Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: ErikKaffehr on May 06, 2014, 04:42:29 pm
-
Hi,
I posted a small article called: Why I cannot see a difference in A2 size prints?
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/82-why-i-cannot-see-a-difference-in-a2-size-prints
I cannot detect a visual difference between MFD and 24 MP full frame 135 in A2 size (16.5" x23") prints. The article looks into possible explanations.
The article is a follow up/companion to this thread: http://www.luminous-landscape.com/forum/index.php?topic=87650.0
Best regards
Erik
-
Oh, come on, Erik! Everybody knows that a trained eye can spot the difference on an 8x10 print from across the room ;) ;D
-
Yes, with the right training ;-)
Best regards
Erik
Oh, come on, Erik! Everybody knows that a trained eye can spot the difference on an 8x10 print from across the room ;) ;D
-
Yes, with the right training ;-)
Best regards
Erik
If you open the MF back, grinning at you from inside will be a kobold holding a Minox.
Edmund
-
Interesting!
You are also ahead of your time: May 18th 2014 and May 22nd 2014
Markus
-
Thanks!
The time is hopefully fixed. I obviously need to fix my OS-installation on that server.
Best regards
Erik
Interesting!
You are also ahead of your time: May 18th 2014 and May 22nd 2014
Markus
-
So in conclusion, the P45+ will find a new home where its capabilities are appreciated a bit more?
If so, great work.
-
Hi,
No, my conclusion is that 24 MP are suefficient for A2 size prints.
You may consider that posting stupid, sarcastic and ignorant responses is probably of little help to anyone. You can consider just ignoring thread or postings that don't meet your demand for artful vision.
Best regards
Erik
So in conclusion, the P45+ will find a new home where its capabilities are appreciated a bit more?
If so, great work.
-
Its a good summary.
You might want to add a graph of 20/20 vision vs viewing distance for the math challenged, say out to 20 ft.
-
Hi,
No, my conclusion is that 24 MP are suefficient for A2 size prints.
You may consider that posting stupid, sarcastic and ignorant responses is probably of little help to anyone. You can consider just ignoring thread or postings that don't meet your demand for artful vision.
Best regards
Erik
eric, you don't need to reply to "he who has to use an alter ego." remember he is all knowledge, so anything you post will never be up to his level.
i for one read these posts for different thoughts and ideas that might be added to make my photography more enjoyable, but then, i am not all knowledgeable. so you and others continue to write for us photographers who are not perfect as it may benefit some of us.
-
Surely the point isn't wether or not you can see the difference, but if there is a difference does it matter?
Other than photographers does anyone bother looking for a difference. Also if you don't have super mp back to compare with side by side less than 24mp is probably plenty good enough to convey what the image is about.
If we get 150mp backs are 80mp backs just going to be not good enough for large prints. Surely the number of pimples or blades of grass you can see is not a measure of wether a photograph is good or not.
I doubt a fine artist working in oils, paints anymore detail than a 6mp camera can capture, yet viewers will marvel at the detail a fuzzy horsehair brush lays down.
To add, I help out shooting horse events, show jumping, cross country. As it's a print on site setup the photography is done as small jpg, it astonishes me the detail that gets printed in a 30x20 inch print when needed, you can see the grain in the leather and hair on the horse. It's more than enough to put a smile on the face of the purchaser. No doubt a side by side 60mp 30x20 would show even more detail, but it just doesn't matter one jot.
-
Spoilsport! :-)
Sorry :-)
Anyone ever had a client reject an image because it wasn't captured with as many pixels as they would of liked?
-
I don't argue on the issue :-)
Shooting equestrian stuff is not easy, also, nobody will care about pixels if the captured image is great.
Most of my best pictures were shot on 12 MP APS/C. I normally print A2, and at that size 12 MP APS-C holds up well.
Best regards
Erik
Surly the point isn't wether or not you can see the difference, but if there is a difference does it matter?
-
A2 is a pretty small size print. Id say 12-16mp is enough given equal everything else. A larger file most likely won't result in more printed detail mainly due to printer/paper limitations.
-
I'm using a tiny point and shoot (Sony RX100) at the moment, and the results are pretty amazing.
There has been technical progress since we embarked on this digital journey, and if we awake from our self-absorption and look out through the windows of the carriage, we should notice that the scenery has changed substantially.
All those software-corrected this, stabilized that, fast on-sensor AF, and other innovations have really changed what the low end can do compared to the high end.
I'm not saying that my RX100 replaces my old P45+, but in fact if I do 8x10 or 13" prints, I guess the results will be quite ok, by Erik's logic. I never managed to sell a big print, but had lots of individual customers for the small ones when I was still selling prints.
And the tourist cam opens up street shooting again, which I simply cannot do with the big-sensors and their big lenses. People get antsy.
So why should I blame the guy for noticing that the emperor has no clothes? It's just kind of funny and some of us who have realized it before are laughing while the newest member joins the club.
Edmund
-
Hi,
My investigation points more into human vision being the limit, and also the fact that human vision is more sensitive to low frequency detail than high frequency detail.
I would agree that differences would be visible at larger sizes if viewed close.
Best regards
Erik
A2 is a pretty small size print. Id say 12-16mp is enough given equal everything else. A larger file most likely won't result in more printed detail mainly due to printer/paper limitations.
