Very interesting, Martin. All the more so as the M8 has a larger sensor than the 5N, so one would have expected the opposite, meaning that field curvature would be more visible on the larger sensor, not less.
So you are saying that pixel density doesn't matter regarding lens performance?
philber wrote:
Very interesting, Martin. All the more so as the M8 has a larger sensor than the 5N, so one would have expected the opposite, meaning that field curvature would be more visible on the larger sensor, not less.
Exactly, and you probably know how much better some lenses perform on the M9 than on NEX-7, for similar reasons. If I'm not mistaken, the D800E has a very thick filter infront of the sensor, because of the double "inverted" AA filter. Maybe that could play a role?
So you are saying that pixel density doesn't matter regarding lens performance?
I don't think higher density per se would do anything, except giving better performance when scaled down. But the micro lenses could of course do things with both sharpness and color casts.
Makten wrote:
The resolution might not change lens character, but other things definitely can. I used the Summicron 28/2 ASPH on an M8 for quite some time and it was awesome. Then I also used it on a NEX-5N, and suddenly there was a terrible curvature of field. Most likely due to a much thicker UV/IR-filter infront of the sensor, changing the optical path from the exit pupil on lenses where it is close to the sensor.
This could also be part of the explanation that Bobu suddenly got mushy corners with the Zeiss 21/2.8 on D800 but not on 5D MkII. It doesn't have to depend on the pixel density. Definitely not if the difference is seen at the same viewing size....Show more →
Indeed many lenses have been criticized for their bad corner performance, even DSLR lenses, and in many cases the thick sensor cover is the culprit. One of the cases is the Sony A900 with its thick cover that also is further away from the sensor (to minimize the dust effect as Sony claimed). Of course some lenses are affected more than others depending on their optical formula and the exit pupil position in relation to the sensor.
The M8 has also shifted micro lenses besides the almost inexistent sensor cover, but no other non Leica camera has these except the Ricoh GXR. So it is not at all starnge that the 28 cron behaves better on the M8.
However, I would not call this field curvature. It is just lens/sensor combination. Field curvature is a lens characteristic that will show up on any sensor/film medium regardless of the cover or micro lenses.
philber wrote:
Yet another. I compared 3 extremely highly reparded lenses on my NEX 7, the highest density reasonably-sized sensor around: the Contax G 45, Summilux 50 and Elmarit R 60. The results are crushing for a lens that was considered one of the best ever designed.
Somewhat offtopic, but what exactly were the results of this comparison?
edwardkaraa wrote:
I would not call this field curvature. It is just lens/sensor combination. Field curvature is a lens characteristic that will show up on any sensor/film medium regardless of the cover or micro lenses.
I don't see how you could call it otherwise… It's a curve in the focus field. Now, it's true that it's caused (mainly) by the sensor (I noticed in quite a few lenses), but it's equivalent to the optical field curvature. I'm not sure it makes sense to distinguish both fenomena.
sflxn wrote:
Are we talking about the same funky patterns? I haven't kept up with the discussion on the x-trans sensor. Last year, there was a discovery of a funky geometric pattern you sometime get in the shadows. It was very geometric and artificial. I was reviewing photos in the x-pro 1 thread, and I believe I can still see them. I have also noticed the bad watercolor effect and 'dirty' shadows. The shadows are especially bad compared to the Sony sensors. I am not sure if all of these are related; however, I am am currently on the side of believing these are inherent problems with the x-trans sensor and can never be fixed without computational interpolation (making up data) on the side of the raw processor. I am hoping this changes over time. Perhaps the demosaicing algorithms need more bake time. I'm not sure. What I am sure of is I really do not like what I see in the Fuji image threads. The sensor or the raw processing... I don't know which is at fault, but I need a reliable solution today. I am not going to go risk a lot of money to "try it myself." When I look at images taken with Sony sensors, they're so clean with lots of details in the shadows. It's up to Fuji to fix this problem if they want my money again....Show more →
Yes, we are talking about the same watercolor funkiness. And yes, I have yet to see it in Aperture-processed RAW files. The tradeoff is one gets high-contrast, high color saturation, and a lot of color artifacting, and moire.
Makten wrote:
It definitely is, even if it's not a property of the lens, but rather the system. What else to call a curved field?
To be honest I don't know the technical name to describe the light rays being bent by the sensor topping or unable to reach the sensel due to extreme angle. But I always thought field curvature is a lens related term. I might be wrong though so maybe someone more knowledgeable than me can chime in.
edwardkaraa wrote:
To be honest I don't know the technical name to describe the light rays being bent by the sensor topping or unable to reach the sensel due to extreme angle. But I always thought field curvature is a lens related term. I might be wrong though so maybe someone more knowledgeable than me can chime in.
