when the IQ250 was new, so it's a while ago so I don't remember the exact shape of the patterns. I remember the yellow blob though, and that blob can be seen on shifted Rodenstocks too.

What you see when comparing LCC shots from different focal lengths is that you see the same pattern with the same colors just more or less "zoomed in". The 24XL is "zoomed out" more than every other lens I've seen so it reveals a larger pattern. Most lenses don't reach past that yellow blob, so the purple blobs above it I have rarely, if ever, seen. I suspect it's visible on the 28XL too, but I don't have the shot here on this computer so I can't compare.

[JaapD]

Sensors with microlenses are in general not a good combination with tech cameras. Personally I hate microlenses from the first moment they arrived. Yes, it gives some added sensitivity, and CMOS needs it more because it’s main sensitivity is in the center of the pixel (CCD has a nearly flat sensitivity), but that’s it. In the development department we used to speak of ‘sensors with spectacles’.

For tech cameras a CCD sensor provides a better solution.

[torger]

Newer Dalsa CCDs, P65+, IQ260 etc also have microlenses.

There's an additional advantage of microlenses which is that aliasing is reduced as the dead gap between pixels where no light is registered is reduced, and thus you don't get as much aliasing as with the older CCDs without microlenses. I wouldn't say it's a huge advantage though, but it's there.

With the Dalsa CCDs the microlenses lead to "microlens ripple" on the wides, which is cleaned up quite effectively by an LCC algorithm designed for the purpose (like in C1). Interestingly enough the Sony microlenses doesn't lead to microlens ripple in the same way. I've seen some suspicious low-frequency waviness which could be due to microlenses but I'm not sure.

In any case I think we could deal with a microlens-equipped sensor, just like with the Dalsas, and clean up possible artifacts -- especially since the Sony microlenses seems to leave less artifacts than Dalsa's. What we need is light-shielded pixels so that there's rather pixel vignetting than pixel crosstalk. The Kodaks in P45+ and my own H4D-50 has light shields and as such works even with the most extreme lenses, like this SK24XL.

Tech wide angle photography is a tiny niche though so getting sensors designed specifically for us is not likely to happen, especially with the CMOS technology which is harder to customize than CCDs (afaik). If we're lucky we'll see new pixel designs which as a side effect gets wider angular response, back-side illumination (BSI) tech is one such thing, but the advantage may be nullified by that pixels become smaller and smaller. It's much easier to make wide angular response of a small pixel than a large. But it's megapixels that sells medium format, not angular response...