Tests
Low magnification test
On this occasion I wanted to make a test involving relatively long focal length lenses and I have done it through different magnification ranges.
The good thing about this long focal length lenses is their good working distances ( WD), these lenses are supposed to become part of my field gear but they can also be used in the studio, of course. I forgot to measure their WD but as an example the Rodagon-D 75/4.5 has around 10cm WD at 2X.
The following lenses were tested:
Olympus MC auto 1:1 macro 80mm f4; a good reputation bellows lens, optimized for 1:1 work and a useful working range from 0.5X to 2X.
Rodenstock APO Rodagon-D 1X 75mm f4; a specialized enlarger lens (in fact it is a duplicating lens) optimized for 1:1 work, it's probably use was film duplication. Its optimum working range goes from 0.7X to 1.5X in 35mm cameras.
Rodenstock APO Rodagon-D 2X 75mm f4.5; other version from same series optimized for 2:1 work, one of its designed tasks may be copying 16mm film to 35mm.
Its optimum working range goes from 1.2X to 2.5X in 6x7 film; as this lens has not got a symmetrical design reciprocal values may apply when reversed so it is also optimized for 0.5X work, having a working range of 0.4 to 0.8X in 6x7 film(It is used at 2X when reversed and at 0.5X when in normal position) I have not found its optimum working range in 35mm cameras but I guess it will have a pretty well balanced behavior from 0.3X to 3X.
I run several stacking sequences at different magnifications: 0.6X (minimum I could get with Olympus 80mm), 1X, 2X and 2.6X (maximum I could reach with Rodagon-D 75/4); at different apertures like f4 (f4.5 for the Rodagon-D 75/4.5), f5.6 and f8.
The test subjects were a couple of old pc cards; a graphics card and a Ethernet card.
I used office paper around the subjects for light diffusion, light source were three ikea led lamps. Stacks were automatized with stackshot (great help for these repetitive tasks) on a EOS 5D mkII in live view silent mode mounted on a nikon PB-6 bellows.
The only post processing applied to the pictures was recovery 20 on Camera raw; images were stacked on Zerene stacker Pmax mode.
0.6X test FOV 59mm
In this image you can see in red the crops we will examine later on.
Three stack sequences were run for each lens, clicking on the f number you can see them full size. Next to each of them you can see the number of steps, their size and the exposure time.
All lenses were used in normal position.
Rodagon D 75/4
f4 16 shots with 1mm steps 1/6
f5.6 11 shots with 1.5mm steps 0.3"
f8 8 shots with 2mm steps 0.6"
Rodagon D 75/4.5 Normal
f4.5 14 shots with 1.2mm steps 1/5
f5.6 11 shots with 1.5mm steps 0.3"
f8 8 shots with 2mm steps 0.6"
Olympus 80/4
f4 16 shots with 1mm steps 1/6
f5.6 11 shots with 1.5mm steps 0.3"
f8 8 shots with 2mm steps 0.6"
Let's have a look to those crops; all three lenses are pretty similar, maybe the Rodagon-D 75/4 struggles a little more in the corners but it also shows better resolution in the center wide open. The weird thing is that it works better at f4 than at f5.6.
Thinking about it I realized I made DOF calculations based on 0.5X magnification (I got confused), so probably this is what is wrong.
However I think that there are enough areas completely focused on the full size pictures
Clicking on the image you will see it original size (100% crops)
1X test FOV 36mm
Here the full image showing the crop areas, same as before.
On this occasion I did three stacks per lens, but I used the Rodagon-D 75 /4.5 both in normal and reversed position. The reason for that being the Rodagon has an asymmetrical lens design and at 1X is in theory out of its optimum magnification range; I wanted to see which way it works best.
Once again next to the f number you have the number of pictures, step size and exposure time.
Rodagon D 75/4
f4 20 shots with 0.35mm steps 1/13
f5.6 13 shots with 0.55mm steps 1/6
f8 9 shots with 0.80mm steps 0.3"
Rodagon D 75/4.5 Normal
f4.5 17 shots with 0.42mm steps 1/10
f5.6 13 shots with 0.55mm steps 1/6
f8 9 shots with 0.80mm steps 0.3"
Rodagon D 75/4.5 reversed
f4.5 17 shots with 0.42mm steps 1/10
f5.6 13 shots with 0.55mm steps 1/6
f8 9 shots with 0.80mm steps 0.3"
Olympus 80/4
f4 20 shots with 0.35mm steps 1/13
f5.6 13 shots with 0.55mm steps 1/6
f8 9 shots with 0.80mm steps 0.3"
Let's examine those crops; again all three lenses have very good performance. In my opinion the Rodagon-D 2X works better reversed rather than in normal position. Probably the Rodagon-D shows a little more detail in the center of the image but the difference is not as big as I would have expected before running the test. The Olympus also performs very well at this magnification. Do not forget that this is a full frame test; 21mpx seem a lot but pixel density is well below what we find in APS-C cameras.
