Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: jerome_m on October 13, 2014, 12:09:55 pm
-
Here an image taken with the HCD 24 and the HTS on an H4D-50. 11.5mm shift to the top. There are some elements on the top right corner for your pixel peeping pleasure.
(https://farm4.staticflickr.com/3942/15493639876_1a15c1f8e6_b.jpg) (https://flic.kr/p/pB7Xm1)
Infineon Campeon technology park, Munich. (https://flic.kr/p/pB7Xm1)
Full resolution here (https://www.flickr.com/photos/jerome_munich/15493639876/sizes/o/).
-
Here an image taken with the HCD 24 and the HTS on an H4D-50. 11.5mm shift to the top. There are some elements on the top right corner for your pixel peeping pleasure.
(https://farm4.staticflickr.com/3942/15493639876_1a15c1f8e6_b.jpg) (https://flic.kr/p/pB7Xm1)
Infineon Campeon technology park, Munich. (https://flic.kr/p/pB7Xm1)
Full resolution here (https://www.flickr.com/photos/jerome_munich/15493639876/sizes/o/).
Image quality looks good. Although at $12,500 the 24mm lens and HTS combination is well into tech camera price territory. For less money one can get a tech camera body, adapter plate for back, cables and a stunning 35mm Schneider XL Digitar lens. That combination will produce much better results.
Of course the 24mm lens by itself is much wider and the HTS might be more convenient to use rather than a different camera system for a Hasselblad H system owner.
-
Nice results.
The 24mm becomes an effective 35mm with the HTS, right?
-
The 24mm becomes an effective 35mm with the HTS, right?
Yes, it does. The HTS includes a x1.5 converter.
-
Image quality looks good. Although at $12,500 the 24mm lens and HTS combination is well into tech camera price territory. For less money one can get a tech camera body, adapter plate for back, cables and a stunning 35mm Schneider XL Digitar lens. That combination will produce much better results.
I am not so sure that the Schneider would produce "much better" results. A quick search on this forum and the competing getdpi forum shows that Schneider Digitar 35 users complain about fuzzy corners when shifted 10mm, magenta shifts, enormous falloff only partially mitigated by the additional center filter, etc... Also: the few published samples are not directly comparable, since they were taken on 39 mpix back. This is a 50 mpix back and therefore a stronger strain on the optics.
I would be interested to see a direct comparaison.
-
I am not so sure that the Schneider would produce "much better" results. A quick search on this forum and the competing getdpi forum shows that Schneider Digitar 35 users complain about fuzzy corners when shifted 10mm, magenta shifts, enormous falloff only partially mitigated by the additional center filter, etc... Also: the few published samples are not directly comparable, since they were taken on 39 mpix back. This is a 50 mpix back and therefore a stronger strain on the optics.
I would be interested to see a direct comparaison.
The SK35 got a bit bad reputation because Dalsa 6um sensors don't handle it well for larger shifts. The Kodak sensors does it better, so Hasselblad 39 and 50 megapixel backs should work very well with the sensor concerning color cast, but all tech wides does require LCC correction. It should not be any bad cast with the Kodaks.
Concerning sharpness the comparison again is Rodenstock Digaron-W, often the 32, which is indeed on yet a sharper level than the SK35 when it comes to shifted positions. On center the SK35 is extremely sharp.
Tech camera users are generally the most insane people you can find concerning being picky about sharpness, so don't take it too seriously when you hear us complain without seeing an example first.
On Dalsa 6um or smaller (IQ160/IQ180) the SK35 should not be used with large shifts at all (too severe casts/crosstalk), on larger pixel sensor and Kodak 50 megapixel you can use it unrestricted in the whole 90mm image circle, but you want center filter. On center and small shifts f/11 is the optimal aperture and then it's as sharp as any Digaron. For large shifts if corner sharpness is important to you, you should shoot at f/16. Added bonus with the symmetrical design is that it's almost distortion-free.
I've attached a 100% crop of a detail from my 7.2um pixel back (33 megapixel, that is a bit lower res than the 6um 50 megapixel in your test) with the SK35 shifted about 12mm, shot at f/16. I could have sharpened more, the render is just a conservative standard setting out of RawTherapee. You do see some trace of chromatic abberation on the cross and it's a little blurrier in the center, but still quite okay. Note that the difference between 7.2um and 6um pixels is not too large when you pixel peep, only 20% difference along the side. You can find SK35 samples with 6um pixels on the web, problem is that I don't think I've found a single one that is shot at something else than f/11 and a Dalsa, and for larger shifts and if you value corner sharpness it's better to step up to f/16, and of course Dalsa 6um get issues with crosstalk unlike the Kodak 6um.
