Spyro P. wrote:
Grnmgfh... the crop factor is killing it. A lens with 75mm field of view and 35mm depth of field is not exactly the combination I was dreaming of.
What M4:3 needs is dedicated primes for the system, not legacy lenses.
But with proper focusing rings.
If zeiss made a small 14mm lens with a small image circle, infinity stop and distance markings I'd buy two M4:3 bodies.
Yes. Crop factors do kill these great lenses. But I would say that the Zeiss G 21 and 28mm Biogons are incredible over the crop area. The corners should be as good as the center. Good, if a bit slow (f2.8) standard lenses. Would be interesting to see how the 21mm Biogon performs relative to the Panny 20/1.7
The m4/3rd system does need more good compact primes. We need a manufacturer to consider that it is a serious and viable system camera rather than a novelty. The more of a novelty they consider the cameras, the fewer good prime lenses we'll get. Olympus may be serious, There is a rumoured "pro" version of the EP-1. Given some of the deficiencies of the EP-1 I might wait for that one and see what they have up their sleeve. Other than for Zeiss or certain niche producers (Cosina, Schneider, Leica) it might be tough for them to see giving us "proper" focusing rings, but we can hope.
Frankly, I'm waiting to see what if anything, Canon, Nikon, Sony and/or Samsung/Pentax (rumoured) respond with. I'd prefer the noise characteristics of a microAPS-C sensor format -- Plus I'd get a little more out from the image circle my Contax G lenses.
That was at F2.8 shot, actually I re-shot the same scene without UV/IR cut filer. And more with
G-Biogon 21/2.8 and CV Ultron 28/1.9. The results are NOT good at all, "the wider the worse".
Lotusm50 wrote:
Yes. Crop factors do kill these great lenses. But I would say that the Zeiss G 21 and 28mm Biogons are incredible over the crop area. The corners should be as good as the center. Good, if a bit slow (f2.8) standard lenses. Would be interesting to see how the 21mm Biogon performs relative to the Panny 20/1.7
The m4/3rd system does need more good compact primes. We need a manufacturer to consider that it is a serious and viable system camera rather than a novelty. The more of a novelty they consider the cameras, the fewer good prime lenses we'll get. Olympus may be serious, There is a rumoured "pro" version of the EP-1. Given some of the deficiencies of the EP-1 I might wait for that one and see what they have up their sleeve. Other than for Zeiss or certain niche producers (Cosina, Schneider, Leica) it might be tough for them to see giving us "proper" focusing rings, but we can hope.
Frankly, I'm waiting to see what if anything, Canon, Nikon, Sony and/or Samsung/Pentax (rumoured) respond with. I'd prefer the noise characteristics of a microAPS-C sensor format -- Plus I'd get a little more out from the image circle my Contax G lenses.
Note that Panasonic has already announced as many m43 normal-range primes (3, the 14 which is announced and the 20 and 45 which have shipped) as Olympus has 4/3rds normal-range primes (also 3, the 25 pancake and 35 and 50 macros).
KOJI wrote:
That was at F2.8 shot, actually I re-shot the same scene without UV/IR cut filer. And more with
G-Biogon 21/2.8 and CV Ultron 28/1.9. The results are NOT good at all, "the wider the worse".
Wow. that sensor really doesn't like anything that isn't hitting it at 90 degrees. I didn't think that the angle of light from the center 50% from the 21 Biogon would not be all that acute. It now looks like Leica/Kodak pulled off quite a feat with the full frame sensor in the M9. It seems to handle this sort of thing a lot better.
What these samples do show from the center crops, however, is how sharp the 21mm Biogon is, particularly in comparison to the CV28 Ultron (both wide open, but f1.9 vs. f2.8).
What these samples do show from the center crops, however, is how sharp the 21mm Biogon is, particularly in comparison to the CV28 Ultron (both wide open, but f1.9 vs. f2.8).
I shot CV Ultron 28/1.9 @ f2 since I knew it is a bit soft @ f1.9, otherwise I agree with your statement above. I did test with CV-SWH 15/4.5 and CV-Nokton 35/1.2, they gave us
much the same results. I wonder how G-Planar 35/2 performs on GF1, I love this lens somehow even though its wide open performance is not stellar.
Thank you for getting back on the topic KOJI.
Your results are well in line with earlier discussions from another board with a more active µ4/3 section. There is an obvious difference between the CV28/2 and the older CV28/1.9 (the later having been reported earlier as "no corner smear").
While not being super sharp the little Panny 20/1.7 seem to be quite an achievement. During the time between announcement and materializing I was afraid Panasonic had promised more than they could keep.
You saved me from ordering the adapter (I think, the Zeiss G 45/2 still may work very well on a G/GH1 camera). I'm sorry you didn't get better results. It would have been wonderful if the small Zeiss lenses could have been used with confidence.
Lotusm50 wrote:
The R-D1 works pretty well in the corners with the CV 15/4.5, 21/4, and 28/1.9. Sure it has bigger pixels, but they're also further away from the center of the lens. The bigger problem is light falloff with the 15 and 21, but even that isn't a big deal.
Lotusm50 wrote:
Hmm. So if the Leica can do it and the Epson RD-1 can do it, what's the problem with the sensor in the Panny G1, and presumably the Oly E-P1??
Well, that's a difference, but not necessarily the problem...
That said, the R-D1 doesn't have fancy offset microlenses like the M8 (and presumably M9), but a plain old D100 sensor (which is pretty damn impressive).