2X test FOV 18mm
We get now to the 2X test, here the Olympus is still in is designed working range, The Rodagon-D 2X is in its optimum range (in reversed position) and the Rodagon-D 1X is well outside of its designed working range. I had seen some tests of these Rodagons which suggested the Rodagon-D 1X had better resolution at 2X, this did not make sense to me; how can a lens designed to work at 1X be better than a lens designed for 2X work, specially when both lenses were designed by the same maker. It is true that those tests were always APS-C tests.
On this stack series I overexposed the shots a little bit, which caused some haloing around the bright white areas; as this is just a test it does not bother me much, as long as we use other areas for the comparison.
Rodagon D 75/4
f4 28 shots with 0.18mm steps 1/6
f5.6 22 shots with 0.25mm steps 0.3"
f8 15 shots with 0.35mm steps 0.6"
Rodagon D 75/4.5 reversed
f4.5 26 shots with 0.2mm steps 1/5
f5.6 21 shots with 0.25mm steps 0.3"
f8 17 shots with 0.35mm steps 0.6"
Olympus 80/4
f4 26 shots with 0.18mm steps 1/6
f5.6 22 shots with 0.25mm steps 0.3"
f8 15 shots with 0.35mm steps 0.6"
Once we look the crops first thing we can see is that the Rodagon-D 2X is very well optimized for 2X work; it was designed with balance in mind, offering almost same performance wide open than with aperture closed 1 or 2 EV. We can also see that both the Olympus and the Rodagon-D 1X struggle in the corners, specially wide open. The rodagon-D 2X is superior even on borders. However the Rodagon-D 1X does show more resolution in the center of the image, which is the reason it seems to be superior in APS-C tests I have seen before.
2.6X test FOV 14mm
Taking the results we saw at 2X we can expect them to be similar on this last test; the Rodagon-D 2X (reversed) still will be in is designed working range, not like the other two lenses.
Rodagon D 75/4
f4 48 shots with 0.09mm steps 1/6
f5.6 30 shots with 0.15mm steps 0.3"
f8 25 shots with 0.20mm steps 0.6"
Rodagon D 75/4.5 reversed
f4.5 45 shots with 0.10mm steps 1/5
f5.6 30 shots with 0.15mm steps 0.3"
f8 25 shots with 0.20mm steps 0.6"
Olympus 80/4
f4 48 shots with 0.09mm steps 1/6
f5.6 30 shots with 0.15mm steps 0.3"
f8 25 shots with 0.20mm steps 0.6"
On this last crop series we can see similar results to those on the 2X test, maybe both the olympus and the Rodagon-D 1X got a little worst, starting to show some mild CAs. The Rodagon-D 2X probes to be very well balanced and in my opinion is best choice for high magnification shots.
After reviewing test results we can see that when makers design a lens it is difficult to make it work well at different magnifications, they have to make choices. APS-C and 4/3 users may think the Rodagon-D 1X is best choice because of the superior resolving power in the center of the frame; however I do prefer the Rodagon-D 2X as it is a very well balanced lens, for me this is an more important aspect than center resolution.
What about the Olympus? Well; a very good quality lens, showing very good performance more similar to that of the Rodagon-D 1X. Its only con it that it is a rather difficult lens to adapt.
On a OM bellows system with this lens and the 38/2.8 you would have all you need from 0.5X to 6-7X. If your bellows system uses a different mount it can be a problem, it is quite difficult to find M42 or Nikon adapters.
You can always put together some pieces and make your own, problem being it also adds some extension; this can be a problem or not, depending on your needs.
OM to M42 DIY adapter
Playing with the light by Nikola Rahmé
Nikola Rhame is well known for his beautiful studio stacks, showing always a perfect lighting technique. He has also conducted a few light experiments in his flickr, here we are going to show a couple of them:
Studio lighting comparison (flash vs. LED)
Since I use two different lighting system for high magnification studio photography, was curious about the difference of the effects.
