Pages: [1] 2   Go Down

Author Topic: Schneider Digitar Question  (Read 4495 times)

JoeKitchen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5022
Schneider Digitar Question
« on: September 01, 2014, 11:03:52 am »

Has anyone heard whether or not Schneider will be or is designing lenses that can compete with the Rodenstock wides on the higher density pixel backs? 
Logged
"Photography is one percent inspiration and ninety-nine percent

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #1 on: September 02, 2014, 06:06:02 am »

That would be in the sub-40 range I guess, above 40 the current range is competitive.

The 28 and 35 is tough on the high density backs, and they have more sharpness falloff towards the edges than the Rodenstock 32. On the plus side is less distortion, lighter weight and lower cost.

It's quite "easy" to make a competitive lens, just make a retrofocus design with lots of glass elements and mount them very precisely. Voila you have a Rodenstock 32 monster ;). The right positioning for the Schneider Digitar brand I think however is to make simpler designs than Rodenstock, make them lighter and lower cost, allow large image circle although with some noticable sharpness falloff. Making exact same type of lens as Rodenstock would be a bit boring (unless they succeed making them cheaper :) ).

The problem with the 28 and 35 however is that they are very limited on high density backs, due to crosstalk issues you can't shift much without getting image quality issues (loss of color fidelity). They are so limiting that some intermediate design would indeed be nice, I think it should be possible to keep most of the Digitar features while still working well on high density backs.

I'm a bit worried if we would see a trend towards more and more retrofocus lenses though. MF sensors have had a strong advantage compared to 135 sensors that they are much less crosstalk-prone and thus allows for better wide angle designs. If MF move into higher and higher pixel densities it may lose this advantage, and tech cams end up as just another mirrorless camera, just bigger. So from my perspective I'm more concerned about digital back makers not supporting "large format style" wide angles, than that Schneider haven't produced lenses that support the current digital backs.

Just as you I'm very curious to hear about rumors, what's in the making or not...
Logged

Enda Cavanagh

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 636
    • http://www.endacavanagh.com
Re: Schneider Digitar Question
« Reply #2 on: September 02, 2014, 09:06:57 am »

I think the problem for the lens manufacturers also is that there isn't a constant for them to work to, as in the sensors mostly settling around the 49 x 37mm size for a number of years than moving towards larger sizes with micro lenses and now back down to a 44 x 33mm CMOS sensor. They are a relatively small company after all. I would love to see a more future proof, more compact and possibly cheaper wide angle Schneider solution though. The 35mm which I love does need to be updated for sure. The 28mm is a great lens on certain backs. I use it on the H3D 39 and have used it on the H5D 50 and love it but the larger sensor backs and newer CMOS sensor backs are a no go area for it. It's one of the reasons I haven't upgraded my back and instead bought the Sony A7r and Canon TS lenses as a second camera system.

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #3 on: September 02, 2014, 09:44:42 am »

I agree that the moving sensor standard is a problem.

The SK28 is a mystery though, it was released in 2011, the P65+ was released in 2008, and by 2011 there was many other backs with Dalsa's 6um technology. Schneider should have known that it would not work well. Maybe it was designed for the future, but unfortunately no sensors with desired properties appeared? Either that or they did not actually test it together with modern backs at the time, which seems strange to omit.

I think 49x37mm was a good size for tech cams, but understandable the back makers had to move to full-frame 645 as they have 645 cameras.

On the other hand, large format film also had various film sizes to work with, mainly 4x5, 5x7 and 8x10. 44x33, 49x37 and 54x41 is tighter than that. More than varying sensor size I think the varying color cast / crosstalk sensitivity is an issue. This is a technology issue though hard to overcome. Pressed by the 135 systems and also providing an argument for upgrade MF backs needs to put more pixels in there, leading to smaller pixels, leading to higher crosstalk, making traditional large format wide angle designs impossible.

In a way I think the lens manufacturers should have seen this coming, a layman could have guessed it.

