Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: EricWHiss on May 25, 2011, 04:02:33 pm
-
This is a follow up to the DR comparison post using the 4 different camera/back set-ups. While I had them here I shot a few different set-ups with each using the same lighting to get a feel for overall look (color and tonality).
The first image is of an old clock. I chose it because it has actually a lot of color variation in the face as it oxidized that was subtle but also a fair bit of detail.
Notes:
1) All cameras shot at base ISO, CFii-39MS and Phase p20 both shot with Rollei 90mm apo macro lens, Leaf Aptus on Phase DF body with 120mm macro, Canon 5D2 with very good Leica 35-70mm f/2.8
2) Because the Phase p20 is square I chose to frame the height to make a more pixels to pixels but in real world you might need to back off and crop - so not exactly fair.
3) All files processed in C1 except the CF which I did in phocus. Other than white balance which I did with X-rite passport card, nothing is necessarily done the same with the files - IOW I used my own judgment on how to handle them. The CF files had the most headroom for adjustment, but exposure is not 100% matched either, and I was not too concerned with blowing out the highlights to the same level on each. The highlight and shadows control has been dealt with in the DR comp post.
4) I'm including both single and multishot examples from CFii-39MS because it may be of interest to those that want to know if the Aptus 12 is truly a 'multishot killer" or not, but be mindful of the differences in lenses and potential focusing accuracy.
Okay now to the images - first here's the full frame - resized.
-
Wish I could attach more than 4 images...
-
Now to crops - this gold region has a lot of color and some stains in it. How much can each camera make it come alive? Hard to pick a winner but its easy to see the canon is not picking up the coloration of the clock face as well - particularly the subtle changes in the stained areas.
-
I dunno what you want to look for... Do you see that 'liquid color' others have mention in the 80mp back? The crops are 100% this time so also the level of detail captured will be available.
There are lots of variants in the processing - input color profiles and curves and custom profiles, color adjustments, etc... I used the canned profiles but did add some small curve adjustments - but don't get hung up on how they differ in color - just feel it.
Whoops here's the 1st canon crop.
-
2nd crop - is the 39 mp Multishot defeating the 80mp Goliath here? or is it just differences in lenses and focus accuracy?
-
Sorry I don't mean to suggest the canon is left behind the others - its just the order I started and LuLa only lets me attach 4 images per reply.
-
3rd crop area to show dark region
-
Is this the way you would like to see tonality comparisons? What do you think about the different cameras?
I myself was surprised how much detail the CF-ii 39MS caught and overall on the downsized image how much they all looked alike.
-
A very important aspect when testing cameras this way I think is to expose manually and really max out the histograms (expose to the right), because this is what most of us do when shooting base ISO from a tripod to maximize image quality. Any camera that has not its histogram maxed out is at disadvantage, it will show less dynamic range than it is capable of.
(Sometimes the histograms are too poorly implemented in the camera to be trusted so one need to check the actual file in suitable software if the brightest point is at saturation.)
Perhaps this is exactly what you've done but I don't see you mention that in the method description.
-
The separate DR thread is here:http://www.luminous-landscape.com/forum/index.php?topic=54348.0
-
Wow, nice work, Eric!
That multishot is amazing. Now I know why you're looking for that CF-528! :)
To me, the multishot followed by the Leaf stand out for detail.
Thanks, this is a lot of work, and I appreciate your sharing it with us.
-
I think a bit of that detail in the CF images probably comes from the Rollei 90mm apo macro. If you look at the thread I posted the shot of the flower taken with that lens and the AFi-ii 12, you also see impressive detail.
-
Properly sharpened, not easy to tell apart the differences between singleshot and multishot.
(http://copiasxl.com/descargas/singleshotvsmultishot.gif)
-
Hm.. I am very suppriced how small the differenses are between these backs.
But the lens used with the Hasselblad-back is normally a lot better than the one you use with the Leaf, isn't it?
Henrik
-
I think your test clearly reveals that MF is a lot better than 35mm with regard to colour rendering and sharpness but more surprisingly how little difference in the quality (apart from resolution) there is between a MF digi back bought in 2004 and one bought in 2011. :o
The same can't be said for 35mm I imagine. :-[
-
Hm.. I am very suppriced how small the differenses are between these backs.
But the lens used with the Hasselblad-back is normally a lot better than the one you use with the Leaf, isn't it?