Well, I choose a tiny leaf miner jewel beetle from my collection and did two stacking sequence. One lit with a Speedlite 580EXII flash through a plactic diffuser around the subject, and another with 2 IKEA table LED lamps through the same diffuser. Find the picture of the setup below!
Shots were taken with a Mitutoyo BD Plan 10x microscope objective on a Apo-Gerogon 9/150 enlarger lens at 9,2x magnification. 105 exposures in each stacks. Shutter speed with flash: 1/100 sec. (live view mode), with LEDs: 1/3 sec. (live view, silent shooting mode).
I found no remarkable difference in the light effects. Perhaps when the flashlight is coming frontwise of the beetle looks more natural than the two separated light reflection. Another strange thing appears: the different color of the LED lamps. I have 4 Jansjö lamps and none of them has the same colors.
Then I looked at the stacks at the original resolution and surprised of the difference. I do not know why, the LED lit version has slightly finer and perhaps sharper details.
Crops from the original images:
Check out the comparison below side-by-side!
How the setup looked like:
Playing with the light
I am a big fan of diffusing the light when photographing insects. Most of the shiny subjects looks far better under drastically diffused flash or other lights. The tiny details of the surface pass off in spotted illumination, so I always prefer using white plastic or white office paper around the insects.
Now I wouldn't show example shots with absolutely pure lights, I hope each of you know the problem. As my favorite family of beetles are the jewel beetles, I have a lot of experiences in shooting them. In the nature they frequent sun-exposed places, like cut woods, flowers or dry trees. Pictures made under these conditions will be overcontrasted, too many black areas and shiny spots, loosing the nice sculpture details. You'd be better to photograph them in overcast or in early morning.
Excercise in the studio gives useful experiences which can be profitable in the field.
Jewel beetles (Buprestidae) are mostly colorful, metallic insects, they always look exciting if the light is well controlled. I obtained some routine with this small fellow a few days ago. It is a prepared leaf miner beetle -Trachys troglodytiformis choosen from my collection. Its length is 2,95 mm.
For the test stacks I used the JML 21/3.5 lens on a bellows set to 7x magnification. Two sequences were made, first with a whole paper cylinder, second with the same cylinder but with a hole on it, and a translucent paper fixed on the hole. Both stacks were combined in two stacking methods of Zerene: DMap and PMax, surprisingly the results are different in terms of lighting effect.
My edification is the excessive diffusing loses from the nature of the subject in some cases. Let's check the examples below!
Fully diffused stack, processed with DMap:
Fully diffused stack, processed with PMax:
Partially light-exposed stack, processed with DMap:
Partially light-exposed stack, processed with PMax:
And a synthetic crossed eye stereogram from the shiny stack:
These articles were originaly published by Nikola Rahmé on his flickr site, below you have the links to the original place
Studio lighting comparison (flash vs LED)
My thougths on MP-E 65mm (APS-C test) by John Hallmen
I have conducted a comparison between the lenses shown above:
Canon MP-E 65mm f/2.8 1-5x (I'm just calling this MPE65 from now on)
SumRay 50/1.9 (Summar 12cm and Raynox +12)
Leitz Focotar-2 50/4.5
EL-nikkor 50/2.8 (Old version)
EL-nikkor 50/2.8N
EL-nikkor 50/2.8N
Leitz Milar 40/4.5
Below you'll find full resolution samples from each lens at three different aperture settings.
Initial notes:
Camera body used: Canon EOS 7D
Magnification: 4.8:1
All but the MPE65 were mounted on the Nikon PB-6 bellows.
Each test image is stacked from 9 exposures in Zerene Stacker. This is to get around the fact that aligning focus exactly the same between different lenses is practically impossible. Also: some lens flaws seem to be amplified by stacking and at this magnification stacking is often crucial to get the best results. Therefore even if it was possible to align focus identically – I find stacks are more telling than single exposures at high magnification.
I have included three EL-nikkors for two reasons::
1) I wanted to know if there is a difference between the old and the new "N" version
2) It's nice to know how much sample variations plays a part in this kind of test and as it turns out it is far from insignificant.
There are a lot of lenses I could have included in this test (such as the Zeiss Luminar 63/4.5 or Leitz Milar 65/4.5) but I wanted to use lenses with roughly the same focal length.
The MPE65 is called 65mm but I'm fairly certain it achieves its 5:1 magnification by means of focal length shortening (the relatively short maximum extension and working distance indicates this). My guess was that the MPE65 is between 40 and 50mm at 5:1 but it could be even shorter – I'll see if I can find this out.
