Posts: 8,003
Threads: 1,851
Joined: Apr 2010
Reputation:
49
[quote name='turboman' timestamp='1294630161' post='5399']
I have a Sony NEX-5 with 18-55mm kit and I am looking for a good 28mm for creative work, like a Leica 28mm Elmarit or Zeiss 28mm Biogon.
I am trying to find out if a good lens can improve performance, or if performance is limited by the sensor.
As a beginner, in order to get a general understanding about performance data, I started to compare lens reviews form different sites, and also manufacturers data. Getting more and more confused. Following is an example:
- The Photozone review Lumix GF1 with Zeiss PlanarZM50mm shows half resolution in the corner at F/2, compared to center.
- The Zeiss website shows that in the corner (u=11mm on 4/3 system), resolution is more or less the same as in the center.
[url="http://www.zeiss.com/C12567A8003B8B6F/EmbedTitelIntern/Planar2.0_50mm_ZM_e/$File/Planar2.0_50mm_ZM_e.pdf"]http://www.zeiss.com...0_50mm_ZM_e.pdf[/url]
Conclusion would be that the corner performance is limited by the sensor, but this is contradicted by the lens review with the Leica 45mm Elmarit with good corner performance on the same camera.
[/quote]
Well, the difference is that the manufacturer data shows the naked lens data whereas we (as well as other sites) show the performance of the system (lens + camera + RAW converter).
[quote name='Klaus' timestamp='1294640488' post='5400']
Well, the difference is that the manufacturer data shows the naked lens data whereas we (as well as other sites) show the performance of the system (lens + camera + RAW converter).
[/quote]
I understand this. That is why I cannot explain the obervations I made, repeated here again:
- Zeiss data (only lens) show same resolution in corner for 4/3 sensor size.
- Photozone (lens Zeiss + NEX-5 sensor)show much lower corner resolution.
Simplistic conclusion would be that NEX-5 sensor has low corner resolution.
But this is contradicted by a high resolution of Leica45Elamrit+NEX5sensor. It is unlikely that the Elmarit could correct the "simplistically perceived low NEX5 corner resolution".
Am I digging too deep?
Posts: 8,003
Threads: 1,851
Joined: Apr 2010
Reputation:
49
[quote name='turboman' timestamp='1294751912' post='5418']
I understand this. That is why I cannot explain the obervations I made, repeated here again:
- Zeiss data (only lens) show same resolution in corner for 4/3 sensor size.
- Photozone (lens Zeiss + NEX-5 sensor)show much lower corner resolution.
Simplistic conclusion would be that NEX-5 sensor has low corner resolution.
But this is contradicted by a high resolution of Leica45Elamrit+NEX5sensor. It is unlikely that the Elmarit could correct the "simplistically perceived low NEX5 corner resolution".
Am I digging too deep?
[/quote]
MTF curves are dependent on a certain target frequency. The manufacturer MTFs have surely a different frequency than the ones in the tests. You may also notice that some manufacturer MTFs have no strict downward trend - they may have local peaks. This may also explain the difference between the MFT and NEX test due to the different sensor size and consequently different test zones. However, I don't really see that the results are so different when looking at the center and border data only actually. We haven't tested the corners on the NEX. That said I would have expected a more even center-to-corner behavior - agreed.
The Leica 45 is a dedicated design for MFT and it's slower which makes things much easier for the designers.
genotypewriter
Unregistered
[quote name='turboman' timestamp='1294751912' post='5418']
I understand this. That is why I cannot explain the obervations I made, repeated here again:
- Zeiss data (only lens) show same resolution in corner for 4/3 sensor size.
- Photozone (lens Zeiss + NEX-5 sensor)show much lower corner resolution.
Simplistic conclusion would be that NEX-5 sensor has low corner resolution.
[/quote]
Sensor design is uniform across its area except in the case of the M8, M9 and the recent Fuji X100 sensor where they have the off-set micro-lenses. So other sensors can't possibly have absolute resolution figures that differ in the corners from the center.
Like Klaus said, PZ numbers are from a camera system and not just the lens alone. Leaving experiment errors aside, it's possible that one lens suits a sensor better than another. For example, one lens might be more tele-centric than the other.
However, how the PZ "border" and "extreme border" values are fairly meaningless in practice as I have explained below:
http://www.flickr.com/photos/genotypewriter/5207301499/
GTW
Posts: 8,003
Threads: 1,851
Joined: Apr 2010
Reputation:
49
[quote name='genotypewriter' timestamp='1294950954' post='5481']
However, how the PZ "border" and "extreme border" values are fairly meaningless in practice as I have explained below:
[url="http://www.flickr.com/photos/genotypewriter/5207301499/"]http://www.flickr.co...ter/5207301499/[/url]
GTW
[/quote]
Sorry but that's merely an opinion but not a fact. If a wide or ultra-wide lens has a poor corner performance this is of course relevant. Possibly not for you but it is THE primary criteria for me here. Unlike a tele lens the wide- or ultra-wides have to provide a uniform performance across the frame IMHO. In fact I don't even care for the center performance here - it is always "good enough" here anyway.
genotypewriter
Unregistered
[quote name='Klaus' timestamp='1295013502' post='5499']
Sorry but that's merely an opinion but not a fact. If a wide or ultra-wide lens has a poor corner performance this is of course relevant. Possibly not for you but it is THE primary criteria for me here. Unlike a tele lens the wide- or ultra-wides have to provide a uniform performance across the frame IMHO. In fact I don't even care for the center performance here - it is always "good enough" here anyway.
[/quote]
Of course, the entire frame is important... my comment was about how PZ chooses "border" and "extreme border" on the image circle radius where they are only representative of a minute spatial area on the entire frame.
Again... this is about:
1. Selecting only 3 points.
2. Two of those points being less important, spatially, on the image frame
I've seen many lens MTF curves dip in the middle, at around the 12mm mark, where the significance of the MTF to the image frame is at its highest. These lenses may have good dead-"center" and "extreme border" values but the final image resolution will be less than of a lens that has low "center" and "extreme border" values but a high MTF at around the 12mm mark.
GTW