Henrik
Yes, I think that's true. The Rollei / Schneider 90mm apo macro is truly in a class by itself and maybe one lens worth going to the AFi for all by itself. The Mamiya lens I used was the 120mm - which I guess is supposed to be good enough? Now that I have the AFi, I could retest the CFii-39MS and AFi-ii 12 and see only differences in the back.
-
Properly sharpened, not easy to tell apart the differences between singleshot and multishot.
Maybe with this crop, but having worked with the MS backs, I can tell you the differences are subtle and sometimes seen not close up at 100% but further back too. I can only describe the differences as this - single shots capture detail, and MS shots capture texture and feel. I shot $2 bills with the Aptus 12 and CFii-39MS. The aptus has twice the pixels and got a very tiny bit more detail, but missed the fingerprint mark on the bill and the crumpled wrinkles in the bill that the multi-shot got. Since Yair and I have gone back and forth about Multi-shot killers, I'll have to post that thread because its just not true. Multishot is still king of the hill at least in my testing.
-
When I looked at the 50MS comparing single and MS acquisitions, I was very impressed with the "better" look of images taken with MS. To my eyes, a key benefit of MS is noise reduction, visible particularly as subtle intensity transitions in dark areas but also a clean edge transitions. Everything is just smoother and that gives images a sense of naturalness. Add to that better color rendition and those were the best quality images I have made. The 200MS must be a wondrous thing.
I did not have time to try something. I would like to average four and nine single shot images and compare them with a four-shot MS. I wonder if averaging would provide at least some of the noise reduction benefit of MS, without requiring quite so much perfection in exposure. Eric, feel like trying this with the 39?
Peter
-
Peter,
That's an interesting idea and I guess I could try it. But you'll have to guide me on what software is available to average the data. I think the advantage of the Multi-shot is actually more that it gets R,G,B data for each pixel location and so the image does not need to go through a bayer interpretation. In a way this is higher resolution not just cleaner.
I'll post some images with very fine detail shot both with the Aptus 12 and the CFii-39MS soon.
Eric
-
Peter,
That's an interesting idea and I guess I could try it. But you'll have to guide me on what software is available to average the data. I think the advantage of the Multi-shot is actually more that it gets R,G,B data for each pixel location and so the image does not need to go through a bayer interpretation. In a way this is higher resolution not just cleaner.
Eric, for the few images I averaged I just used the average function in batch processing under Photomatix Pro. Apparently, it is simple to average layered images in PS as well, but I do not know how.
I fully agree about the improved resolution with MS, but to me there appears to be a lovely pixel to pixel smoothness as well. Perhaps that is just a function of higher spatial resolution but I hope noise reduction through averaging could fake some of the effect. Sadly, not much of my work allows the tight light control that MS needs so I look for ways to get part way there.
-
The MS shots are impressive.
Stupid question: The MS mode is like a scanning back and requires continuous lighting, right?
-
Not a bad question. These images were shot with my profoto strobes, but you can use all kinds of lighting provided its consistent enough from shot to shot. I've shot with my metz camera mount flashes, daylight, etc. I think fluorescent is okay if the exposure time is long enough to average out the fluctuations in brightness, but that is the only kind of lighting where I got errors.
-
Interesting tests. I'm not surprised that the difference in resolution from different backs is so minimal.
I recently tested various backs from Leaf, Sinar and Hasselblad against my trusted Leaf Aptus 65.
I really couldn't see much difference between them, they are all so bloody good that I would be happy to own any of them! If you are new to camera backs, you will have great difficulty in deciding and your decision could be made on points such as practicality, a good demo, whether you like and believe what the rep says, whether you like the software and so on, rather than the ultimate quality. I notice that reps are not bashing the competition as much as they used to, although some still do, in subtle ways. There is an increased acceptance by reps that the competition is very good, so reps score brownie points by being friendly, helpful and honest with well prepared answers that don't sound like waffle.
The needs of a still life photographer shooting tethered in the studio are different from those of a people's photographer shooting untethered under diverse light conditions, etc.