This composite is showing all lenses at f5.6. If you click it you'll get a larger version where each is scaled down to 2000x1333px. At this size you can't make out any differences in terms of resolution but this gives you an idea of how they compare when it comes to white balance and contrast.
18MP FULL RES SAMPLES
Leitz Focotar-2 50/4.5 f4 f5.6 f8
EL-Nikkor 50/2.8 N(Sample 1) f4 f5.6 f8
EL-Nikkor 50/2.8 N(Sample 2) f4 f5.6 f8
EL-Nikkor 50/2.8 (Old version) f4 f5.6 f8
Lighting was provided by a single Nikon SB-24 flash unit (see below). It was set to 1/16th, 1/8th and 1/4th of full power respectively for each of the three aperture settings.
The Milar and the Focotar are both f4.5 at their widest so for these I've used this setting instead of f4. Lighting was the same though so theoretically their "widest" image should look half a step or so darker.
When conducting the test I made the mistake of thinking of the "SumRay" as a f2.0 lens while f1.9 is closer to the truth (see here for calculation). Consequently the SumRay yielded slightly overexposed results since the effective aperture was a little bit larger (also meaning a little less diffraction).
App for Comparisons
I've built a flash app that makes comparing test images easier by syncing the zooming and panning. It's original purpose is to view stereograms (which explains the anaglyph-button) but it works well for comparisons too.
Ejemplo 1: MPE-65 vs SumRay a f5.6
Ejemplo 2: MPE-65 vs El-Nikkor a f5.6
Zoom in and out with the "+" and "-" buttons (keyboard or on-screen) and pan using the mouse or arrow keys.
Click "Change images" and copy and paste from the URL:s above to compare two images of your choice.
(Flash can only juggle 3000x3000 pixels so if the files are larger than that (as in this case) only the top, left 3000x3000pixels of the images are visible)
This app is still to be considered a beta-version!
OBSERVATIONS
One of my main tasks here was to find out how good the venerable Canon MPE-65 is performance-wise compared to other optics.
Before we begin the pixel-peeping let me just say this: as someone most familiar with the Nikon system and not having used the Canon system for real since the analogue nineties I'm very pleasantly surprised. If Nikon keels over tomorrow I can now rest assured that there is at least one viable alternative out there.
The MPE65 oozes quality and is a pleasure to use; it's perfectly solid even at maximum extension. The 7D + MPE65 is a kit that should satisfy most macro junkies out there, myself included.
MPE-65 image quality at 4.8X
For this comparison I didn't have time to thoroughly test the lens over its entire magnification range but of course this is what you need to do if you want the complete picture.
So, does the MPE65 sweep the floor with the competition?span>
No. Lets just say that If there is something magical about this lens from an image quality standpoint (as you sometimes are lead to believe) it eluded me in this comparison.
The MPE65 is great when it comes to producing clean, contrasty results with consistent edge to edge performance at wide apertures (f4). At f4 its blacks are blacker than all the other lenses in this test. The micro contrast is also very high and when you put these qualities together the results are very appealing – at least at first glance. You could say that the images appear to have an extra level of clarity, straight out of the camera.
So what's behind those caveats? Well, in terms of resolving power it's not better than the competition in this test. And while CA (chromatic aberrations) are generally under control the Leitz Milar and Leitz Focotar-2 and to some extent the "SumRay" are slightly better while the EL-nikkors are slightly worse. CA is only an issue at f4 – at smaller aperture it is insignificant even in the EL-nikkors.
The resolution champion in this test is the SumRay.
MPE-65 a f5.6 vs Sumray a f5.6 (Click for flash comparison)
At f5.6 the SumRay clearly outresolves the MPE-65 over the most part of the frame (see above). Granted, the SumRay is "cheating" a little since its effective aperture is slightly wider than f5.6 (perhaps f5.4?) and thereby it is slightly less affected by diffraction. The difference isn't big but not insignificant either since you can make out that the SumRay image is a little brighter. Remember that the flash output is exactly the same in both cases so any difference in effective aperture would correspond to an equivalent amount of brightness difference. Looking at these images it is clear that it is far from a full stop difference but lets compare the MPE65 @ f4 with the SumRay still at the "less than f5.6" setting anyway:
Here, the MPE65 @f4 to the left gets half the amount of flash output (1/16th) compared to the SumRay (1/8th). So in this case the MPE65 certainly has the benefit of larger effective aperture. Still, resolution-wise the SumRay is on par with the MPE65 or even a little bit better.
