Pages: [1]   Go Down

Author Topic: Very good tech wide color stability on Hassy 50MP CCD backs  (Read 1468 times)

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Very good tech wide color stability on Hassy 50MP CCD backs
« on: January 21, 2015, 02:39:43 pm »

I've finally got my battery and just need to find a gap in my time to be able to shoot my H4D-50 outdoors. Meanwhile I've continued my lab tests to see how it performs, and tonight I made the "wide angle color stability test", which I think is a good thing to do for anyone into tech wide angle photography when they get a new back. You do like this:

1) set up a color checker under a good stable CRI light
2) mount the widest lens you have (SK35 in my case) and shoot one reference shot with the color checker in center
3) shift all you can and re-adjust so you capture the color checker far out on the image circle, make a shot with horizontal and one with vertical orientation.

If your back has low color cast you can skip LCC shots and just use a color picker for white balance the color checker, it should cover a small area so color should be reasonably stable over that local area. If you have a really colorful cast, shoot and apply LCC.

If you have relevant levels of crosstalk you will notice that at least some of the colors get desaturated. You should see it with your naked eyes, if you don't probably noone else will see it either and then there is no crosstalk or very low levels so it's irrelevant.

Even with my large pixel 33 megapixel Leaf Aptus 75 I have visible desaturation on extreme shifts of the SK35XL, in vertical orientation (not in horizontal). In real shots it's very rarely a problem of course, but I have come across it on some extreme compositions, like the one attached. There you can see that the yellow leaves in the upper part of the image is a bit desaturated and greener than the ones closer to the center of the image. The trained eye also see that the greens of the pine trees are affected too. I've also seen this effect in real images with the SK60XL as it has a larger image circle and allows for more shifting.

Now I repeated the lab test with my H4D-50 and the SK35. This sensor has light shields between pixels in both directions, and it gives results. There's no visible desaturation in either horizontal or vertical orientation at the image circle edge. I was expecting this result as I've got real shots from other users showing excellent results from the SK28XL, but now I have really made sure with my own lab test. I'm not going to put together the color checker shots for show so you just have to take my word for it :).

Although I can't say I've been limited by the Aptus 75 performance in this regard, I do see it as a plus that regardless what movements I do with the H4D-50, color will be stable. It should be said that the Aptus 75 has much much less issues than anything 6um Dalsa, I think few even knew that the Aptus 75 had any issues of this kind at all. And the H4D-50 is even more stable.

As Hassy hasn't been big on tech cams this excellent wide angle compatibility has been unknown to most. I think it's a bit unfortunate as it has also caused the popularity of the Schneider Digitar lenses drop as they haven't worked that well with the more recent Phase One's offerings.

This may be the last sensor of this kind we see for a while, so to anyone that thinks this level of performance suits them I recommend to keep this alternative in mind when evaluating alternatives. Having test files from a lot of backs I know backs based on this sensor (KAF-50100) is truly unique in combining relatively high resolution with the ability to take the whole Schneider Digitar range with its symmetrical wides with any shift amounts.
« Last Edit: January 21, 2015, 02:47:11 pm by torger »
Logged

voidshatter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 400
Re: Very good tech wide color stability on Hassy 50MP CCD backs
« Reply #1 on: January 22, 2015, 05:42:59 am »

The Kokak CCD is perhaps the best choice for technical cameras if color consistency is important for movements. I would assume that this is important for scientific purposes (besides arts). The P45+ can also do long exposure, and along with the nearly fullframe size it provides best versatility.

Personally I weight SNR (DR) and resolution to be the most important, while color consistency is of less interest as I do not shoot scientific images and color is something way a lot more simpler to deal with in post processing, while SNR and resolution are not. I can easily change the color (or leave color uncorrected to give a romantic atmosphere / unique look), but I cannot save poor shadow details or soft images out of focus (due to not having modern LiveView for critical focus).

While I worry less about color fidelity on the CMOS backs, mazing artifacts (demosaicing failure) is something like flare which can totally ruin an image. Unfortunately I do not use Mac so I cannot try your software.
Logged

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Very good tech wide color stability on Hassy 50MP CCD backs
« Reply #2 on: January 22, 2015, 06:29:47 am »

Good points, surely color is only one aspect and it depends on what priorities one have. The Kodak CCD's weak spot is indeed noise. But it is at the level that was "great!" all the way up to D800 was released, so it's not exactly bad but it will depend on your shooting and post-processing style what your needs are. If your shooting style often involves tonemapping and shadow pushing and you look more at noise than color then there's a risk that the CCD will be disappointing. In fact I think the color is often disappointing in heavily pushed areas of Sony sensors with a tendency towards monochrome brown (low read noise, but too few captured photons I guess) so I think the DR advantage is a bit exaggerated - ie you still need a good exposure.

I don't agree that it's about "scientific" color though. I would avoid tech wide angles for scientific or reproduction applications all-together. The center filter is not 100% neutral color wise, and as soon as you need LCC correction you're hurting precision a bit, white balance shift etc. I haven't used any measurement software when evaluating the results, I just look with my eyes. If you see desaturation you also get reduced color separation capability, and color is a property you can see in your artwork from any distance, therefore I value it much. Excessive noise does affect color reproduction of course, but mild noise is a pixel peep thing, the visibility further reduced if printing. So I guess I'm the other way around, I worry more about color and less about noise.