There is no way you could see the tiny, tiny differences on printed material, or offset printing, other than small variations in colour due to differing colour profiles. What's more, I didn't see any of the current backs were any better than my 5 year old Aptus 65, apart from file size, although I have to say that they were all tested at base ISO, as I'm not interested in higher ISO's. If you wanted to push me really hard for a superior file, I would probably say that in fairness, the 50MP Hasselblad multishot had the edge on certain detailed subjects, but only a tiny edge at that, something that may be of great importance to pixel peepers. But I am looking at practicality overall and as I say, once you go into print, they all become equal, although for big interpolations of files, you are naturally better off starting with a bigger file.
Much as I wanted to buy a better upgrade to minimise my tax burden, I came to the conclusion that I should stick to what I have, at least for the time being. The major difference between these backs is in the software, and how practical it is for the way you work.
My conclusion is that without a doubt, the Leaf Live View is still the best of all the ones I've seen, unquestionably, although it is still pretty poor on all medium format backs. And Live View is of major importance to me to be able to focus quickly, accurately and with certainty. Other photographers couldn't care less about Live View.
Also, to me, being able to work directly in Lightroom (via a hotfolder) is of great importance, as I prefer Lightroom to any other manufacturers software. Both Leaf and Sinar allow you to import files automatically to Lightroom via a hotfolder. I was told that Hasselblad is working on it, but no dates yet.
Edward
-
Hi Edward,
In your type of work, the great equaliser I think is the fact that you often have to stop down well into the diffraction-limiting aperture range.
Your current backs, at 7.2µ, still cope well at f16 or even f22 however the backs with the smaller pixels, while still "seeing" more and producing larger files, do not appear to be sharper.
A similar test, done nearer to the lens's "sweet spot" may lead to different conclusions but of course you will then have to deal with limited DOF...
-
Yair,
As you know, All my tests were done at f16 using the Schneider 120 Macro Digitar. I could work at say, f8, and maybe I would see some difference, I certainly don't see any difference in sharpness in my Aptus 65 opened at f8, (other than less DOF) though, as you point out, the pixels on my back are bigger. When I first bought this wonderful macro lens, I tested it at different apertures and I came to the conclusion that f16 is good, but at f22 you start to see a tiny bit of diffraction (slight overall blurriness). I quite agree that on smaller pixel backs, diffraction is going to be more of an issue at such small f-stops.
However, shooting regularly at f8 would require a different approach to the way I've always worked, i.e. more stack shots to be joined to increase the DOF, something to consider if the end result warrants it. More testing, aargh!
Edward
-
Hi Edward, looking at your work I really can't imagine, apart from larger output files, how a new back could possibly improve your work as it is already top draw. As you shoot a lot of Jeweley that probably goes through a lot of retouch work I would image much of the gains in detail and colour depth of the newer chips would be lost.
MFD has improved slowly over the years but I feel most of these improvement are appreciated by architectural and landscape photographers who upgrade to the newest sensors. When digi backs get REAL live view I think still life photographers will feel the same appreciation.
-
Definitely at smaller f/stops you can see the diffraction effects become very pronounced with the Aptus 12. For these comparison images in this thread f/13 was used. I did test the whole range vs the CFii-39MS and will post that on another thread.
If you are shooting table top stuff and need to shoot with small apertures then that's the type of work where mulitshot backs really sing.
-
Hi Eric,
really nice comparrison, thanks for sharing.
Thats what I was looking for.
I think the multishots do stand out, but the differnces are kind of small. Even to the canon 5D.
The Leaf shots are kind of disapointing.
But I guess a few clicks in Photoshop and most differences will dissapear....
Was thinking about multishot, but handling and cables and electric shutters and stuff makes it harder to use. Especially on a viewcamera, cause I need shift and tilt.
So I guess price, handling and lenses one already are the factors for finding a decision.
-
Hi Eric,
really nice comparrison, thanks for sharing.
Thats what I was looking for.
I think the multishots do stand out, but the differnces are kind of small. Even to the canon 5D.
The Leaf shots are kind of disapointing.
But I guess a few clicks in Photoshop and most differences will dissapear....
Was thinking about multishot, but handling and cables and electric shutters and stuff makes it harder to use. Especially on a viewcamera, cause I need shift and tilt.
So I guess price, handling and lenses one already are the factors for finding a decision.