In fact, at f8 the Canon has the lowest resolution of all the lenses in this test. Also, at this setting the contrast/clarity advantage this lens showed at the wider settings is completely gone.
The MPE65 is a good performer but at this magnification we seem to be pushing its upper limit. While it has the edge in contrast/clarity at the wider settings this is really only an issue if you want to do a minimum of adjustments in post – a simple levels adjustment would cancel most difference between the MPE65 image and the results produced by the other lenses in this test. Also, at f4 and 4.8X, stacking is almost a must and if you are the kind of photographer who accepts focus stacking as part of your workflow you are unlikely to shy away from a simple levels adjustment.
Subjective performance
Determining which lens performs better is quite often a matter of taste. Take this case for instance:
EL-Nikkor a f4 contra el Focotar a f4.5
The EL-nikkor @f4 resolves finer detail than the Focotar-2 consistently throughout the frame. On the other hand, the Focotar-2 is almost completely free from CA whereas the EL-nikkor shows relatively much of it. In a case like this I would pick the Focotar-2 without hesitating, but I can see how others might be less sensitive to CA and always opt for maximum resolution!
Another thing that is often and easily overlooked is sample variation.
Two samples of EL-Nikkor 50mm a f4
This test confirms my suspicion that these two virtually identical lenses behave quite differently at f4; One is clearly better than the other. At this setting the difference between these two samples of six element design lenses is larger than the difference between the "good" sample and the three element Milar 40/4.5. Is one of the EL-nikkors to be regarded as a lemon? I'm not so sure! At f5.6 ( flash >>) and f8 they are indistinguishable – if anything the "bad" one shows a little less (!) CA.
Some additional thougths and rants
The MPE65 is good but at this magnification you can achieve equally good results with much cheaper alternatives, especially if you intend to stop down beyond f5.6 or can tolerate a small amount of post processing. The MPE65 is the only lens in this test that offers automatic aperture but this is only a real benefit if you're stopping down and then IQ might not be the best.
For handheld shooting at 1:1 – 3:1 the MPE65 seems really nice but frankly I would be surprised if the Micro-nikkor 105/f4 + Raynox diopter wouldn't prove to be as good in most relevant aspects.
I really don't mean to bash the MPE65 but I'd like to point out some weaknesses just to counterbalance the praise this lens often receives (and for the most part deserves). The following crop (shot at 1:1 and f16) shows another less than perfect aspect of this lens:
The straight, six bladed iris of the MPE65 can render out of focus highlights in a rather unattractive way. More often than not this will not be a problem but there are occasions where this kind of OOF highlights can become obtrusive in my opinion. This is especially true if there is a lot of dew drops or very moist, textured substrate in the frame. Apart from being unattractive this sort of highlights can become a problem when stacking: the distinct shapes and patterns can be "perceived" by the stacking algorithms as in focus detail and thereby severely interfere with the actual detail rendition.
The Summar 12cm has no less than 15 rounded diaphragm blades and even when stopped down all the way its opening forms a perfect circle:
This wouldn't be possible with an auto-aperture lens though since the friction between the blades would prevent it from stopping down fast enough. In fact, in cold weather it takes quite a lot of torque to change aperture on the Summar.
If I was a Canon user I would most likely buy the MPE65. It's a unique, flexible and well built lens and it offers a lot of convenience especially if you don't want to lug around a bulky and sensitive bellows unit.
If you are a Nikon macro shooter I don't think the MPE65 should be reason enough for you to change systems. The old Micro-nikkor 105/f4 combined with Raynox diopters offer a great alternative. However, there are other things that Canon does better than Nikon in my opinion: Canon's Live view and Mirror lock up is superior to Nikon's. And I much prefer to be able to set the nominal aperture rather than the effective aperture that Nikon forces you to use (with modern optics).