Added a second crop which is from a position which corresponds to edge of 20 mm shift up of a horizontal 49x37mm sensor. Here the picture is starting to get a bit worried and smeared.
A more direct comparison to the HTS would of course be better, with some very fine detailed high contrast structure at the shifted edges. I'm quite sure that the SK35 will be markedly better and also with lower distortion, but if it matters in real image making is another question. If you predominantly shoot with the H system it can be nice to stay with one system. I do have samples from an H3DII-50 shot on SK35 and SK28 with extreme shifts and the result is in line with my own crops, but as it's not my pictures I can't publish crops from that. Hasselblad 50 megapixel backs are unique in that by being both high MP and playing well with the two SK wides 28 and 35 with larger shifts, and compared to other MF wide angle solutions it's probably the best price/performance you can come by.
The most demanding tech users would use the Rodenstock Digaron-W 32 which performs markedly better than the SK35 concerning sharpness in shifted positions, but is on the other hand much more expensive and not distortion-free.
-
I guess we would really need a direct comparison to know for sure how large the difference between the lenses is. The HTS system has the advantage that all corrections are automatic: one drops the raw image in Phocus and that program gets the lens type, the amount of tilt and/or shift and directly corrects distortion, vignetting and chromatic aberration.
-
Yes it would indeed be an interesting test. Tech cams due to the lack of exif integration don't have any automatic lens corrections. The HTS will surely narrow the gap due to that.
-
Yes it would indeed be an interesting test. Tech cams due to the lack of exif integration don't have any automatic lens corrections. The HTS will surely narrow the gap due to that.
True.
Even so I see some nasty CA and softness in the corners in the shot posted with the 24 and HTS. My 40mm HR-W is pin sharp in those areas and clean with just an LCC.
-
My 40mm HR-W is pin sharp in those areas and clean with just an LCC.
Could you post some samples, please?
-
Could you post some samples, please?
Ok, Il look up some files tonight when I get back. I have mostly landscape shots but think I have some with buildings.
-
Could not find a shot with buildings using the lens shifted. But here is a straight on shot using the 40mm HR-W on the Arca with the IQ160. No LCC applied (did not do one for this shot). Also shown are crops from the left edge and bottom left corner. No CA correction. Pretty minimal adjustments and a touch of sharpening.
-
Ken,
You for sure have a good version of the 40mm HR-W, I am surprised you got that shot without an LCC. My 40 HR-W looks about the same without shift, however on 12mm to 15mm of shift I do get a bit of CA, which surprised me a bit. However it's very easy to correct in either LR or C1. I am using the 260 or 160, as I have shots taken with the 40 from both.
As you point out, great all around lens, and so much easier to work with than the 32 HR-W. Somewhere in the back of my mind, I still want a 32, but it's a beast.
Paul
-
I can't see the benifit of using the 24 with the HTS. I see no reason why Hasselblad can't make a tech body one designed to slap on their backs and provide lens boards. The HTS is so limiting with its 1.5x factor for many situations. Also make lens converters to use other manufactures lens. If Hasselblad could follow some of the camera manufactures road maps with equipment introductions based on this century innovations the company might be self sufficient and not need outside funding. After all it's about making money?.... Isn't it?
-
here is a straight on shot using the 40mm HR-W on the Arca with the IQ160.
We would need a shot with about the same equivalent of shift for a better comparison. Straight shots on the HTS don't show much CA either.
-
Here an image taken with the HCD 24 and the HTS on an H4D-50. 11.5mm shift to the top. There are some elements on the top right corner for your pixel peeping pleasure.
(https://farm4.staticflickr.com/3942/15493639876_1a15c1f8e6_b.jpg) (https://flic.kr/p/pB7Xm1)
Infineon Campeon technology park, Munich. (https://flic.kr/p/pB7Xm1)
Full resolution here (https://www.flickr.com/photos/jerome_munich/15493639876/sizes/o/).
Do you have any examples with this set up of linear subjects at a much greater range of distances from the camera? That is, I would like to see do what degree straight lines are rendered as straight at various distances from the camera, in the same shot.
-
Lens corrections can do much about chromatic aberrations and distortion, but it can't do that much concerning general sharpness. If the lens smears the result will be smeared. The top left corner of the building in the example has quite soft low contrast surface which is nice on the lens even if it smears a bit.