I don't think a company like Schneider have lost much money on the SK28 design though. This type of lens is probably quite fast from design to production, and production is more or less "made to order", so maybe that's an explanation why the lens design does not seem to be that thought through.
« Last Edit: September 02, 2014, 10:01:52 am by torger »
Logged

Enda Cavanagh

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 636
    • http://www.endacavanagh.com
Re: Schneider Digitar Question
« Reply #4 on: September 02, 2014, 11:13:34 am »



The SK28 is a mystery though, it was released in 2011, the P65+ was released in 2008, and by 2011 there was many other backs with Dalsa's 6um technology. Schneider should have known that it would not work well. Maybe it was designed for the future, but unfortunately no sensors with desired properties appeared? Either that or they did not actually test it together with modern backs at the time, which seems strange to omit.

I remember when I bought the 28mm the marketing spiel highlighted the fact that it WAS designed with the new larger sensor backs backs and that it could handle the extra resolution etc. What makes it even more frustrating is surely Phase One have a good working relationship with Schneider. (due to them making the lenses for their camera). Some where along the way there was an almighty cock up that has been brushed under the carpet to al large extent. Hasselblad users generally don't work with view cameras. I'm one of the few but you barely hear that the H4D 50 and H5D 50 works fine with the Schneider wide angler lenses ( I tested the H4D 50) I was just messaging with Torger and I said hasselblad marketing just seem to be oblivious to a missed opportunity for view camera users.


On the other hand, large format film also had various film sizes to work with, mainly 4x5, 5x7 and 8x10. 44x33, 49x37 and 54x41 is tighter than that. More than varying sensor size I think the varying color cast / crosstalk sensitivity is an issue. This is a technology issue though hard to overcome. Pressed by the 135 systems and also providing an argument for upgrade MF backs needs to put more pixels in there, leading to smaller pixels, leading to higher crosstalk, making traditional large format wide angle designs impossible.


But than you also didn't have the enormous cost of a digital back to deal with. With digital MF, when buying a system photographers pick it best on what suits their needs. Regarding lenses most photographers (not wealthy amateurs) have no plans to completely change their lenses every few years because a wide angle is no longer a wide angle or because your very expensive wide angles can no longer be used. And that's before shelling out 10's of thousands for the backs.


In a way I think the lens manufacturers should have seen this coming, a layman could have guessed it.


As I said people made a huge cock up



I don't think a company like Schneider have lost much money on the SK28 design though. This type of lens is probably quite fast from design to production, and production is more or less "made to order", so maybe that's an explanation why the lens design does not seem to be that thought through.

I'm not sure sure. I'd ask the question. If the Schneider 28mm and Rodie 32mm could both be used on the large Dalsa sensors and Sony Sensors how many would go for the lighter, cheaper, wider angle lens with minimal distortion and a huge image circle. I know which one I'd choose. Schneider messed up on this one. They didn't do enough testing. They surely have a good working relationship with the main digital back manufacturer plus they did market it as a lens for the future.
« Last Edit: September 02, 2014, 11:15:07 am by Enda Cavanagh »
Logged

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #5 on: September 02, 2014, 11:25:54 am »

people made a huge cock up

Okay, I have to agree that your theory is better than mine :)

I realize that when you described how market speak was, I had kind of forgot about that. It's a bit symptomatic that tech cam lens performance has always been down to user testing. There were many involuntary lens switches from Schneider to Rodenstock with the P65+ to IQ180 upgrades (the IQ180 performs even worse with symmetric wides), the information was not there in advance, the users discovered it after upgrade.
« Last Edit: September 02, 2014, 11:28:58 am by torger »
Logged

Paul2660

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4066
    • Photos of Arkansas
Re: Schneider Digitar Question
« Reply #6 on: September 02, 2014, 12:00:03 pm »

I still regard the testing Guy Mancuso did back about 1 year ago with the 2 28's a great article.  He used a IQ160 as I recall and you could see where the 28XL was going to have trouble after 8mm or so of shift.  The 28mm Rod as good as it is, (IMO) is sooo limited by the 70mm image circle and the stupid image circle market that Rodenstock places inside the lens which creates the hard vignette when hit.  My 28mm would easily go to 10mm of shift without that in most instances which for a 28mm is a lot of shift. 