You can make local color adjustments to fix issues though, so when the issue is small like on the Aptus 75 it's not a big deal. But when it comes to the Sony CMOS it's virtually unusable on the symmetrical wides, and with Dalsa 6um you can use them but with very real limitations. So in a practical meaning I guess it's more about being able to use the Schneider Digitar lens range and having competitive resolution at the same time. (The Hassy advantage over the Aptus 75 in this context is of course not a big deal, but it is what I compare with as it's the back I've used). The advantages of the Schneider range is smaller size, lower weight, lower cost, lower distortion and more "large format like" in behavior. But sure, you do get slightly more resolving power out of the Rodenstock retrofocus wides, I just don't think it's worth it. To me large format is more about movements than resolving power, but of course that is also a personal choice.

Price is no object photographers will probably do the IQ260 and Rodenstock wides, and it does have the better image quality. But step down to a Hassy deal (which are quite frequent) of this 50MP CCD and Schneider's on the wide end you are very slightly behind in real image quality at a much lower cost, and the gear is more portable too which too me is important (I carry seven lenses into the wild...). And you never need to curse about tiling or ripple which comes up now and then with the Dalsas.

I don't think the other options are bad though, and the CFV-50c with Rodenstock 40 has many great advantages. There is no single perfect sensor today that can do it all. I'd say the Hassy/Kodak alternative is the least known and I think it can be a good choice for many.

If it only comes down to mazing it's an easy fix, I'm surprised that Capture One hasn't included it yet but probably they will eventually. You can fix mazing in RawTherapee too if you want to experiment with a windows software, use the "green equiliberation" feature. Be sure to apply flatfield (LCC) correction too.
« Last Edit: January 22, 2015, 06:47:54 am by torger »
Logged

voidshatter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 400
Re: Very good tech wide color stability on Hassy 50MP CCD backs
« Reply #3 on: January 22, 2015, 07:09:38 am »

I think this is like the worldview and religion we bear while we grew up. If I started with Canon I probably will be happy with the IQ260 + Roddie combo (which is definitely an upgrade for Canon). Unfortunately I started with the Nikon D800E so I was never happy with the IQ260. I can't keep using a medium format which has even less DR than a Nikon D4S... It is such a pain that you get obviously inferior image quality - a severe downgrade for 10x the price. I just had to make the call to switch to the IQ250 otherwise I would have sold all my medium format digital gear. Gerald Donovan seems to prefer his IQ250 than his IQ180 too. I also know lots of guys buying an IQ150 after the purchase of an IQ180, or buying a Credo50 after the purchase of a Credo 80.

With the IQ260 you would have to shoot at ISO 140 for long exposure, which produces even more shadow noise than the P45+ which can shoot at ISO 50 for long exposure. You also get tiling issues with the IQ260.

If I could choose again I would pick the P45+ (or H4D-50) along with the Schneider SK XL lenses for portability. Size and weight do matter for me, as I also prefer to work with gear that I can easily carry. Unfortunately for me I already had the IQ260 and the Roddie, and I had to switch to the IQ250 as it was the only option offered by my dealer other than selling everything.

It seems that you are optimistic about software implementation of compensating mazing artifacts. In such case, the current Roddies may even work on future fullframe CMOS sensors so I have to worry less about depreciation.
« Last Edit: January 22, 2015, 07:12:09 am by voidshatter »
Logged

Gigi

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 548
    • some work
Re: Very good tech wide color stability on Hassy 50MP CCD backs
« Reply #4 on: January 22, 2015, 08:12:36 am »

Torger -
This work you are doing is much appreciated. I don't know of much else out there taking a real world look at the combinations of sensors and lenses, and the impacts of the selections for the "mid-level" shooter. At the high end, there are good opinions, pushing the envelope, but many of us work with older, smaller, or perhaps more reasonably priced gear, and much thanks to you for pursuing this.

Might you ever be tempted to put a comparison chart together, as to which sensors work with different lens families - comparing compatibility with the new Rodies, the good small Schneiders , and perhaps analog lenses? Would be of interest. Thanks again!
Logged
Geoff

torger

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3267
Re: Very good tech wide color stability on Hassy 50MP CCD backs
« Reply #5 on: January 22, 2015, 09:09:56 am »

I can unfortunately only test with the lenses I got :-\ (I have a bunch of test files too through my role as raw software developer though, but the lens selection in those files is not that broad), but concerning analog lenses they're generally symmetric designs. Some of the Schneider Digitar designs are actually analog designs with the only difference that a tighter precision requirement was employed in manufacturing, I think the SK47XL is such a lens.

A rough guide is that symmetric designs shorter than say 50mm can start show issues on the current lineup of recent sensors when you start to employ movements. Dalsa with 6um and smaller pixels have some issues, and Sony CMOS have much issues. Older sensors with larger pixels don't have any major issues. The 50MP Kodak CCD used by Hasselblad is unique in that it can handle the symmetric wides and has a 6um pixel size, ie it's currently the highest end you can get if you want to use symmetric lenses, analog or digital, with full freedom of movements. The color cast is low compared to lots of other sensors, but you still need LCC, today there is no sensor that can do without it.

The analog designs usually provide sharp images for the longer focal lengths (say 90 - 200mm), but on the wide end they are often quite soft compared to the digital, and tele lenses I've seen some examples of are okay but not digital class.

(A quick note on long exposure on the H4D-50, while Hassy hasn't developed the long exposure technique of the P45+, they do a very good 128 sec exposure. If the back has not become warm (which it becomes when powered on for some time) the 128 sec exposure is almost indistinguishable from a short exposure in terms of noise. There's a bunch of hot pixels of course as for all CCDs, but those are automatically cleaned by Phocus so you don't see them. This was a happy surprise for me, my old Aptus does max 30 seconds and then with a noticably degraded image quality.)
« Last Edit: January 22, 2015, 09:13:52 am by torger »
Logged
Pages: [1]   Go Up