I think the HTS is a nice and interesting concept, the only problem I have with it is that it costs a large amount of money and I don't think the price level matches the image quality expectation at least I have for an optical system at that cost. From an it's-practical-to-use-for-someone-already-owning-an-H-system standpoint it's entirely logical though, and the absolute quality is certainly not bad. I guess a tech cam lens that provides comparable quality and field of view would be the Rodenstock Apo-Sironar 35mm, which costs $2.2k.
-
Do you have any examples with this set up of linear subjects at a much greater range of distances from the camera? That is, I would like to see to what degree straight lines are rendered as straight at various distances from the camera, in the same shot.
No, I did not take this kind of image during that test. But there is something you should realise: that image has seen its distortion corrected by Phocus. It is a standard feature: the HCD 24 and HCD 28 lenses have a noticeable amount of distortion (with or without HTS). Phocus uses the info from the exifs (which includes shift and tilt for the HTS) to correct that distortion. The result, as far as I can tell, is straight lines at any distance.
-
From an it's-practical-to-use-for-someone-already-owning-an-H-system standpoint it's entirely logical though
It is even more logical if you can rent it. It is much easier to rent an HTS or an HCD 24 lens where I live than a complete view camera. For somebody who has an H camera and an HCD 28 lens (a very useful lens on its own), just renting the HTS for an architecture project is a financially attractive proposal.
-
No, I did not take this kind of image during that test. But there is something you should realise: that image has seen its distortion corrected by Phocus. It is a standard feature: the HCD 24 and HCD 28 lenses have a noticeable amount of distortion (with or without HTS). Phocus uses the info from the exifs (which includes shift and tilt for the HTS) to correct that distortion. The result, as far as I can tell, is straight lines at any distance.
Rectilinear distortion varies with the distance from the subject to the camera, and there is only so much software can do when there are straight lines at very different distances from the camera. This is why architectural photographers tend to prefer lenses with inherently low level of rectilinear distortion, regardless of the ability of software to correct for some of the distortion. Essentially what I am asking is "what is the rectilinear distortion profile of this combination before applying any software correction?"
-
As an aside I have a pizza client and they want hires images of pizzas sharp from front to back. I use the HTS with an 80mm and 7 degrees of tilt and it works well. My point being it's not just for architecture.
Nick-T
-
Rectilinear distortion varies with the distance from the subject to the camera, and there is only so much software can do when there are straight lines at very different distances from the camera.
I don't think that this statement is quite true. Distortion varies with the position of the focus ring, yes. But not directly with the distance between subject and camera. Of course, indirectly, the position of the focus ring is linked to the subject distance, but I think that distortion is the same for out of focus elements than for the focus plane.
Phocus, the software from Hasselblad, reads the position of the focus ring from the exifs. It uses it to correct distortion. In my experience, the results are great with straight lines looking straight on all subjects. Basically, when you use Phocus, distortion is non-existent for all H lenses.
-
As an aside I have a pizza client and they want hires images of pizzas sharp from front to back. I use the HTS with an 80mm and 7 degrees of tilt and it works well. My point being it's not just for architecture.
Yes, indeed. I would even say that the HTS has been originally designed with this kind of use in mind and not for architecture.
-
Yes, indeed. I would even say that the HTS has been originally designed with this kind of use in mind and not for architecture.
The fact is, though, a significant minority of architectural photographers who use medium format are using Hassleblads, whether with HTS 1.4 or not I am not sure. I would be interested to know why they are doing that instead of using "tech" cameras.
-
I have no horse in this race, but I recall a series of videos a german magazine made pitting a phase one kit against a Hasselblad kit. Their conclusion was that the Hasselblad had the edge in the studio because of Trufocus, but the Phase on a tech cam did better for architecture than the blad with HTS.
I don't think this is necessarily a bad thing. If you look at the blad ecosystem, it is very much geared towards a studio user. Things like Trufocus and tight integration of body and back are things a studio person appreciates and a tech user couldn't care less about. The displays on the backs haven't been updated because they probably feel that these cameras would almost always be used tethered. Their promotional materials almost always feature advertising/ fashion portraiture and very rarely architecture and landscape. I believe the HTS was developed as an ingenious (And relatively low cost) solution for T/S for a Blad user who has invested into their ecosystem. Again, more for product/ studio work rather than field work.
I believe that Blad has chosen the niche they want to focus on and tot heir credit, they have come up with some unique solutions that work really well for that market. You can make them run another race, but it's not what they focus on and for sure, the results won't be as good as brands/ products that are specialized for that task.