The big question that only a few know, is what the future of the big MF digital sensor is. 

Now that Phase One agrees that a CMOS chip CAN look the same as CCD (pretty much their standard line since the advent of the IQ250),  and it's also clear that even with all the "extra" that went into the IQ260 to give it longer exposures up to 1 hour, that the next chip will be CMOS since CCD is really only optimized for base iso, and that in many instances is must too limiting.  Way to limited especially with a tech camera which only has a Copol 0 shutter and limits you to hard shutter speeds no 1/3 or 1/2 speeds, that you gain with a Alpa FPS. 

If I were Rodenstock and Schneider, I would either be at Phase One's door asking what's next, i.e. Full frame CMOS micro lenses or not??  That's the big question as it's very clear to me and anyone who reads reviews and looks at images taken with the IQ250, that that chip is amazing. 

But it's a 1:3 crop and won't work well with shifts on tech wides and in some instances even the 60XL.  The key is what the next chip is and will the current line up of lenses out there which BTW none are cheap. 

If the answer is yes we have a new chip, CMOS Full frame at 60MP to 80MP BUT you have to get all new lenses, I am pretty much done.   

Paul

Logged
Paul Caldwell
Little Rock, Arkansas U.S.
www.photosofarkansas.com

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #7 on: September 02, 2014, 03:14:28 pm »

It will be very interesting to see where tech camera land is in say 5 years. I have no idea. There are both bright and dark scenarios.

While tech cam manufacturers really depend on lenses and backs that play together, the lens makers (Schneider and Rodenstock (=Qioptiq)) will survive even if they discontinue the whole lens lines, the "large format digital" category is more about tradition and prestige than moneymaking. Copal shutters have already been discontinued, and while stock still exist and you do get a Copal shutter if you buy new today, it won't last forever, and no attractive electronic shutter alternative has yet appeared. DHW HS-1000 is promising, but compared to the Copals they are crazy expensive (~$2100 per shutter was last I heard, with a pancacke camera mount on that you'll be paying more for shutter and mount than glass for most lenses).

There's lots of interesting sensor technology growing in the small sensor market (ie mobile phones these days) but if it will ever reach MF sized sensors is unclear. In the best case we do get back-illuminated instead of the current front-illuminated designs:



which opens up for great wide-angle performance, even with relatively small pixels.

In the darkest of scenarios the sensors will just continue to become worse concerning color cast and crosstalk, making even Rodie wides limited, and Schneider and Qioptiq don't make anything new and just wait until the market self-dies, at least the wide-angle side of it. I guess Sinar-style studio product photography with longer lenses and clunky electronic shutter systems will still have its niche.

In a bright scenario we do get a wide-angle-capable 645-sized CMOS, we get a worthy replacement of Copals (or a restart of Copal manufacturing), and lenses like the SK28 and SK35 start working great again, and maybe Schneider even makes new versions for those that needs even sharper images.

If Copal no longer will be the standard shutter and instead some electronical shutter, I'd like to see an electronic interface from the tech cam bodies for registering movement settings in the EXIF data. Although I find the 100% mechanical operation charming, I'm sure "regular joe" would prefer having lens/aperture/shutter-speed/tilt/swing/rise/fall/shift/focus settings all registered (would mean that people could skip LCC by the way, and open up for automatic digital lens corrections like we have on regular cameras). That together with live-view and high ISO CMOS would actually make tech cams user friendly :)
« Last Edit: September 02, 2014, 03:16:43 pm by torger »
Logged

buckshot

  • Guest
Re: Schneider Digitar Question
« Reply #8 on: September 02, 2014, 06:13:51 pm »

If the answer is yes we have a new chip, CMOS Full frame at 60MP to 80MP BUT you have to get all new lenses, I am pretty much done.  

Amen to that. As much as I'd love a CMOS 60/80MP back, if it means a huge $$$ hit/investment in terms of glass, I'll stick with my P45+ (still the best sensor design for a shifting on a tech cam, ever), my S/K Digitars, and probably look to add something like a used Leaf II-12 for single hi-res shots.