-
Hi,
I also have an RX100 and I have printed A2 from that one, too. I did not make direct side to side comparisons. My impression was that I have noticed some weakness in that print, but I would hung it at an exhibition without the slightest doubt. But, I would not state it was from 1" sensor camera and invite photographers to discuss print sharpness.
Best regards
Erik
I'm using a tiny point and shoot (Sony RX100) at the moment, and the results are pretty amazing.
There has been technical progress since we embarked on this digital journey, and if we awake from our self-absorption and look out through the windows of the carriage, we should notice that the scenery has changed substantially.
All those software-corrected this, stabilized that, fast on-sensor AF, and other innovations have really changed what the low end can do compared to the high end.
I'm not saying that my RX100 replaces my old P45+, but in fact if I do 8x10 or 13" prints, I guess the results will be quite ok, by Erik's logic. I never managed to sell a big print, but had lots of individual customers for the small ones when I was still selling prints.
So why should I blame the guy for noticing that the emperor has no clothes? It's just kind of funny and some of us who have realized it before are laughing while the newest member joins the club.
Edmund
-
Let me ask some silly questions.
I notice that the article limits itself with 20/20 vision. It just so happens that I have 20/14 vision. If 24 mpix are sufficient for others, does it means I am allowed to use 36 mpix? I also have access to an A1 printer. Does this means that I should use a camera with 72 mpix?
It just also happens that I bought a Sony A900 with 24 mpix in early 2009. I have printed pictures taken with it in A2 size for the past 4 years and am generally pleased with the results. I use the finest Sony, Minolta and Zeiss lenses with it, but I noticed that I have difficulties getting perfectly sharp corners with wide angle lenses on that camera, while the HC28mm has no such problems. Is there anything you can suggest about this?
I notice that your study only concerns itself with flat subjects. While I have photographed historical papers for reproduction, most of my subjects are actually tridimensional, with parts which are in focus and other parts less so. I notice that the rendering of out of focus areas is very different between the lenses I have for the A900 (which include the legendary 135mm STF, Sony/Zeiss 135mm f/1.8 and other portrait lenses) and lenses from Hasselblad / Fuji (HC lenses). How can I solve that?
Last but not least, I see the Sony RX100 mentioned earlier in the thread. While the camera indeed gives results similar to the ones I get with the A900, this only happens at the widest end on the zoom range and when using iso 80. Even then, I noticed coloured fringes appearing where I do not always want them and fine details have an unnatural appearance. How do you explain that?
-
Let me ask some silly questions.
I notice that the article limits itself with 20/20 vision. It just so happens that I have 30/20 vision. If 24 mpix are sufficient for others, does it means I am allowed to use 36 mpix? I also have access to an A1 printer. Does this means that I should use a camera with 72 mpix?
Hi Jerome,
It depends on whether you keep your viewing distance constant or not ... That's why I built some user adjustable variables into my DOF output quality planning tool (http://bvdwolf.home.xs4all.nl/main/foto/dofplan/dofplan.html).
... I noticed that I have difficulties getting perfectly sharp corners with wide angle lenses on that camera, while the HC28mm has no such problems. Is there anything you can suggest about this?
Depends on the type of sharpness fall-off, which is of course lens specific. Maybe you can use a 2-layer sharpening approach, one sharpened for the center, and one for the corners, and then use a radial mask (blurred for a gradual transition). A Raw converter like Capture One Pro has a user adjustable sharpness fall-off correction built in.
I notice that your study only concerns itself with flat subjects. While I have photographed historical papers for reproduction, most of my subjects are actually tridimensional, with parts which are in focus and other parts less so. I notice that the rendering of out of focus areas is very different between the lenses I have for the A900 (which include the legendary 135mm STF, Sony/Zeiss 135mm f/1.8 and other portrait lenses) and lenses from Hasselblad / Fuji (HC lenses). How can I solve that?
The lens characteristics are a given. The only thing you can do is blur the OOF regions even more, maybe starting with a too narrow aperture to begin with if that improves initial bokeh (quality of OOF blur). When the shot is very important it would not be too objectionable to spend some time on it with a tool like Topaz Labs Lens Effects which can also use a depth map for accurate DOF transitions into the fore- / background. Nothing beats the ability to see the actual DOF effect change in real time, even after the shot has been taken.
Cheers,
Bart
-
I see that you answered my silly questions very seriously. Now I find myself almost obliged to ask some more.
It depends on whether you keep your viewing distance constant or not ... That's why I built some user adjustable variables into my DOF output quality planning tool (http://bvdwolf.home.xs4all.nl/main/foto/dofplan/dofplan.html).
Let me see: I already said that I have the Minolta/Sony 135mm STF lens. As you certainly know, this lens uses an apodisation element to shape the distribution of out of focus highlights. This has an effect on the perceived depth of field. Does your tool account for that?
Another lens I have is the 100mm f/2.0 DC lens. As you probably also know, it has a ring to adjust the amount of spherical aberration, thereby changing the apparence of the depth of field. How to I enter the value of the ring in your tool?