-
I don't think that this statement is quite true. Distortion varies with the position of the focus ring, yes. But not directly with the distance between subject and camera. Of course, indirectly, the position of the focus ring is linked to the subject distance, but I think that distortion is the same for out of focus elements than for the focus plane.
Phocus, the software from Hasselblad, reads the position of the focus ring from the exifs. It uses it to correct distortion. In my experience, the results are great with straight lines looking straight on all subjects. Basically, when you use Phocus, distortion is non-existent for all H lenses.
See figure 5 on this Web page: http://toothwalker.org/optics/distortion.html
An automatic software lens profile correction cannot deal with this situation. Admittedly, however, the example represents an extreme case, with the ruler held very close to the lens.
-
See figure 5 on this Web page: http://toothwalker.org/optics/distortion.html
An automatic software lens profile correction cannot deal with this situation.
The distortion in that fig. 5 is similar to the one exhibited by the HCD 24 or HCD 28 lenses. Phocus corrects it automatically. The correction system in Phocus is much more powerful than you appear to think and includes thousands of data tables directly computed from the lens ray tracing software.
-
The distortion in that fig. 5 is similar to the one exhibited by the HCD 24 or HCD 28 lenses. Phocus corrects it automatically. The correction system in Phocus is much more powerful than you appear to think and includes thousands of data tables directly computed from the lens ray tracing software.
You can't seem to wrap your mind around what I am talking about. I think maybe we went over this before somewhere else and you still didn't understand. I can't try any harder to explain. If I am wrong, I would expect the Hassleblad rep to correct me immediately, since he seems quite active in this forum.
-
You can't seem to wrap your mind around what I am talking about. I think maybe we went over this before somewhere else and you still didn't understand. I can't try any harder to explain.
Obviously, yes: I don't understand your concerns about distortion. I have used the HCD 28 extensively. It has a fair amount of "moustache" distortion, uncorrected. Corrected, I have never seen it produce anything else than straight lines. Let me add an example, since this thread lacks pictures:
(https://farm9.staticflickr.com/8262/8624208423_fca2f8dc4a_b.jpg) (https://flic.kr/p/e96jGF)
Tree twig and glass wall. (https://flic.kr/p/e96jGF)
Maybe you are not convinced because everything is at the same distance on the picture above? Let me try something else:
(https://farm9.staticflickr.com/8542/8624210245_cf8aa51f20_b.jpg) (https://flic.kr/p/e96kf6)
Art piece without self portrait. (https://flic.kr/p/e96kf6)
Or, maybe, this one. You are welcome to align a ruler to the floor tiles:
(https://farm9.staticflickr.com/8247/8625316390_894eb9f7a5_b.jpg) (https://flic.kr/p/e9c14y)
Underground passageway perspective. (https://flic.kr/p/e9c14y)
If I am wrong, I would expect the Hassleblad rep to correct me immediately, since he seems quite active in this forum.
I have not noticed a "quite active" Hasselblad representative in this forum.
-
Obviously, yes: I don't understand your concerns about distortion. I have used the HCD 28 extensively. It has a fair amount of "moustache" distortion, uncorrected. Corrected, I have never seen it produce anything else than straight lines. Let me add an example, since this thread lacks pictures:
(https://farm9.staticflickr.com/8262/8624208423_fca2f8dc4a_b.jpg) (https://flic.kr/p/e96jGF)
Tree twig and glass wall. (https://flic.kr/p/e96jGF)
Maybe you are not convinced because everything is at the same distance on the picture above? Let me try something else:
(https://farm9.staticflickr.com/8542/8624210245_cf8aa51f20_b.jpg) (https://flic.kr/p/e96kf6)
Art piece without self portrait. (https://flic.kr/p/e96kf6)
Or, maybe, this one. You are welcome to align a ruler to the floor tiles:
(https://farm9.staticflickr.com/8247/8625316390_894eb9f7a5_b.jpg) (https://flic.kr/p/e9c14y)
Underground passageway perspective. (https://flic.kr/p/e9c14y)
I have not noticed a "quite active" Hasselblad representative in this forum.
You second and third examples are in the range of what I am talking about. They are not ideal for the purpose, however. Nevertheless, if you look at the bottom of the underground passage photo, you will see pincushion distortion, which also appears to a lesser degree at the extreme sides, whereas further away in the scene there does not appear to be any significant curvilinear distortion.
-
if you look at the bottom of the underground passage photo, you will see pincushion distortion, which also appears to a lesser degree at the extreme sides, whereas further away in the scene there does not appear to be any significant curvilinear distortion.