Jim
« Last Edit: September 02, 2014, 06:15:26 pm by buckshot »
Logged

JoeKitchen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5022
Re: Schneider Digitar Question
« Reply #9 on: September 02, 2014, 09:28:01 pm »

Wow, not a whole lot of love with this topic.  

Personally, I asked because I am thinking about the SK 28mm as a possible lens that I would like to purchase (along with the SK 43mm).  Between the two 28mm on the market, I think that the Schneider version is the optimal lens.  On my P45+ I would get to shift that lens as much as it has to offer.  If I were to upgrade to a IQ260, I still would be able to shift that lens more then the Rod 28mm has to offer because of the much smaller IC of the Rod version.  

The caveat here is that my SK 35mm is my sweet spot for interiors, and I regularly shift that lens to its max.  A few days ago I was able to test the 35mm on my P45+ and my girlfriend's IQ260, and I was kind of bummed out.  After about 10mm of shift, loss of color saturation due to lens cast was an issue.  Also, loss of detail due to cross talk and noise (due to greater observed light falloff with a stronger correction) became a real issue around 14mm of shift.  

If I were to upgrade to an IQ260 in the next year or so, the current obvious solution would be to replace my 35mm with the Rod 32mm.  This is sort of okay (kind of expensive and I would now have to deal with distortion), but doable.  The only issue would be that although a 28, 35, 43 would be a great line up, a 28, 32, 43 would not be so great.  I kind of feel the 32 is too close to the 28 to make it worth buying the 28mm.  

So, my hope is that Schneider is working up some new 35 (or 36 or 34) lens that is an in-between design that would work well with the higher density backs.  

One can only hope though.

Insofar as CMOS taking over CCD, I do not think that it is going to happen anytime soon, or at least not in the next generation.  CMOS is incredible more difficult to make than CCD, so the idea of Sony spending the money to make a sensor full framed for a very small market is not forceable, especially if Nikon and Canon are not players yet.  Also, there will always be a need for CCD due to applications that CMOS can not be used for, such as for use in satellites.  
« Last Edit: September 02, 2014, 09:30:10 pm by JoeKitchen »
Logged
"Photography is one percent inspiration and ninety-nine percent

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #10 on: September 03, 2014, 03:51:52 am »

Just to inform, loss of color saturation is a crosstalk effect. With crosstalk the color channels get mixed, that is evened out, that is closer to monochrome, that is loss of saturation. While crosstalk also will lead to softened detail (lower contrast between two adjacent pixels due to the mixing) it should be much less than the softening caused by the natural lens sharpness falloff. The visible contrast lowering should stretch one pixel, and lens softening is generally larger than that in the area where crosstalk occurs. Noise is due to the general pixel vignetting, color cast due to varying pixel vignetting.

The P45+ has the Kodak KAF-39000, and as I've recently heard from Enda the 50 megapixel (6um pixels) KAF-51000 used in Hasselblad CFV-50 and H5D-50 also performs well with these lenses, unlike the Dalsa 6um (IQ160, P65+ etc). Going from 39 to 50 megapixels may seem like a too small upgrade though (and you'll lose long exposure). For me that is on 33 megapixels it would be a suitable upgrade though. Actually I find 50 megapixels and the 39x47mm sensor size to be a very nice sweetspot of size relative to the common 90mm image circle (nicer than full-frame 645), and good resolution too. As my shooting style require f/11-f/16 there's not too much gain to go with smaller pixels than that.

As I have expertise in crosstalk analysis I'd like to further investigate the actual performance of the CFV-50/HxD-50 on these lenses, but it's hard to find tech users with any of these backs and lenses that can provide me with raws for analysis :-\. If it actually does perform well I'd say that the CFV-50 is a very under-appreciated back and it's a pity that it has not been more popular in the tech cam community, where Phase One instead rules. Hasselblad is totally clueless about tech cameras though, so it's their fault not showing it's superiority in tech cam use. And now they have discontinued it, and the replacement CFV-50c is near useless for tech wides. The H5D-50 is still current though, but not sold back alone (I assume). The KAF-51000 sensor is still current in TrueSense Imaging's product catalog, so anyone can still make new backs with it.