Portrait lenses, for example 85mm f/1.4 or faster or the HC 100mm f/2.2 have a noticeable amount of uncorrected spherical aberration. In effect, they are a bit like the DC lens, with the ring glued into the max position. This is a wanted feature by portraitists, because it gives a flattering appearance to human skin. It also increases the apparent depth of focus. How do I adjust that in your tool?
Depends on the type of sharpness fall-off, which is of course lens specific. Maybe you can use a 2-layer sharpening approach, one sharpened for the center, and one for the corners, and then use a radial mask (blurred for a gradual transition). A Raw converter like Capture One Pro has a user adjustable sharpness fall-off correction built in.
Well... at the time, I tried conversion software from DXO. It includes special routines to improve lens sharpness from supported lens/camera combinations. While these gave excellent results for distortion or lateral chromatic aberration, the far corners stayed rather fuzzy. Is Capture One Pro really better?
The lens characteristics are a given. The only thing you can do is blur the OOF regions even more, maybe starting with a too narrow aperture to begin with if that improves initial bokeh (quality of OOF blur). When the shot is very important it would not be too objectionable to spend some time on it with a tool like Topaz Labs Lens Effects which can also use a depth map for accurate DOF transitions into the fore- / background. Nothing beats the ability to see the actual DOF effect change in real time, even after the shot has been taken.
How does one enter the depth map? Some of my subjects routinely have extensive, complicated depth variations and I think that OOF regions should at least a bit correspond to these. Do you use a 3-D scanner or similar tool?
-
Hi Jerome,
I don't think these are silly question. What I reported here is essentially an investigation of my observations on moving from 12 MP/APS-C to 39MP/MFD, withs ome 80MP/MFD images thrown in.
Answers below:
Let me ask some silly questions.
I notice that the article limits itself with 20/20 vision. It just so happens that I have 20/14 vision. If 24 mpix are sufficient for others, does it means I am allowed to use 36 mpix? I also have access to an A1 printer. Does this means that I should use a camera with 72 mpix?
I don't have the slightest idea about how 20/14 vision affects contrast sensitivity function. I have done a few tests at larger sizes like around 86 cm in the long dimension) and there I have seen a discernible difference. But I don't think that it could be detected side by side on full size images, more probably on crops with one on the top.
I don't know how the SQF function would be affected by 20/14 vision. I am actually a bit confused by the SQF function calculated by Imatest as it seems that the difference stays small even at large sizes.
It just also happens that I bought a Sony A900 with 24 mpix in early 2009. I have printed pictures taken with it in A2 size for the past 4 years and am generally pleased with the results. I use the finest Sony, Minolta and Zeiss lenses with it, but I noticed that I have difficulties getting perfectly sharp corners with wide angle lenses on that camera, while the HC28mm has no such problems. Is there anything you can suggest about this?
I don't really have the same experience. The Sony 24-70/2.8 is generally weak in the extreme corners at 24mm, but my Zeiss lenses on the Hasselblad have significant field curvature except the Sonnars. One of the shots at A2 that I could not tell apart is 24-70/2.8 at 35 mm vs. Hasselblad Zeiss Distagon 50/4. Printing the same size crops corners on either P45+ or DSLR.
I have a shot with the Planar 80/2.8 which is very sharp in the centers and awful in the corners. One of the few cases where the Alpha 99 actually outperforms the Hasselblad when the image is blown up to around 40 MP. But, although the Hassy image has horrible corners on screen, the print still looks almost perfect in A2-print.
I have noticed that foreground often improves with the Hasselblad/Zeiss lenses while background sharpness is lost on the edges. Probably field curvature at play.
I compared MTF curves from Hasselblad and Zeiss and in almost all cases the new H-lenses are better performers than the older Zeiss lenses. I guess that it is one of the reasons Hasselblad decided to design their own lenses.
I notice that your study only concerns itself with flat subjects. While I have photographed historical papers for reproduction, most of my subjects are actually tridimensional, with parts which are in focus and other parts less so. I notice that the rendering of out of focus areas is very different between the lenses I have for the A900 (which include the legendary 135mm STF, Sony/Zeiss 135mm f/1.8 and other portrait lenses) and lenses from Hasselblad / Fuji (HC lenses). How can I solve that?
The SQF data is based on flat subjects, but one of the subjects is three D. It is quite true that rendering on out of focus areas is different, even if the Sony is preferable in my case. The Hasselblad lenses have hexagonal apertures while the Sony lenses have circular apertures. The Hasselblad lenses have significant axial chromatic aberration while my most used Sony/Zeiss lenses have very little.
These differences are not related to format or MP, but to lens design. But, you are right, differences in out of focus areas are visible, even if the Sony is superior to the Hasselblad V in this sense. I can add that I don't think this is caused by bad samples, but depend more on less use of AD glass in the older lenses. I am actually on my third Sonnar now, and all exhibit this behaviour.
Last but not least, I see the Sony RX100 mentioned earlier in the thread. While the camera indeed gives results similar to the ones I get with the A900, this only happens at the widest end on the zoom range and when using iso 80. Even then, I noticed coloured fringes appearing where I do not always want them and fine details have an unnatural appearance. How do you explain that?