The illusion of pincushion distortion comes from the fact that the ground is not flat in that underground passage. It is designed this way so that water flows in the gutters left and right. If you look at the straight lines on the ceiling (which is flat), they look straight.
-
The requirement of 100% distortion-freeness, a classic large format advantage, is becoming extinct. The only wider lenses that does it decently today is the SK Digitar wides, and they're not as popular as the Rodenstock Digaron-W wides which have a bit better sharpness but more distortion. The trend goes towards "large format digital" becoming as distorted as any other format, and you have to rely on raw converter lens corrections.
-
The illusion of pincushion distortion comes from the fact that the ground is not flat in that underground passage. It is designed this way so that water flows in the gutters left and right. If you look at the straight lines on the ceiling (which is flat), they look straight.
This is not a good example then. Also, the lines in the ceiling that you reference are not easy to see. And, by the way, thank you for not letting us know in advance that some lines which we might assume to be straight in real life are not. But, do you know that for a fact about the floor? It makes sense, but I see no gutters or drains along the sides for collecting water. In any case, I also pointed out some pincushion at the very edges of the sides. You are not saying those lines are curved in real life, are you?
Also, it would have been helpful if you had corrected the slight vertical convergence.
-
This is not a good example then. Also, the lines in the ceiling that you reference are not easy to see. And, by the way, thank you for not letting us know in advance that some lines which we might assume to be straight in real life are not.
Obviously, I did not remember that the floor was arched, or I would not have posted that particular picture. Sorry.
The lines on the ceiling can be seen on the original picture, which you can download in high resolution by clicking on it. On the second post, I drew a red line with Photoshop and it covers the original ceiling mark completely.
In any case, I also pointed out some pincushion at the very edges of the sides. You are not saying those lines are curved in real life, are you?
I drew a red line on the right hand side of the picture above. It appears to cover the seam between the tiles perfectly.
-
I am hired to shoot a lot of Architecture projects.
I find that some of the most important things I need are extreme wide angle coverage (the 14mm L II lives on my 5D3 for most Arch projects although on occasions I use the 24mm TSE II) but be able to use longer and fast lenses for shallow dof shots, working speed (most times the client asks for a LOT of shots and I have very limited time available in the project), processing speed (most clients need the images finished right away, incredibly a simple thing like dust cleaning can really set me back but with the Canon 5D3 Canon has managed to keep the sensor incredibly clean with the sensor shake even after years of heavy use also stitching and image blending are mostly out of the question but I use them in a pinch sometimes) also good dynamic range is key. That is one area where the Canon lacks a bit but I have learned to expose the sensor to extract the most out of it and most times it is enough.
So even though I have a tech camera and a PhaseOne IQ160 I rarely if ever take it on those jobs even though the image quality is beyond comparison. The Phase has just so much more dynamic range and resolution it is unreal.
I though about using the IQ160 on a SLR system for those jobs (I have a H1) but neither the Hasselblad H system or the PhaseOne system have an SLR lens that has the angle of view comparable to a 14mm or 15mm lens on a 35mm DSLR. I would need to get the uber expensive and somewhat flare prone but awesome Rodenstock 23mm HR-W or one of the (also very expensive) FPS solutions that will allow the use of my 24mm TS-E with the back (might be wide enough).
but there is still the speed issue...
If I had the time to work then the Arca RM3di I have is just about the perfect camera for Architecture. Even with the 40mm HR-W I have I could stitch images to gain precious horizontal angle of view and more.
All that said, the HTS is a viable alternative although the 1.5x crop is somewhat of a dealbreaker for me at least.
-
+1
1.5x is the problem
-
It seems that many people want extra wide angles for architecture. I am not so sure about that: I find that the 24mm on the HTS already exaggerates the perspective too much when shifted, and the resulting focal length is only 36mm.
Also: I do not understand the advantage of tech cam on that specific point. The widest lens from Schneider Kreuznach is a 34mm, which is about the same focal length as what I get on the Hasselblad. Rodenstock has a 23mm and a 28mm Digaron lens, but they have reduced image circles and can only be shifted so much. Only the 32mm has an image circle large enough to be shifted decently.
Last but not least, I tried to compare the MTF between a few lenses. It is not easy, one must first find curves computed for the same l/mm values and aperture, and then scale them to the same image circles. Here is the result between the HCD28 + HTS combination (in red) and Schneider Kreuznach Apo-Digitar XL 43mm (in grey), both at f/11. Values are given for 20 l/mm and 40 l/mm.