Josef,
I don't think the leaf files are disappointing at all, but do think that the backs walk away from the 5d2 in this test. This is the kind of test where differences in lenses really can steal the show, and I think the Rollei 90mm apo gave a boost to the CF-39MS shots and same is true for the canon with the leica lens, even though most people saw it as being well behind. The lens used on the 5D2 is a very sharp lens, not likely you will get exactly the same results with one of the canon L lenses. If you are in the field then I could see multishot being a lot more difficult, however in the studio I don't think its a problem at all since I would be shooting tethered to a computer anyhow. Of course everyone works differently and has different applications.
Eric
-
Forgot to add that I think the race to more pixel count has gotten past the needs of most photographers and added some technical difficulties such as these diffraction effects to work around. I'm predicting that these new 80mp backs with the small sensel sites will drive the introduction of more lens solutions with tilt/shift which would be a good thing.
-
Eric,
you are right. I guess the lenses are starting to get a probleme here. At this fine resolution only the best will do.
I didnt mean the Leaf is bad, but I was expecting a bit more.
Its just compared to the rather old P20, the difference is too small for my opinion, having the price in mind.
You can clearly see an advantage, the more money you throw in but air is getting thin up there and 15k $ more or less is worth thinking.
-
I think we have probably hit the top of what's possible in terms of image quality, both optically and in pixels. Yes, there is no doubt that those who need big enlargements are better off with the bigger files that the latest backs produce.
I used to think that packing more pixels in a given area would be the answer to my prayers, but, as we have seen, there is an optical limit as to how small the pixels can be. The 50 MP multishot Hasselblad is the best I've seen for image quality, but you can't always shoot multishot and you can't always see the difference plus there are many other considerations, such as time, movement, software processing...it's not clear cut by any means.
For most photographers, the improvements must come from IQ at higher speeds, better software design, better Live View, quicker processing, faster file transfer, etc.
It is ingrained in the photographer's psyche to seek better and better quality but there is a point where we have to be content with what we already have in terms of image quality and look for improvements elsewhere.
For my part, I'm seeking improvements in areas that would make my daily shooting experience more pleasurable. For instance, I got myself a nice Foba stand, an absolute pleasure to operate. Now I'm looking at an Arca Swiss camera, better studio furniture, classier frames on the walls, a tidier and more discerning work schedule, learning to say 'NO", more holidays (good for thinking or giving your mind a rest) etc. the little things that are conducive to more creativity or at the very least, a better quality of life...
Edward
-
Agreed. I do see the higher pixel counts as mostly just held the IQ while adding pixels and other features which is a technical achievement considering the lower well capacity etc. I hear a lot about improved color but I don't know how to really observe that. The live view on the aptus 12 is pretty nice, certainly better than the CFii-39MS.
What I'd like to see is a big square sensor 56mm x 56mm with 9um pixels, CMOS or CDD, that had higher DR. Of course I am probably dreaming!
-
What I'd like to see is a big square sensor 56mm x 56mm with 9um pixels, CMOS or CDD, that had higher DR. Of course I am probably dreaming!
Eric
Now you're talking! But sadly, it probably is only a dream. And I would like such a thing in monchrome only too, please.
John
-
All sensor sizes are starting to meet diffraction. Just few years ago people only thought about lens limitations coping with this megapixel sensors. All companies have vastly improved lens drawing in all aspects. But now, diffraction seems to be the Kraken. I think its time for bigger sensors. I'd say true 56X42mm are just around the corner. Why not 6X6? I see 6X7 sensors on the streets in 10 years.
Digital is clearly superior to film now. Unfortunately digital brought its own disadvantages and complexities. It is time now, to go back to simplicity and bug-free photography like it was before. We all learned that the f-stop was our best friend there. Now, it is a foe.
Many photogs complain that they don't want more pixels but truth is they're selling quite well. It's time for bigger sensors. It is time for photography to go back to basics. Time to frame and shoot. It's overdue.
Eduardo
Agreed. I do see the higher pixel counts as mostly just held the IQ while adding pixels and other features which is a technical achievement considering the lower well capacity etc. I hear a lot about improved color but I don't know how to really observe that. The live view on the aptus 12 is pretty nice, certainly better than the CFii-39MS.
What I'd like to see is a big square sensor 56mm x 56mm with 9um pixels, CMOS or CDD, that had higher DR. Of course I am probably dreaming!