No one stated that the Sony RX could keep up with the Alpha 99 or the P45+ at A2 size. What I wrote is that I have made an A2 size print with the RX100 that I would hang at an exhibition, but I would not challenge pixel peeping. That image happens to be a close up of a flower. In this case the small size of the RX100 sensor was helpful as it got most of the subject in focus while still giving a nice out of focus background.
Best regards
Erik
-
Hi,
I have noticed that applying selective sharpening is very good at fixing corner/edge sharpness fall of on the Hasselblad Zeiss lenses. I have a strong presumption that much of the sharpness loss is coming from field curvature on those lenses, so the result is mostly out of focus images.
Unfortunately, I have not seen measured MTF curves on the Sony lenses, but I feel that they may respond less well to sharpening, but I have few issues with the Sony lenses.
Depends on the type of sharpness fall-off, which is of course lens specific. Maybe you can use a 2-layer sharpening approach, one sharpened for the center, and one for the corners, and then use a radial mask (blurred for a gradual transition). A Raw converter like Capture One Pro has a user adjustable sharpness fall-off correction built in.
-
I don't have the slightest idea about how 20/14 vision affects contrast sensitivity function.
Erik, I don't have a reference for this, but when I was regularly attending SPIE conferences in the early 90s, I remember a paper that was presented that said the following: In young subjects requiring no visual correction, there is a very large variation in the packing density of cone cells in the foveola. I remember 6:1 linearly measured, but that seems so large as to strain credulity. The researchers found that visual acuity, not surprisingly, varied in proportion to linear packing density.
The 6:1 ratio would indicate that if there are some young people capable of fully-corrected 20/8 vision, others would only be capable of 20/48 vision.
Thus it would be reasonable to suggest that the CSF of a person with 20/14 vision would be shifted to the right by a factor of 20/14=1.4 (before the logarithm is taken). Now, maybe the whole thing isn't shifted to the right, because the factors that cause the droop on the left side of the peak may not scale with cone density. However, the right side past the peak will probably shift rightwards proportionally. The visual acuity test are usually performed at high contrast, so the relevant portion of the CSF is well down on the right hand slope.
Thus at the same viewing distance, we'd need twice as many pixels in an image viewed by a person with 20/14 vision to have the spatial frequencies at threshold suffer the same attenuation.
That all assumes that the lens is capable of resolving to a greater level of detail than the foveola, which seemed to be true in the experiments reported on at the SPIE conference.
Jim
-
I don't have the slightest idea about how 20/14 vision affects contrast sensitivity function.
It just means that, on the standard eye test (the ones with progressively smaller lines of text), I am capable to read lines smaller than the one marked "20/20" (which is actually marked 1.0 in Europe). Basically I "see" more pixels than the average human. It can go even higher, 20/8.9 was the highest score recorded in a study (http://www.optometricmanagement.com/articleviewer.aspx?articleid=71635) of some US professional athletes.
I don't really have the same experience. The Sony 24-70/2.8 is generally weak in the extreme corners at 24mm, but my Zeiss lenses on the Hasselblad have significant field curvature except the Sonnars. One of the shots at A2 that I could not tell apart is 24-70/2.8 at 35 mm vs. Hasselblad Zeiss Distagon 50/4. Printing the same size crops corners on either P45+ or DSLR.
The Sony SAL-2470F28Z lens is quite good at 35mm, but I was thinking wider angle than that. The HCD 28mm is close to 21mm is "fullframe" equivalent. Do you have any lens that wide?
Still: talking about 35mm focal length, the equivalent in Hasselblad H would be the HC 50mm. I think I have posted a comparison (http://www.luminous-landscape.com/forum/index.php?topic=83294.0) between that lens (version II) and the best Nikon lens I had on the D800. The HC lens was noticeably better, but the comparison was equivalent to a A0 print.
I have a shot with the Planar 80/2.8 which is very sharp in the centers and awful in the corners.
Are you using that lens at close distances? It should not have poor corners when the subject is far away.
I compared MTF curves from Hasselblad and Zeiss and in almost all cases the new H-lenses are better performers than the older Zeiss lenses.
Well, yes. Optics made some progresses in the past 40 years.
The Hasselblad lenses have significant axial chromatic aberration while my most used Sony/Zeiss lenses have very little.
That is surprising. Do you try to use your MF lenses at the maximum aperture? They are not designed to be used that way. Do you know why?
No one stated that the Sony RX could keep up with the Alpha 99 or the P45+ at A2 size.
I thought I did. But I immediately qualified that statement. Why and how I qualified that statement was the intended message.
-
Jim,
Thanks for the information.
I presume that using a closer viewing distance like 28 cm instead of 40 cm would compensate for 20/14 vision.
I am somewhat confused by the small variance of the SQF with print size, but I guess it depends on sharpening causing MTF to keep above 90% to about 50 lp/mm. I was just using the "Landscape" sharpening default on the Sony A7 ZA 85/1.4 combination.
I guess sharpening plays a very significant role, until enlargement makes that the picture starts to fall apart.
Best regards
Erik
Erik, I don't have a reference for this, but when I was regularly attending SPIE conferences in the early 90s, I remember a paper that was presented that said the following: In young subjects requiring no visual correction, there is a very large variation in the packing density of cone cells in the foveola. I remember 6:1 linearly measured, but that seems so large as to strain credulity. The researchers found that visual acuity, not surprisingly, varied in proportion to linear packing density.