The critical crosstalk angle is unfortunately not specified in the data sheets, but if I compare angular response (indicates color cast, which is often related to crosstalk, ie more color cast more crosstalk) between the KAF-39000 (P45+) and the KAF-51000 (CFV-50) they look about equal. On the other hand Dalsa's 6um is not far behind so these diagrams don't say that much as they only look at quite narrow span. Dalsa only measure +/-30 degrees while Kodak/Truesense have +/-40 degrees in their diagrams, which might indicate that Kodak handles a wider span, but a stronger indication is that the Kodak lacks microlenses while Dalsa have them. I think the SK28 will deliver light at a 55-60 degree angle at the edge of the image circle.

The KAF-39000 support 39 degrees before crosstalk occurs (http://www.optcorp.com/pdf/FLI/31Mp_and_39Mp_Full-Frame_CCD_paper.pdf). It should be noted that small amounts of crosstalk don't disturb the image too much, so pushing a little bit past that is not too bad (unless you're doing reproduction work where color accuracy is extremely important, but you wouldn't use ultra-wides for that anyway I guess). I haven't found any number for the KAF-51000.

Concerning sensor performance in ideal conditions the Phase One Dalsa backs is in measurements better than the Hasselblad Kodak backs, slightly better DR and better color separation, however that will be more than nullified on tech wides after LCC correction has been applied.
« Last Edit: September 03, 2014, 04:12:14 am by torger »
Logged

Enda Cavanagh

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 636
    • http://www.endacavanagh.com
Re: Schneider Digitar Question
« Reply #11 on: September 03, 2014, 11:18:29 am »

Joe
I use and love the 28mm and 35mm on the H3D 39. If you go for the IQ260, I would suggest the Rodie 32 and 40. You end up with the same field of view approximately as what you are used to and you don't get the colour cast issues. Another suggestion is to go for a hasselblad H4D or H5D 50 (H4D is much cheaper and you get the same sensor. Also you can use the quantum batteries to power it to give you countless hours of operation. The H5D cameras use a different voltage which prevents it from been powered. It just shuts down after a few minutes. Grrrrrrr) If you go for the Hassie option you can use the Schneider 28, 35, and 47mm lenses without any issues re. colour cast. It would be by far the cheapest solution. You will max out at 128 seconds but than you don't have to deal with that blood dark frame of the Phase One backs (I know I sound like a broken record on this but it does drive me nuts : :P )  

Torger I just sent you a message but I'd be glad to send you any calibration shots I have from when I had the H4D 50 on loan. I shot with  the Schneider 28, 35 and 47mm lenses with a lot of shift/rise and no colour cast issues. (vignetting due to really pushing it to the limit of the image circle yes but no more than the H3D 39 that I'm used to)

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #12 on: September 03, 2014, 02:43:59 pm »

I got some test shots from Enda, it will take some time to make more scientific analysis of them and make comparison with Dalsa, but oh boy, just looking at them in the raw converter and doing some basic tests reveals the following:

- no tiling issues whatsoever (virtually all Dalsa sensors have tiling which becomes extra tough on wides, a plague for high contrast B&W processing)
- no microlens ripple or similar issues (not surprising as I now know this sensor lacks microlenses unlike Dalsa 6um)
- most likely some crosstalk on the SK28 (not measured yet) but not enough to show any obvious desaturation, so I guess it's just as good as the P45+.
- like a zillion times better than P40+/P65+/IQ140/IQ160/IQ260 or any other Dalsa 6um back for symmetrical wides like the SK28.

Why did noone (except Enda I guess :) ) tell us? How could this have been such a well-kept secret? A Hasselblad 50 megapixel is the tech cam back for wide angles. Someone at Hasselblad's marketing department should be fired. Or just shot ;). They could have done their FUD thing against their competitors as usual and for once back it up with fact.

For a system like mine, 48x36mm 33 megapixel with SK lenses, an upgrade to a 50 megapixel Hasselblad seems a lot more painless and cheaper than going for full-frame 60 megapixels.