-
Concerning diffraction -- if you do not want less depth of field there is no gain in increasing the sensor area. Larger sensor -> longer focal length -> shorter Dof. Some easy mathematics show that to achieve the same angle of view and DoF with a larger sensor you need a smaller aperture such that you will get the exact same problem with diffraction.
For practical reasons a larger sensor format may make sharper images anyway (less resolving power required from lenses, possibly easier-to-design focal lengths etc), but diffraction is not the issue. With higher megapixel count comes the diffraction problem regardless of sensor size, if DoF and angle of view is kept the same.
Diffraction can be fought to some extent with deconvolution, but a problem with that process is that is disturbed by noise (and also increases noise in the final picture).
With those new 80 megapixel backs it must be a tough challenge to make those near-far everything-is-sharp compositions while still making use of the full resolution. I guess focus stacking will become more popular.
Small pixels/sensels are not necessarily bad either. 4 um sensels need only 1/4 full well capacity of a 8 um sensel for the same dynamic range since they will get 1/4 the light. There's no law of physics that makes small pixels worse than large if you take into account that all pixels together form the image (yes more photon shot noise *per pixel*, but when combined it is the same), but traditionally due to limits in current technology the larger pixels have been better. This is evening out though.
-
Yes, deconvolution too! Maybe all the MFDB makers will also introduce some kind of routine in their RAW handling files that will pick up the aperture from the EXIF. It seems like that would be useful and relatively easy.
-
Having a 2nd look at this and used LC 11.5 to process the Leaf Aptus 12 file. I think more detail was released from the MOS file, but it may also be oversharpened. Anyhow thought it was worth posting this.
-
Can I add my thanks to that of others for the work involved and your bravery in posting any comparison without a five page disclaimer. Also as a P20 owner for making me very happy and much less upgrade inclined. Looks like I can improve my output by spending the budget elsewhere than on back hardware, for a while ;D
-
Henrik,
The lenses from Schneider and Rodenstock actually are much better than those of Canon and Nikon and equal if not better than many of the Zeiss designs made in japan.
I have a camera which allows testing of DSLR's with Schneider and Rodenstock Digital LF lenses and yhis is easily seen when viewed on computer assuming one can focus correctly.
Arca-Swiss DSLR 2.
Rod
-
I recently tested various backs from Leaf, Sinar and Hasselblad against my trusted Leaf Aptus 65.
I really couldn't see much difference between them, they are all so bloody good that I would be happy to own any of them! If you are new to camera backs, you will have great difficulty in deciding and your decision could be made on points such as practicality, a good demo, whether you like and believe what the rep says, whether you like the software and so on, rather than the ultimate quality.
The vital is to know what to look for and regrettably that is not that obvious in marketing from the MFDB makers or from reviews, e.g. as published here on LuLa. In my personal experience a demo gives very little apart from the feeling of handling the back, same as a test drive of a BMW 5 or Mercedes gives little without knowing or understanding the differences between the characters, performance and handling of these cars. The essential in my view is to prior any testing knowing what key characters differ between different backs, and which helps narrow down and see what really differs between them, and in particular so during test driving raw files, perhaps with images captured with our own set of conditions or of captures similar to our own shooting and critical conditions.
I have had the Aptus 65 which is an excellent back. The Aptus-II 10 apart from more pixels appear to me to add slight more DR and improved color performance. On other hand I find the Aptus-II 12 files to apart from pixels be a larger step from the Aptus-II 10 than that was from Aptus 65 in improvements in color profiles, DR, and the fine gradations of colors. Yair was kind to pin point me towards this, but I experienced the improvements with the Aptus-II 12 as larger than he described. It is important though to draw our own conclusions per our own test drives. Small size images do as much justice to a back as photos in a Motor Trend do of the handling of a Porsche or BMW on a twisty road...
And no, pixels apart, no DSLR measure up in optimum image quality at low ISO.
Regards
Anders
-
What I'd like to see is a big square sensor 56mm x 56mm with 9um pixels, CMOS or CDD, that had higher DR. Of course I am probably dreaming!
Yes, I'd like this one too, exactly as described, with a Hy6 / Hassy V swappable mount. Hasselblad has reasons not to make it, it seems; Sinar what are you waiting for all the time???
-
Sinar what are you waiting for all the time???
The sensor does not exist, that's the problem. And is not likely to, either, given the tiny demand there would be for it. You would be looking at a price way in excess of the Dalsa 80MP.
John