If I was to pick a winner in this test it would have to be the SumRay. It outresolves the rest and shows very little CA. Not only is it a stellar performer but it is also very flexible: Changing the Raynox +12 diopter for a Raynox +6 diopter gives you an excellent 70/f2.6 (very useful for bugs). With a Raynox +25 diopter you get a high magnification 30/f1.2 (that becomes excellent, stopped down a bit). On its own the Summar 12cm is wonderful for lower magnification macros (below and around 1:1) and portraits. If you can tolerate a fair bit of edge softness you can even use it for landscape work. I have a feeling my beloved Sigma 150/2.8 will have to work to defend its place in my bag the upcoming season ;-)
The Raynox lenses continue to amaze me with their optical brilliance. I rate them as among the finest pieces of glass I've ever used. Quite astonishing considering their price tags! They perform a little bit differently depending on which lens you put them on though so you can't smack them onto anything and expect wonders. For instance I tried the +12 diopter on the Canon 100/2.8 USM macro lens: The results were not bad but nowhere near as good as from the MPE65
These are my opinions and conclusions. Go ahead and pixel peep the samples and please feel free to disagree with me! And if you do, please tell me – I'd like to hear your opinion!
Test set up
Basically I used the same setup as described in my last lens test (more info here)
However, one very significant difference was the lighting setup. This time I've taken measures in order to shade the front elements of the lenses and hereby flare was drastically reduced!
The shading was done by adding a small "snout" around the flash head and (more importantly) a "mask" around the diffuser-cup. Simple but extremely effective! Just compare the contrast from EL-nikkor between this test and the last one I did
I mounted the MPE65 on the same micrometer controlled linear stage as I used with the bellows:
Here you see MPE65 compared to Zeiss Luminar 63/4.5 (not included in this test) at the same magnification and plane of focus. As you can see there is a significant difference in extension. This also illustrates another point: a small lens leaves much more room for lighting the subject. Another good thing with small lenses is that they are less likely to scare skittish subjects away. With a small lens and a set of narrow extension tubes you can take this to the extreme
Considering how small the front element of the MPE65 is, I don't see why Canon couldn't have made the barrel a little slimmer. It doesn't have to look like an endoscope (or a sniper rifle) but shaving off a little bit of girth at the front end of the lens would have many benefits:
1- easier to light the subject
2- easier to align the lens correctly without looking through the viewfinder (crucial when you need to work with fast moving subjects)
3- easier to monitor the subject from behind the camera (fat lenses with short WD obstructs the view)
4- easier to get better (=lower) angles with subjects on flat surfaces
5- easier to construct a highly effective lens hood: if the MPE65 had a filter thread the same size as the front element (≈25mm) a 1 cm long cylindrical lens hood would be enough to make a huge difference.
Why Summar 12cm f4.5 + Raynox 12diopter becomes SumRay 50mm f1.9
Well the "SumRay" part is obviously just made up but the thinking behind the numbers is as follows:
120mm f4.5 means an entrance pupil of 120/4.5= 26.7mm. The Raynox is +11.8 diopter which means a focal length of 1000mm/11.8=85mm. The combined focal length is calculated as 1/F=1/f1 + 1/f2 => 1/F= 1/120 + 1/85 => F= 50mm. Meanwhile the entrance pupil is unaffected so the f-number of the lens is 50/26.7≈1.9
This might not be 100% accurate but I've been assured it should work perfectly well as an approximation!
This test was originaly published in John hallmen's flickr on the 29th of January of 2010; click here too Read the original article with aditional info and comments
Twelve lenses at 5X (full frame test)
12 lenses at 5X
 |
|
To start our test section we have a 5X test involving 12 different lenses; camera used was a EOS 5D mkII From the begining we will state what kind of camera has been used for the test, either full frame or APS-C, as depending on what kind of format we use some lenses may work better than others. Full frame cameras are more sensitive to quality in the corners and because of this the range of useful lenses is more limited. APS-C cameras are not so sensitive to corner/border quality as they crop the better, center part of the image circle but as this kind of sensor tends to have a higher pixel density they need high resolving lenses (In my opinion the pixel density of some APS-C cameras is just insane, like with the 18mpx of the EOS 7D and specially the 24mpx of new sony cameras) Something very important to bare in mind are aberrations of lenses, chromatic aberrations can be a nightmare when focus stacking As I said, this test was conducted on a EOS 5D mkII, in silent live view mode (EFCS); lighting was provided by three Jansjo IKEA led lamps diffused through a paper cylinder Lenses with iris were used at f4 (when posible), the Olympus 28mm also at f2.8. Mitutoyo 5/0.14 was used together with the 172mm morfanon tube lens. First I run all images on Zerene stacker Pmax, the I choose a group of lenses and redid the stacking in Dmap mode to get the highest quality possible
|
|
JML 50MM Here you can see the full image; red squares show the Pmax crops, blue squares the Dmap crops and green squares a second series of Dmap crops
|
| Pmax crops |
|
First row shows center crops, second row low left corner crops and third row top right corner crops
(click on the image to see it full size )
 |
| Pmax crops, sharpening and levels applied |
From this first round the following lenses are withdrawn:
Fuji 24X, worst of them all, low contrast and resolution (John hallmen owns this lens and got quite good results from it, maybe mine is a big lemon)
Mikrotar 30mm and Bd plan 5/0.1 Both lenses because of same reason, low corner performance. Both lenses would do a good job in APS-C cameras
Rodagon 50mm 2.8 and APO rodagon 50mm 2.8, both lenses were a big dissapointment; I expected more from them, specially from the APO version
JML 26mm the reason I drop this one is because of some low CAs at some high contrast areas, nonetheless it is a very good lens as good as the 21mm 3.5 version.