I think this is pretty big news for all of us at a budget using 22/33/39 megapixel backs and thinking about an upgrade. The are other upgrade paths than the Phase One and only...
« Last Edit: September 03, 2014, 02:50:43 pm by torger »
Logged

Enda Cavanagh

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 636
    • http://www.endacavanagh.com
Re: Schneider Digitar Question
« Reply #13 on: September 03, 2014, 02:58:05 pm »

I got some test shots from Enda, it will take some time to make more scientific analysis of them and make comparison with Dalsa, but oh boy, just looking at them in the raw converter and doing some basic tests reveals the following:

- no tiling issues whatsoever (virtually all Dalsa sensors have tiling which becomes extra tough on wides, a plague for high contrast B&W processing)
- no microlens ripple or similar issues (not surprising as I now know this sensor lacks microlenses unlike Dalsa 6um)
- most likely some crosstalk on the SK28 (not measured yet) but not enough to show any obvious desaturation, so I guess it's just as good as the P45+.
- like a zillion times better than P40+/P65+/IQ140/IQ160/IQ260 or any other Dalsa 6um back for symmetrical wides like the SK28.

Why did noone (except Enda I guess :) ) tell us? How could this have been such a well-kept secret? A Hasselblad 50 megapixel is the tech cam back for wide angles. Someone at Hasselblad's marketing department should be fired. Or just shot ;). They could have done their FUD thing against their competitors as usual and for once back it up with fact.

For a system like mine, 48x36mm 33 megapixel with SK lenses, an upgrade to a 50 megapixel Hasselblad seems a lot more painless and cheaper than going for full-frame 60 megapixels.

I think this is pretty big news for all of us at a budget using 22/33/39 megapixel backs and thinking about an upgrade. The are other upgrade paths than the Phase One and only...

Great to see the results been so positive and glad to help :)

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #14 on: September 03, 2014, 03:56:11 pm »

Forgot to mention that the SK28 test shot is ~14mm lens rise and ~12mm sideways shift and sensor in vertical (portrait) orientation, ie a pretty big shift, the sensor corner is at the very edge of the image 90mm image circle, and even there there is no strong crosstalk.

Actually I can't really measure the crosstalk amount as I need a special red-filtered LCC for that, but I can see side effects by green channel separation in the LCC shot, and there is some of that but to a mild extent. Should generally not produce any significant color fidelity issues. The test shot has very muted colors so it's not a good test for that particular thing, but as said it's likely that there is not any big issue.

As far as I know this shot is made with center filter, but there is quite steep vignetting loss at the image circle edge, 2 2/3 stops, which is only about 1 stop on the Dalsa. Why? Because on the Dalsa there is microlenses and huge crosstalk, and on the Kodak there's light shields so light not getting into the right pixel don't get registered at all, ie pixel vignetting.

This light loss is the only downside, and in some cases it can be problematic. Due to this pixel vignetting there's noise a the image circle edge, which could be solved by HDR merging if the situation requires. Shifting the 28 to the edge like here should not be that common though. At 80mm image circle you lose 2 stops (ie "ISO200" quality out there if you shoot at ISO50). Much easier to live with than crosstalk, but still means that there's no perfect sensor available for the SK28.
« Last Edit: September 03, 2014, 04:15:54 pm by torger »
Logged

Enda Cavanagh

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 636
    • http://www.endacavanagh.com
Re: Schneider Digitar Question
« Reply #15 on: September 03, 2014, 05:06:02 pm »


As far as I know this shot is made with center filter, but there is quite steep vignetting loss at the image circle edge, 2 2/3 stops, which is only about 1 stop on the Dalsa. Why? Because on the Dalsa there is microlenses and huge crosstalk, and on the Kodak there's light shields so light not getting into the right pixel don't get registered at all, ie pixel vignetting.

This light loss is the only downside, and in some cases it can be problematic. Due to this pixel vignetting there's noise a the image circle edge, which could be solved by HDR merging if the situation requires. Shifting the 28 to the edge like here should not be that common though. At 80mm image circle you lose 2 stops (ie "ISO200" quality out there if you shoot at ISO50). Much easier to live with than crosstalk, but still means that there's no perfect sensor available for the SK28.