As long as you don't blame the lens, I see no problem with using "curvature of field". The problem is that you can't know if it's because of the lens or not, if you don't try it on several cameras. This is why I think we should be careful when claiming things like that smaller pixels reveals flaws in the lens, because it could as well be flaws of the system.
Makten wrote:
As long as you don't blame the lens, I see no problem with using "curvature of field". The problem is that you can't know if it's because of the lens or not, if you don't try it on several cameras. This is why I think we should be careful when claiming things like that smaller pixels reveals flaws in the lens, because it could as well be flaws of the system.
Apart from the curvature of field term, I fully agree. More often than not, it's the system not the lens. Zeiss MTF measure the lens performance independently from sensor/film, so at least at infinity, we know the real performance, and any discrepancy is probably due to the system.
I think astigmatism is the proper term for sensor toppings causing the issues. If I remember correctly, that is the term Zeiss used in their wide angle white paper.
douglasf13 wrote:
I think astigmatism is the proper term for sensor toppings causing the issues. If I remember correctly, that is the term Zeiss used in their wide angle white paper.
Hi Douglas. Yes, now that you mentioned it, I believe you're right.
douglasf13 wrote:
I think astigmatism is the proper term for sensor toppings causing the issues.
That might be true, but you must distinguish the cause and the issue. The issue is clearly curvature of field, but the cause could be anything from a warped sensor to an optical property of the lens design.
A flat tyre is a flat tyre both if it was caused by not filling it or by it being punctured.
Head to head, Touit vs Fuji 35 and 14mm comparison. These were taken from the same spot instead of moving a little closer for the same framing with the Touit lenses, so it's not 100% equal, but you can still see how they render and the sharpness at the pixel level.
To my eyes at least, the Fuji lenses look to be superior to the Zeiss in most every way. Sharper at wider apertures (though the 12 and 14 are very close when stopped down, but the 14 still looks slightly sharper to my eye), but the Fuji 35 has MUCH better bokeh than the Touit 32mm (and what looks to be noticeably more blur, even at the same aperture).
douglasf13 wrote:
I think astigmatism is the proper term for sensor toppings causing the issues. If I remember correctly, that is the term Zeiss used in their wide angle white paper.
astigmatism is when tangential and sagital focus planes change their position relative to each other and is a type of curvature of field. if both tangential and sagital focus planes curve together than you have field curvature with no astigmatism. both are properties of optical systems not sensors. whether one is caused by the lens or the sensor toppings does not change its name or what it is. since sensor cover glass is basically just a flat refracting material it will offset rays that come at a more oblique angle more than rays that come at a perpendicular angle. thus the offset increases as you move tangentially from the center causing a greater shift in the tangential focus plane than the sagital giving you astigmatism.
Jman13 wrote:
Head to head, Touit vs Fuji 35 and 14mm comparison. These were taken from the same spot instead of moving a little closer for the same framing with the Touit lenses, so it's not 100% equal, but you can still see how they render and the sharpness at the pixel level.
To my eyes at least, the Fuji lenses look to be superior to the Zeiss in most every way. Sharper at wider apertures (though the 12 and 14 are very close when stopped down, but the 14 still looks slightly sharper to my eye), but the Fuji 35 has MUCH better bokeh than the Touit 32mm (and what looks to be noticeably more blur, even at the same aperture).
Wow, there is such a striking difference in favor of the Fuji lenses, that it makes me wonder what the issue is with the Touit's. Part of it relates to differences in focus, at least with the 12 vs 14 where the last 2 images are front-focused. I don't understand that as it's CDAF and there should be no focus problems. But even when looking at material in the focal plane, it's almost like in-camera sharpening was ramped up with the Fuji's and not with the Touit's, or something to that effect.
I'm looking forward to some images on a NEX-7 where it's more personally relevant. I see what everyone is talking about regarding the Fuji moire, with all of the color aliasing artifact in the branches on the 14mm samples, I've got enough of that with the Ricoh GXR. But if the Touit's indeed perform like this on a NEX body, they are worlds behind most rangfinder glass, at least on a NEX-6. That's why I remain suspicious until we see more samples in the wild.
uscmatt99 wrote:
I see what everyone is talking about regarding the Fuji moire, with all of the color aliasing artifact in the branches on the 14mm samples, I've got enough of that with the Ricoh GXR.
Yep, that's pretty much a perfect example of the trade off by using Aperture with X-trans. You occasionally get these purple/green areas with little white dots. Overall, I still find it preferable to the painterly effect from LR, which seems to happen more often.
The problem is that you can't know if it's because of the lens or not, if you don't try it on several cameras. This is why I think we should be careful when claiming things like that smaller pixels reveals flaws in the lens, because it could as well be flaws of the system.