El Noname I I did not include it in the next round but it could have been included no problem. After seeing the pictures it could be better than the Noname II. I have used it both in the studio and field with very good results
All pictures in the next round were run through Zerene stacker in Dmap and mixed with the problematic areas with the Pmax outputs, as I normally do with all my shots. Click on each lens name to see the full size image.
Mitutoyo 5X
JML 21mm
Olympus 38mm f2.8
Olympus 38mm f4
Noname II
JML 50mm 2.8
| Dmap crop I |
|
First row shows center crops, second row top left corner crops and third row low right corner crops  |
| Dmap crop I with levels and sharpening |
 |
| Dmap crop II |
|
Some more center crops showing loose scales and a pollen grain
 |
| Dmap crop II with levels and sharpening |
 |
In my opinion the Mitutoyo and the JML 21mm are the best, differences between them being quite small. The JML covers corners no problem, the mitutoyo might have some corner degradation but has great center and border resolution. Both lenses are CA free
All lenses in this final round are good optics (Noname I and JML 26mm could well be here too), the Olympus 38mm gives me mixed feelings but to be fair it seems that lighting conditions changed when I used it. At 2.8 it seems to work worse than at f4 and at f4 it behaves more like f4.5 because of its pupil ratio so any differences in resolution have more to do with effective aperture than with optical flaws. It is a great lens and I always get good results with it. It is the most expensive of the lenses on this test.
Best surprises have been both noname lenses (I do not know who made them or with what purpose) and the JML 50mm 2.8 (it says JML but I do not think it is a JML optical lens). All of them have good resolution and contrast and good working distances so they can be used both in the studio and in the field; the JML 50mm is going to be part of my field gear.
Working distance (WD) can be an important aspect of this kind of lens, in the studio is not that important but for field work it is important.
The WD of these lenses are:
JML 50MM -------------------------------42.5MM
APO RODAGON 50MM--------------- 42.5MM
RODAGON 50MM ----------------------40.5MM
OLYMPUS 38MM ----------------------36.5MM
MITUTOYO 5X -------------------------33.5MM
NONAME I ------------------------------32.0MM
MIKROTAR 30MM --------------------30.0MM
NONAME II ----------------------------28.0MM
NIKON BD 5X -------------------------24.0MM
FUJI 24X------------------------------- 22.0MM
JML 21MM-----------------------------16.5MM
JML 26MM---------------------------- 16.5MM
JML 21mm and 26mm have very short WD, so they are not that good for field work. The Mitutoyo has a very good WD but it is quite big and needs a tube lens so it ends up being quite bulky and I would not use it in the field.
Best lenses for field work are the JML 50mm 2.8 and the Olympus 38mm, this last one's tip is cone shaped which makes lighting quite easy. The Noname lenses also would work well but they have no aperture ring.
There are many lenses that could be in this test, like Canon's MP-E 65mm (which would be among the best) but can only be used on canon bodies or the Nikon 4/0.20 plan APO that would outresolve all these lenses in the center of the frame but would probably fail in the corners (remember, this is a full frame test); many enlarger lenses could be in this test too.