Yes there is a strong vignette but if you consider the extreme amount of rise and shift in the shot it's pretty understandable. I do merge the images using the fusion option in Photomatix to get a natural affect rather than LSD HDR affect :D and that does help with the noise you speak about. If you consider the focal length and the fact it can get 14mm rise and 12mm shift it is an amazing lens. You are right of course that it's not something you have to do all the time but it comes in so handy where you can't walk back that extra meter or 2.

You're completely right about Hasselblad taking their eye of the ball. There's a lot more I could say about that but I won't. I have mentioned it multiple times from memory on the forums but most medium format/view camera users on LuLa and Getdpi seem to be Phase One users  and the sales guys on the forums are also in the PhaseOne corner too. But again Hasselblad should be aware of it.

JoeKitchen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5022
Re: Schneider Digitar Question
« Reply #16 on: September 03, 2014, 09:59:32 pm »

Very interesting that the 6 micron Kodak sensor does not have the intensity of the problems with lens cast and cross talk that the Dalsa sensor does.  If only Kodak actually said "yes Phase One, we would love to try and make a full framed CCD for you." 

Problem with the Hassy backs is that I frequently do very long exposures, sometime more then two minute, for my interiors.  So Hassy is not an option. 

In any event, I will be waiting until the next generation of P1 back to be released before I make any decisions on whether or not to upgrade.  Maybe, and it is more a dream than a probability, Phase One will actually listen to us Tech camera users and try to make a better back.  Maybe an IQ350 with the 50 MP Kodak chip?  Probably not going to happen, but personally I would rather have a better sensor that produces better files and handles lens cast better than more mega-pixels. 

Insofar as a new lens, I am kind of stuck.  I really like the SK 28mm and SK 43mm, but I am not sure if buying them right now would be a good move. 
Logged
"Photography is one percent inspiration and ninety-nine percent

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #17 on: September 04, 2014, 04:18:00 am »

Phase One should have made a P50+ in the P+ series when the KAF-51000 was released in 2008 and release it together with the P65+, ie identical to P45+ but with this newer sensor, having long exposure and all. However I'm sure they made a business decision that there should be only two sizes, 44x33 and 54x41, and having a third size 49x37 inbetween would only make pricing more complicated, that size is not likely to ever return into their product line. Hasselblad still have all these three sizes in current products though.

Guessing a bit I think what makes the difference between Dalsa 6um and Kodak 6um here is that 1) there is no microlenses and 2) there is proper shielding between pixels so the sensor rather vignettes than crosstalks. So even if Dalsa would make a version of the sensor without microlenses it would not work well, there need to be shielding too. Having microlenses is in cases it works better though, you get more coverage ie a bit less aliasing and better quantum efficiency. Probably the shielding makes you lose some coverage too. Choosing between two evils I prefer the Kodak design currently, but what we want in the long term is a new sensor which is back-illuminated or in some other way has drastically reduced stack height so the pixels gets a natural wide angular response as a side effect.

One problem is that the only application for wide angular response may actually be tech cams. I don't think the scientific and industrial applications which use these CCDs need that, so there's little motivation for Dalsa to design a sensor that has that property.

It's a matter of taste too, you can perfectly well argue that wide angle lenses should be retrofocus, we should once for all drop symmetrical (or near-symmetrical) design. Retrofocus lens distortion can be corrected in the computer (although a bit messy if you forgot your shift settings), and with retrofocus you get lower vignetting and larger aperture, ie better for live view focusing. With that view there's nothing wrong with the current Dalsa sensors, it's just SK lenses that are wrong, not only the 28 and 35 but also the 43XL 47XL and 60XL which are not retrofocus enough for their larger image circles.