Then again, the 12mp M43 cameras have a 5MP/cm^2 density while the R-D1 has a 1.6MP/cm^2 density.... the R-D1 pixels are over TRIPLE the size.
And the post by the suede, and mentioned earlier: Panasonic and Olympus stack a supersonic filter on top of the AA and UV thingies in there, and the microlenses are probably not positioned in an "offset" manner.
What I would like to see is a good 10-12mm prime for the µ4/3 system. It doesn't need to be very fast but it should perform as well as the ZD7-14. Then the system would have decent primes; a wide, a normal and a short tele (yes, that's my needs).
...and right now I read that Olympus will release a µ4/3 9-18mm lens to be for sale sometime before next summer. OK then. It's not a prime, it's not fast but it sure look small and neat. BJP
The edge smearing is the result of spherical aberrations introduced by glass layers above the sensor (AA/IR filters and dust shaker). A simple explanation and illustrations of this effect could be found in the pdf from this link http://tinyurl.com/ydqsz2s. This pdf explains why Biogon 38mm is not recommended by Hasselblad for the use with digital backs.
The amount of the smearing depends on the distance of the lens exit pupil from the sensor. For the symmetrical lenses (and many RF lenses are symmitrical) this distance is simply the focal length. More generally, this distance could be calculated as the focal lengths multiplied by the exit pupil diameter and divided by the entrance pupil diameter. These diameters could be measured just be looking through the lens from both ends. Zeiss used to publish this data, so they are available for all Contax G lenses.
This distance should be compared to the frame diagonal which is 22.5 mm for m4/3. It is not very clear what is the good value for this ratio. As a reference point I included the data for Olympus 17mm which is the only m4/3 lens I own. It would be interesting to see the data for other lenses.
Lens Distance Ratio
Biogon 2.8/21 22.4mm 1
Biogon 2.8/28 26.6mm 1.18
Planar 2/35 51.6mm 2.29
Planar 2/45 53.1mm 2.36
Olympus 2.8/17 35mm 1.6
The 35mm and 45mm are in good shape. The numbers for the wide angle lenses are not promising.
We should expect that in practice we may see deviation from this theory. For example, a certain lens might have residual aberration which have the opposite sign and, at least partially, cancel the ill effect of the filter layers. In fact, I wouldn't dismiss the possibility that the m4/3 lenses are designed to have such residual aberrations.
I think Alex is onto something here. I had been assuming that the edge smearing issue was primarily an issue with the wide Leica M mount lenses and was caused by the lens' rear element being so close to the sensor. However, using Alex's approach above, this can be extended to adapted SLR lenses. I have the Oly 21/2 and have used it a bit on my E-P1 via a EOS adapter. I hadn't done any tests on the lens, but I worked out the ratio as Alex outlines (based on a block diagram since I don't have the lens in front of me). The Oly 21/2 ends up with a value of ~.5 which would indicate an edge smearing issue. As I said I haven't tested the lens...brick walls are in short supply ... but I did find this one shot that looks suspiciously like smearing.
How could an SLR lens have an exit pupil closer to the sensor than the flangeback distance of the system it's from?
The value for the 21/2 should be more than 1, not less than 1. In fact, the flangeback for the OM system is 46mm, so the ratio can't be less than 2.1 for any of the lenses.
My mistake. I erroneously assumed that the entrance and exit pupil diameters would the diameters of the elements. A little more research would have given me the correct answer, and obviously you need to actually have the lens in your hands to measure them. The 21/2 has an exit pupil diameter of ~21mm an entrance pupil diameter of ~10mm. This computes to a ratio of 1.96 so any edge issues in my shot can probably be chalked up to bad technique.
I should note that using these diameters give a distance to the sensor of 44mm, which seems about right since the rear element seems to be about even with the surface of the EF adapter that's attached.
Alex, thank you. Now let's see if theory and practise is the same in this case. Like CVickery I think you are on the right track. Maybe the exact numbers need to be adjusted but it all sounds plausible to me. Well, most complicated stuff do that as I only understand half of it. But here it also fits well with my experience.
Jonas,
The exit pupil is the lens aperture as it appears from the back side of the lens. For all practical purposes the position of the exit pupil relative the focal plane determines the angle of the light when it strikes the sensor.
The procedure to determine this position is actually quite easy. You look through the back of the lens and measure the diameter of aperture as it appears from the back side (closing the aperture a little may help to identify the boundary). It is the diameter of the exit pupil. The same measurement from the front side will give you the diameter of the entrance pupil (don't change the aperture between measurements). You divide the exit pupil by the entry pupil and get the pupil magnification (typically it is >1). You multiply the focal length by this number and get the distance between the focal plane and the exit pupil.
You mentioned two CV 28 which behave differently. If you have access to these lenses it would be interesting to get the numbers. In fact it would be interesting to get numbers for all widish RF lenses.
Thanks again Alex. I think I understood how to measure the pupil sizes. What I don't understand exactly is how the limit is calculated. But it doesn't matter a lot - I just like that you presented a possible, probably, working model for how to estimate the lens' functionality in this case.
Unfortunately I don't have any of the lenses. I returned the CV28/2 after having seen the result I posted at page 1. KOJI posted a sample from the CV28/1.9 (a link) at page 1 as well. There clearly is a difference and yes, it would be interesting to learn about the entrance and exit pupils for a comparison. Maybe some friendly soul over at a rangerfinder forum could help with this?
... but I did find this one shot that looks suspiciously like smearing.