If I could only keep one of this lenses probably I would keep the Olympus as in my opinion it is well balanced; works well in the studio and in the field, has an aperture ring, has good working distance, color and contrast, a cone shaped tip and it is CA free
If I had to keep one lens for the studio that would be the JML 21mm, has excellent color, contrast and resolution and it is CA free too. The advantage over the mitutoyo is that it has better corner coverage and it can be used on fullframe bodies from 3X
Flash tool to compare images
A good way to view/compare images is with this tool developed by John Hallmen
This tool only allows to view files with 3000px on the long side, so I did some crops that have the center part and the top left corner (for corner pixel pipping)
You can see that images are quite similar, quality is of a high standard
Both lenses give high quality output (even that JML 26mm is working with a higher effective aperture) but in some high contrast areas you can see the JML 26mm shows some CAs (nothing serious)
This images can be replaced copying any of the links bellow and swapping them pressing "change image" and pasting the link from the image you want to compare
Pmax
APO Rodagon http://farm7.static.flickr.com/6238/6308819213_13cdb88087_o.jpg
Rodagon http://farm7.static.flickr.com/6219/6309340468_57777cce0a_o.jpg
Olympus 2.8 http://farm7.static.flickr.com/6031/6308822057_f15bb5c8bc_o.jpg
Olympus f4 http://farm7.static.flickr.com/6098/6309343738_ba93d2b020_o.jpg
Noname I http://farm7.static.flickr.com/6237/6308826751_43e6ebe634_o.jpg
Noname II http://farm7.static.flickr.com/6060/6309345286_c9c70129ea_o.jpg
Nikon BD5 http://farm7.static.flickr.com/6059/6308828181_82d89b64e8_o.jpg
Mitu 5X http://farm7.static.flickr.com/6227/6308829829_7f5353ddf7_o.jpg
Mikrotar http://farm7.static.flickr.com/6232/6308831253_00afee58aa_o.jpg
JML 50mm http://farm7.static.flickr.com/6225/6308832733_f07d386c10_o.jpg
JML 26mm http://farm7.static.flickr.com/6045/6309354456_61c6fcd785_o.jpg
JML 21mm http://farm7.static.flickr.com/6118/6309356036_638d1a88c6_o.jpg
Fuji 24x http://farm7.static.flickr.com/6037/6308837925_7a1fd98aa3_o.jpg
Now some examples from the second round
In my opinion this two lenses are the best ; which one is best? I do not really know. The Mitutoyo might resolve more detail but shows some CAs that probably come from the morfanon (easy to fix in Camera Raw). If you have any of these you do not really need any other lens from the test (at least for studio work)
Here the mitutoyo proves to be a better lens
Once again the Mitutoyo is better but the Noname performs quite well for an unknown lens
The mitutoyo shows once again a better behaviour, the olympus may be affected by a higher effective aperture because of its pupil ratio of 0.89 so it behaves like a f4.5 (aprox) lens. I have to say that the Olympus is one of my favorite lenses and I always get good results with it.
The people who designed it knew what they were doing and people from canon could learn something from its design (cone shapped tip)
I still had to see if the Olympus worked better at f2.8 (because of pupil ratio something like f3.2)
In my opinion even if it resolves more detail (which is not clear) is not worth the loss of contrast, the higher CAs and the higher number of shots needed
I might be wrong but I would say the JML 50mm resolves more detail than the Olympus 38mm; it has 8 iris blades for 6 of the Olympus and around 6mm more WD (with a little M42 hood this advantage dissapears). Javier Replinger has used it at 2-3X and says corner performance is good so it can be a nice lens for the studio and for field work.
The Olympus is a good lens but there seems to be better options; however when I have to take a picture in this range my eyes always look at it and there must be a reason for it.
Once again bellow you have the links so you can compare images as you wish.
Dmap
JML 21mm http://farm7.static.flickr.com/6047/6308924439_8f07598853_o.jpg
JML 50mm http://farm7.static.flickr.com/6109/6308926095_c458acb4f4_o.jpg
Mitu 5X http://farm7.static.flickr.com/6115/6309448234_56ceac5687_o.jpg
Noname II http://farm7.static.flickr.com/6041/6308929963_14e79f57a9_o.jpg
Olympus f4 http://farm7.static.flickr.com/6060/6308931667_6d7dc2ddf8_o.jpg
Olympus f2.8 http://farm7.static.flickr.com/6220/6308933275_ac37d24911_o.jpg
One last note
This kind of test run on flat subjects is good to show resolution all over the frame and corner performance. On a subject with volume and hairs (like a fly head) lenses like the Olympus or the JML 26mm could perform very well because of higher apertures. So less shots needed but specially less halos and less transparencies. John Hallmen also did notice some telecentrity on the Mitutoyo 10X and maybe the Mitutoyo 5X behaves in the same way (I did run a test with scale set to 0 on Zerene Stacker and it worked well, but it is still inconclusive)