Anyway, for myself and many others in a similar situation like mine, ie overall pleased with the SK lenses and their design and performance (cost & weight are factors too), but using an old discontinued back which won't be supported forever, I see now through Hasselblad's 50 megapixel Kodak sensor backs an upgrade path that buys us quite some time. In the best of worlds it will stretch far enough into the future so we get that new sensor technology that can keep tech cams vital and unique. I've said it before and I can say it again, I rather not want to see tech cams become just an enlarged Sony A7r with a little shift. I think having sensor technology that allows unique tech cam lens designs is a key factor in separating the camera type from the rest.
Logged

Paul2660

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4066
    • Photos of Arkansas
Re: Schneider Digitar Question
« Reply #18 on: September 04, 2014, 07:14:48 am »

I hope that Phase One will not be planning on releasing any more base ISO optimal backs i.e CCD.

Take a sec and go back to the Doug Peterson testing in the library. Even at base ISO the IQ260 could not handle any push without excessive noise. The IQ250 had much less aliasing also.  Right now with I am working with ambient outdoors temps in the 90 to 100 F degree range and at ISO 100 a IQ260 produces more noise than I am comfortable with especially on shifts.  

To me, one of the greatest opportunities Phase One has, is to work out a software solution that can negate the effects of crosstalk. Sadly they don't seem very interested in this.  Such a fix would give some the Schneider lenses new life on CMOS.  It may something Phase is working on and may come out in some form of LCC algorithm.  Such a solution would enable better shifts on both CCD and CMOS.

Torger has shown in his work just how much of an issue that crosstalk is with current MFD CMOS chips from Sony and Dalsa.

Paul



 
« Last Edit: September 04, 2014, 07:29:02 am by Paul2660 »
Logged
Paul Caldwell
Little Rock, Arkansas U.S.
www.photosofarkansas.com

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Schneider Digitar Question
« Reply #19 on: September 04, 2014, 08:56:36 am »

Unfortunately I have to be pessimistic about software solution to crosstalk. I have worked hard to achieve results, and I do achieve them when there are medium levels of crosstalk.

However, SK28 shifted to the edge with a Dalsa 6um CCD, it's just game over man :-\. The core of the problem is that to calculate the next pixel you need to know the actual value of the previous, and you get a long chain of values which depend on eachother. With strong levels of crosstalk this chain becomes very unstable, and the microlens ripple does not exactly help :). It might still be possible to make some rougher approximative solution to the problem (haven't given up totally yet), but I doubt it.

With the Sony MF CMOS I haven't tried as much as with the Dalsa 6um CCD, but thanks to a helpful user I've had a fair bit of experimentation with it, and this sensor under crosstalk is behaving non-uniform to an excessive amount, which makes it hard to succeed with an algorithm. In other aspects it behaves better than the Dalsa 6um though, it has no microlens ripple and minimal to no tiling which helps. I think I would have to do some specific algorithm design to this specific sensor type to succeed though, if possible. Phase One engineers should have an advantage by having access to how exactly the Sony CMOS is laid out (CMOS sensors are more complex structurally than CCDs), and that way make it easier to target an algorithm for it.

I have succeeded correcting crosstalk for 7.2um Dalsa (33 megapixel), which indeed has crosstalk with visible desaturation on extreme shifts on SK35 and even the 60XL but to a much milder extent. The commercial value of that is not that big. Probably the algorithm works for the Dalas 6um with less crosstalk than on a SK28, but I haven't had time to make tests for that yet.

The Sony CMOS has so great dynamic range and low noise, so possibly it would work out just by adding proper pixel shielding to the sensor, ie you would get pixel vignetting instead of crosstalk as the Kodak sensor on the Hasselblads. It would require a new sensor though, but probably one that can be manufactured with today's technology. I think you can keep the microlenses, since Sony's type don't cause any ripple anyway (not sure though if they can contribute to crosstalk).

If I would guess I don't think the reason the Sony CMOS crosstalks this much is because it's hard to include light shields around the pixels, but simply because it was deemed unnecessary in the design process. The sensor designers don't look at the needs of the tech cam segment at all, on the whole it's tiny. So even if the problem is solvable with current manufacturing technology simply by adding light shields (like in the Kodak sensor) I don't find it likely to happen, instead I find it more likely that we can get wide angular sensor response as a side effect of technology improvements which is desired for other reasons, eg back illumination as it improves sensitivity and coverage.
« Last Edit: September 04, 2014, 08:59:41 am by torger »
Logged
Pages: [1] 2   Go Up