The 6:1 ratio would indicate that if there are some young people capable of fully-corrected 20/8 vision, others would only be capable of 20/48 vision.
Thus it would be reasonable to suggest that the CSF of a person with 20/14 vision would be shifted to the right by a factor of 20/14=1.4 (before the logarithm is taken).
Thus at the same viewing distance, we'd need twice as many pixels in an image viewed by a person with 20/14 vision to have the spatial frequencies at threshold suffer the same attenuation.
That all assumes that the lens is capable of resolving to a greater level of detail than the foveola, which seemed to be true in the experiments reported on at the SPIE conference.
Jim
-
I presume that using a closer viewing distance like 28 cm instead of 40 cm would compensate for 20/14 vision.
I have better vision than 20/20, so I would need to go further away. About 60cm.
-
I see that you answered my silly questions very seriously. Now I find myself almost obliged to ask some more.
That was my punishment for asking silly questions on a respectable forum like this ;)
Let me see: I already said that I have the Minolta/Sony 135mm STF lens. As you certainly know, this lens uses an apodisation element to shape the distribution of out of focus highlights. This has an effect on the perceived depth of field.
Does your tool account for that?
It may change the shape of the blur PSF, but that doesn't necessarily mean that you need to use the provided (circular/disk) COC as a limiting factor to plan a shoot. But more seriously (I did warn!), one would hardly use such a lens for a specific DOF limitation planning, but rather a creative shape of the Bokeh. Even the camera's viewfinder will provide that at narrower than f/2.8.
Another lens I have is the 100mm f/2.0 DC lens. As you probably also know, it has a ring to adjust the amount of spherical aberration, thereby changing the apparence of the depth of field. How to I enter the value of the ring in your tool?
The spherical aberration will add veiling glare to the image. That will not necessarily change resolution as such (although it might at the wide end), but will change the shape of the MTF curve (overall lower contrast). Again, more of a creative addition than a DOF planning issue.
Portrait lenses, for example 85mm f/1.4 or faster or the HC 100mm f/2.2 have a noticeable amount of uncorrected spherical aberration. In effect, they are a bit like the DC lens, with the ring glued into the max position. This is a wanted feature by portraitists, because it gives a flattering appearance to human skin. It also increases the apparent depth of focus. How do I adjust that in your tool?
As above, although resolution may be lower due to potential defocus of certain wavelengths. My 85mm f/1.2 has a lot of longitudinal spherical aberration, which I personally don't like, but at f/1.2 almost everything is OOF and that is what one is usually after by using such a lens (I prefer it for its bokeh and resolution at f/2.8 and narrower, and the bright viewfinder).
Well... at the time, I tried conversion software from DXO. It includes special routines to improve lens sharpness from supported lens/camera combinations. While these gave excellent results for distortion or lateral chromatic aberration, the far corners stayed rather fuzzy. Is Capture One Pro really better?
It's different, DxO attempt to correct unsharpness by spatially variant deconvolution. That should produce good results. Capture one uses an unknown kind of sharpening. I don't think its deconvolution, but it is producing relatively little halo, so it probably isn't simple USM either. It seems to vary the amount of sharpening radially. So it might be less effective for a complex PSF, but it does help with less problematic cases. I usually export withaout any sharpening, because I prefer to do it as a postprocessing step on a layer in Photoshop (allows to disable it for downsampling, and/or mask out it where required, e.g. in sky areas).
How does one enter the depth map? Some of my subjects routinely have extensive, complicated depth variations and I think that OOF regions should at least a bit correspond to these. Do you use a 3-D scanner or similar tool?
The 'Lens Effects' plugin allows to use an automatically generated map as a basis for editing, but editing is usually required because it is hard to guess depth from a 2D original. It therefore also allows to create a map by hand painting in the depth clues with Edge Aware masking (one does get somewhat proficient with experience). If the image is important enough, it's probably easier (not easy, but easier) to create a depth map in Photoshop and load that in 'Lens Effects'.
I've experimented with the generation of depth maps of a scene based on stereo image-pairs, but the technology leaves a lot to be desired. Research is progressing though, and single image derived depth maps (http://make3d.cs.cornell.edu/) are getting better all the time, so it's a matter of time before we can laugh at our current struggles. Even my Android tablet can create somewhat re-focusable images (http://googleresearch.blogspot.nl/2014/04/lens-blur-in-new-google-camera-app.html) with a poor-man's Lytro like technology but based on images from different viewpoints.
Cheers,
Bart
-
The spherical aberration will add veiling glare to the image. That will not necessarily change resolution as such (although it might at the wide end), but will change the shape of the MTF curve (overall lower contrast). Again, more of a creative addition than a DOF planning issue.
Spherical aberration changes depth of field. I am sorry to insist.
Spherical aberration makes it so that some rays of an object at a given distance are in focus and some other not. This is the classical diagram:
(http://upload.wikimedia.org/wikipedia/commons/9/92/Spherical_aberration_2.svg)
What this also means is that rays going through the center of the image are focussed at a distance f and rays going through the edge of the lens at a distance f+ɛ. The plane of focus is extended to a thicker locus of focus.
My 85mm f/1.2 has a lot of longitudinal spherical aberration
You are probably meaning longitudinal chromatic aberration. This is different, and usually not a problem with medium format. I'll let you find out why.
It's different, DxO attempt to correct unsharpness by spatially variant deconvolution. That should produce good results.
In my experience, it does not. Sorry. Software does not replace good lenses.
I've experimented with the generation of depth maps of a scene based on stereo image-pairs, but the technology leaves a lot to be desired.
How surprising. ::) So you are saying that it is technologically challenging to emulate lenses which just work in medium format?
-
I'm using a tiny point and shoot (Sony RX100) at the moment, and the results are pretty amazing. ...
I have the RX100M2, and I have printed a (cropped, handheld) image from that camera as a tripterych measuring a total of 76x130cm.
At that size, upon inspection, it is obvious to me that there could have been more spatial details. Perhaps more attention to post processing is needed (I recently purchased Focus Magic, and I am considering some Topazlabs tools after Barts recommodation). I am confident that better technique and/or a better camera _could_ have given better results for less postprocessing work. But that shot was not planned, and a larger camera or stand would not have been feasible.
-h
-
Hi,
LoCa is definitively a problem with medium format lenses of older design, at least the ones I have. Lloyd Chambers tested a Rollei Hy6 recently and little doubt that the images had lot of axial chroma.
Very much possible that your Hasselblad lenses don't have axial chroma, due to modern construction. The Minolta 80-200/2.8 APO zoom I have has lots of axial chroma but the 70-400/4-5.6G has virtually none. The Zeiss Otus is constructed so it has no axial chroma at full aperture.
I would say that modern lenses with liberal use of AD glass have little axial chromatic aberration while older lenses using more normal glass and fewer elements have a significant amount.
By the way, I am pretty sure Bart is right on axial spherical aberration. Your drawings actually show an axial aberration not a lateral one.
Best regards
Erik
You are probably meaning longitudinal chromatic aberration. This is different, and usually not a problem with medium format. I'll let you find out why.
-
Hi,
The reasoning is that if we calculated SQF for 26 cm the spacial frequencies used in the calculations would be 1.4 times higher correcting for your vision.
But yes, it is correct if a common man could observe differences at 40 cm than you would be able to discern the differences at around 56 cm.
Best regards
Erik
I have better vision than 20/20, so I would need to go further away. About 60cm.
-
LoCa is definitively a problem with medium format lenses of older design, at least the ones I have. Lloyd Chambers tested a Rollei Hy6 recently and little doubt that the images had lot of axial chroma.
Very much possible that your Hasselblad lenses don't have axial chroma, due to modern construction. The Minolta 80-200/2.8 APO zoom I have has lots of axial chroma but the 70-400/4-5.6G has virtually none.
But if you check in your a treaty of optics how longitudinal chromatic aberration changes with aperture, you will understand why your observations, although true in absolute terms, are not relevant in practice.
The Zeiss Otus is constructed so it has no axial chroma at full aperture.
A miracle of technology and one which comes at the cost of 12 elements, 6 of special glass and one aspherical. And it costs and weights as much as a MF camera.
You see: this is one of the things your tests do not show, because they are designed to ignore these elements. To get a picture with a given, pleasing, depth of field and rendering of out of focus areas which are pleasing and devoid of color casts one has two choices:
-use a MF camera and an aperture of f/4, where the problems are minimal (due to the slow aperture) or
-use a small sensor camera and an aperture so fast it needs a lens corrected so well it costs a fortune (the Zeiss Otus).
By the way, I am pretty sure Bart is right on axial spherical aberration. Your drawings actually show an axial aberration not a lateral one.
The Canon 85mm f/1.2 exhibits spherical aberration and longitudinal chromatic aberration.
As to axial versus lateral spherical aberration, both are the same aberration (http://en.wikipedia.org/wiki/Optical_aberration#Aberration_of_axial_points_.28spherical_aberration_in_the_restricted_sense.29), just measured differently. On the other hand, axial versus lateral chromatic aberration are different, with the first of the two being the only one responding to aperture.
-
Hi,
My Sonnar 150 and Sonnar 180 show significant colour fringing on unsharp backgrounds at f/8. As I said I had two Sonnar 150/4s and now a Sonnar 180/4.
I exchanged the first Sonnar that was a CB for a CF because I didn't like the haptics on the CB lenses. Later I replaced the second 150/4 Sonnar, because I found I used the 120/4 more that the 150/4 so I wanted a longer lens instead.
I am aware that axial and longitudinal mean the same thing, it is just I prefer the term axial and lateral. Both terms are correct. Perhaps influenced by Modern Photography test reports I read back around 1975, they used that terminology in their optical bench reports.
Regarding the Otus it costs 25400 SEK here in Sweden and the Hasselblad 80/2.8 costs 19260 SEK (both without sales tax).
Best regards
Erik
But if you check in your a treaty of optics how longitudinal chromatic aberration changes with aperture, you will understand why your observations, although true in absolute terms, are not relevant in practice.
A miracle of technology and one which comes at the cost of 12 elements, 6 of special glass and one aspherical. And it costs and weights as much as a MF camera.
You see: this is one of the things your tests do not show, because they are designed to ignore these elements. To get a picture with a given, pleasing, depth of field and rendering of out of focus areas which are pleasing and devoid of color casts one has two choices:
-use a MF camera and an aperture of f/4, where the problems are minimal (due to the slow aperture) or
-use a small sensor camera and an aperture so fast it needs a lens corrected so well it costs a fortune (the Zeiss Otus).
The Canon 85mm f/1.2 exhibits spherical aberration and longitudinal chromatic aberration.
As to axial versus longitudinal spherical aberration, both are the same aberration (http://en.wikipedia.org/wiki/Optical_aberration#Aberration_of_axial_points_.28spherical_aberration_in_the_restricted_sense.29), just measured differently. On the other hand, axial versus longitudinal chromatic aberration are different, with the first of the two being the only one responding to aperture.
-
I have the RX100M2, and I have printed a (cropped, handheld) image from that camera as a tripterych measuring a total of 76x130cm.
At that size, upon inspection, it is obvious to me that there could have been more spatial details. Perhaps more attention to post processing is needed (I recently purchased Focus Magic, and I am considering some Topazlabs tools after Barts recommodation). I am confident that better technique and/or a better camera _could_ have given better results for less postprocessing work. But that shot was not planned, and a larger camera or stand would not have been feasible.
-h
Herbert Keppler wrote about his friend who did the Minox advertising images - the secret was the huge stable tripod :)
Setting a 1/2000 shutter speed if one can, and holding the camera like an SLR, under the lens, may help.
The files are not amazing, but it is certainly a camera and not a toy. Modern 35mm, I guess.
Edmund
-
My Sonnar 150 and Sonnar 180 show significant colour fringing on unsharp backgrounds at f/8. As I said I had two Sonnar 150/4s and now a Sonnar 180/4.
Axial colour aberration at f/8? Now I am genuinely surprised. Or is this "colour fringing" lateral chromatic aberration? That does not respond to aperture.
I am aware that axial and longitudinal mean the same thing, it is just I prefer the term axial and lateral. Both terms are correct.
Sorry, I meant axial and lateral. I corrected the post.
Regarding the Otus it costs 25400 SEK here in Sweden and the Hasselblad 80/2.8 costs 19260 SEK (both without sales tax).
I meant to say that the Otus, which one must buy new, is as expensive as a second hand MF.
-
Axial colour aberration at f/8? Now I am genuinely surprised. Or is this "colour fringing" lateral chromatic aberration? That does not respond to aperture.
Sorry, I meant axial and lateral. I corrected the post.
I meant to say that the Otus, which one must buy new, is as expensive as a second hand MF.
Who needs Otus, when they can have Art?
Edmund
-
Hi,
Bart demonstrated the tools in processing one of my images starting at this link: http://www.luminous-landscape.com/forum/index.php?topic=87650.msg716838#msg716838
I have repeated his processing on a few of my images but I have not printed any of those images yet.
Best regards
Erik
I have the RX100M2, and I have printed a (cropped, handheld) image from that camera as a tripterych measuring a total of 76x130cm.
At that size, upon inspection, it is obvious to me that there could have been more spatial details. Perhaps more attention to post processing is needed (I recently purchased Focus Magic, and I am considering some Topazlabs tools after Barts recommodation). I am confident that better technique and/or a better camera _could_ have given better results for less postprocessing work. But that shot was not planned, and a larger camera or stand would not have been feasible.
-h
-
Hi,
Bart demonstrated the tools in processing one of my images starting at this link: http://www.luminous-landscape.com/forum/index.php?topic=87650.msg716838#msg716838
I have repeated his processing on a few of my images but I have not printed any of those images yet.
Best regards
Erik
Actually, I just found a solution in Photoshop CC > Sharpen > Shake reduction.
If Sony used the incredible Foveon sensor we would have the ideal point and shoot :)
Edmund
-
If I may come back to your article and questions.
As I said: I have been printed in A2 size from the Sony A900 for the past 3 or 4 years. The results are generally very good. Therefore your article does not come as a surprise.
From the photographs you have posted, you seem to be mainly interested in landscape photography. With landscape photography and your A99, the only defect anyone is likely to notice on A2 prints would be lens defects, because 35mm wide-angle lenses are rarely perfect. If you limit yourself to the following wide-angle lenses, you will have excellent A2 prints:
-the Sony Minolta 20mm at f/8, if you correct lateral chromatic aberration
-the Sony Zeiss 24mm from f/4 onwards.
-your Sony Zeiss 24-70mm between 35mm and 50mm at f/8.
Under these conditions, your A99 will give A2 prints which will be as good as anything from a MF camera.
The difference with a MF camera will only show:
-on landscape if you print larger than A2 and look at the prints from the same distance
-on portraits or generally tridimensional subjects with limited depth of field, because the bokeh is different. On this subject I should add that classic Minolta lenses and some lenses like the Sony Zeiss 24mm already have very good bokeh, so that the difference may not be dramatic.
-
Hi,
I got my Sony Alpha 900 in december 2008, and I was surprised at that time how well my pictures from the 12 MP Alpha 700 in APS-C held up to the full frame Sony Alpha 900. At least in one of the early test shots I could no observe any difference between the 12 MP Alpha 700 and the 24 Alpha 900. The difference in file quality was impressive, but in A2 size prints, the difference was nil. I tested that shot on quite a few observers.
Since that time I have repeated that kind of test many times. Essentially I found that 12 MP are adequate for A2 size prints, with 24 MP offering a small advantage if the stars are in proper alignment. For instance at one time it was windy and dark. On the Alpha 900 I didn't want to use high ISO, but with the Alpha 55 (16 MP APS-C) I could risk a bit higher ISO, I also had live view so I could have pin point focus, so I could use a larger aperture, and because of the crop factor I could use my 24-70/2.8 instead of the 70-300/4.5-5.6.G.
In the final evaluation the A2 print from the Alpha 900 was a bit sharper when viewed obscenely close, but the Alpha 55 picture had a better DoF and was sharper overall because of the larger aperture and the shorter exposure time.
Now that I have the P45+ I found a similar experience. The images from the Alpha 99 are good enough and although the files from the P45+ are better most of the advantage is lost in print.
Now, what we see in print is depending on eyesight. I have 20/20 vision with corrective glasses, and it is my understanding that you have 20/14. I am near sighted so I can focus quite close without glasses, and I still don't see differences in A2 prints, but I can clearly observe differences with a loupe. What I would see with younger or better eyes, I cannot tell.
In general, I try to verify my findings with friends and colleagues but they tend to be of similar age than me.
Getting back to the lenses, I have a few of them.
What I use mostly is the 24-70/2.8. It is weak in the corners at 24 mm, but I seldom have an issue with that. At the long end it seems to loose a lot of sharpness of axis, but is still mostly OK. I have also a Sigma 12-24/4.5-5.6 (I think), a Samyang 14/2.8 which happens be one of the sharpest lenses I own, but with an awful distortion. The Minolta 20/2.8 is sort of acceptable, but I feel it needs to be stopped down to f/11 for good corners. It is a bit better in the corners than the 24-70/2.8 but I may feel the 24-70/2.8 may be more "snappy".
I also have a 28-70/2.8 Minolta (made by Tamron). Little doubt it outperforms the 24-70/2.8 "Zeiss" in the corners at large apertures. For some reasons I use the Zeiss. The SSM and the "Zeiss" label may be part of that. I don't know…
When I got my Alpha 77, I found that neither of my zooms, the 80-200/2.8 or the 70-300/4.5-5.6G were good enough, so I bought the 70-400/4-5.6G. In initial tests it performed similar to my 400/4.5 APO. I was quite happy with that lens as I could handle weight limits when flying.
With the Hasselblad 555 I have now I have 5 lenses:
- Sonnar 180/4 CFi, excellent across the field. Only issue is a lot of longitudinal chromatic aberration (LoCA)
- Macro Planar 120/4 CFi, main problem is field curvature needs to be stopped down to f/11 - f/16 at infinity. A bit low contrast? *)
- Planar 80/2.8CFE, very good at center but dismal in the corners, at least what I see now.
- Distagon 50/4 FLE, a bit soft, possibly
- Distagon 40/2.8 FLE, I don't know, sometimes very sharp sometimes not. I guess field curvature is part of the explanation.
I'm mostly shooting with those lenses and not doing that much testing.
My findings agree decently well with published MTF data from Zeiss and Hasselblad. I guess that it was one of the reasons Hasselblad chose to design their own lenses for the H-series.
It seems that Zeiss made a couple of new lenses for the Hasselblad, including the Distagon 40/4 IF which is dramatically improved over the older designs. I am interested in that lens. Another interesting lens is the 100/3.5 which is extremely well corrected. The 100/3.5 is an older lens.
Best regards
Erik
*) I have switched quite a few of my lenses. First I had a Sonnar 150/4 CB but later I found that I preferred the CF designs. So I upgraded my 150/4 CB to 150/4 CF. Later I felt that I used the 120/4 a lot, and found that the 150/4 was a bit to close so I replaced it with an 180/4CFi. I have a very helpful dealer ;-) The aperture in my 120/4 had issues, so I replaced it with the 120/4 CFi, which according to Zeiss (but not their MTF data) is much better. If it is better I still need to find out.
If I may come back to your article and questions.
As I said: I have been printed in A2 size from the Sony A900 for the past 3 or 4 years. The results are generally very good. Therefore your article does not come as a surprise.
From the photographs you have posted, you seem to be mainly interested in landscape photography. With landscape photography and your A99, the only defect anyone is likely to notice on A2 prints would be lens defects, because 35mm wide-angle lenses are rarely perfect. If you limit yourself to the following wide-angle lenses, you will have excellent A2 prints:
-the Sony Minolta 20mm at f/8, if you correct lateral chromatic aberration
-the Sony Zeiss 24mm from f/4 onwards.
-your Sony Zeiss 24-70mm between 35mm and 50mm at f/8.
Under these conditions, your A99 will give A2 prints which will be as good as anything from a MF camera.
The difference with a MF camera will only show:
-on landscape if you print larger than A2 and look at the prints from the same distance
-on portraits or generally tridimensional subjects with limited depth of field, because the bokeh is different. On this subject I should add that classic Minolta lenses and some lenses like the Sony Zeiss 24mm already have very good bokeh, so that the difference may not be dramatic.