kosmoskatten wrote:
...In my experience I have never seen a solution where focusing past "true" infinity of a lens has produced acceptable results, quite the opposite. Stopping down, yes, focusing past infinity no. Especially if you could and would have to focus some ways past infinity...
Hi. I wasn't meaning to imply that focusing past infinity on the 5N was a worthwhile solution in normal use. I was just wondering how far past infinity that Sam needed to focus to sharpen up those corners. With the CV 35/1.4 on NEX, focusing the center past infinity very much sharpens up the corners at, even when stopped down, and that is indicative of a field curvature issue.
Henrik - I think you can get solid results on the NEX by focussing 'deep' with field curvature however the margin of error (e.g. going too deep and losing central DOF and sharp corners) is pretty fine though.
Having said that, F8 is a great equalizer!
I know the CV 15mm can show good results to the edges with the NEX (i'm a little envious of Ron's copy as it appears to be very consistent wide open on the 5N) but the Ricoh may simply achieve it at wider apertures and more consistently.
I'm being critical of the CV 15 and the 5N a bit in my post, and highlighting the weaker areas but honestly for landscape stuff its working better for me than any modern UWA zoom i've used (i've not used the monster Nikon 14-24 though).
Given I usually shoot at F8 and am a tripod fan i'm happy to roll with what I have for now, although i'm hoping to head off to Asia for a while next year to work on some travel shooting - for that i'd like to go very lightweight and get technical results at those wider apertures.
To me that means an A16 module, a lens upgrade or ideally... both
After seeing Rechmann's new M9 vs. NEX-7 comparison, Sam's test is striking me as more and more odd. In MR's center crops, I'm not sure how the full frame M9 and aps-c NEX-7 can be so similar in center resolution, despite one being FF with no aa and the other having a few more megapixels, yet, in the test in this thread, the 5N isn't even coming close to the GXR in the center. I'm wondering if maybe sharpening isn't optimized in this test?
In case anyone is interested in my results with the 5N & CV15. But my images might be difficult to compare to Sam's. I didn't shoot anything at infinity.
I agree with Sam that my results with the CV15 on the 5N look a lot better at wider than f/8 apertures than I've ever seen with my Canon L wide zooms (16-35 v.1 & II). I feel I could shoot the CV15 wide open on APS-C and be very happy with the results, and going beyond 5.6 doesn't really appear to increase sharpness.
Regarding the representation of field curvature: obviously the field curvature of the lens doesn't change from body to body, rather the interpretation of the lens's rendering is different from sensor to sensor. Without the AA filter, it could be that the A12 sensor is resolving edge details/contrast better than the 5N, which may be afflicted by the dual impact of AA softening and increased astigmatism (though then the blur would be much busier). While the resolution difference between the two sensors is not great, I do wonder how much of an influence it has on the appearance of depth of field/blur. Higher resolution should record more subtle blur transitions while a lower resolution sensor will clip those subtleties into a more high contrast interpretation. And viewing both at 100% means we're enlarging the image more with the 5N, which should reveal more softness in fine details.
It would be interesting to see the 5N scaled down to 12MP for perhaps one of those scenes. Maybe the last one of the tree because it has the greatest depth of field range.
Doug, the part of the answer might be in the manner in which Sam has sharpened the images:
Sosua wrote:
All images were processed with my standard minor capture / deconvolution sharpening settings in Lightroom 3 (Amount 35 / radius 0.5 / detail 100 / threshold 15). These ended up being a little strong for the Ricoh and pretty weak for the Sony. No disortion, fringing or vignetting correction at all.
Michael Reichmann:
I debated the issue of sharpening. The M9 does not have an AA filter, and therefore theoretically, at least, requires less sharpening than the NEX-7. But, sharpening is something that one does by eye and by feel, so there is a totally subjective aspect that could skew the test. In the end, for better or worse, I left the images unsharpened.
Another factor is that if I interpret Reichmann's methodology correctly, he cropped the FF sensor of the M9 to effectively 12MP to cover the same area as APS-C. To show the same subject size in the 100% crops, the full NEX-7 image would have had to have been down sampled to 12MP from 24MP, which is a HUGE jump that will eliminate a lot of not so nice details.
Maybe I misunderstood how it did it (because he constantly referred to resizing the NEX to 18MP, which in fact he should have stated 12MP).
By showing us 100% native crops from both, Sam is showing the 5N sensor resolution for what it is natively.
Very true, Ron. Using the same sharpening for cameras with and without an AA filter is akin to putting the same racing fuel in diesel and gasoline engines. One really has to meticulously optimize sharpening for each before we can get an idea of the difference in real usage (as well as resize to the same size.) Thanks though, Sam, for getting this ball rolling. I know it's a ton of work.
douglasf13 wrote:
After seeing Rechmann's new M9 vs. NEX-7 comparison, Sam's test is striking me as more and more odd. In MR's center crops, I'm not sure how the full frame M9 and aps-c NEX-7 can be so similar in center resolution, despite one being FF with no aa and the other having a few more megapixels, yet, in the test in this thread, the 5N isn't even coming close to the GXR in the center. I'm wondering if maybe sharpening isn't optimized in this test?
this should answer your question:
Michael Reichmann wrote:
One other difference from the last test – I decided that since what people were asking for was a "real world" comparison, then that's the way I would treat the files. Both files were loaded into the latest Lightroom and processed using my normal technique, and that includes sharpening. You wanted "real world", you got real world. At least, my real world.
he optimized sharpening for each camera prior to downsizing the NEX-7 image.
That's referring to "Part II" of his NEX-7 rolling review. The original 7 vs. M9 images were done with no sharpening.
The Part II images though are interesting for the similarities. And related to the discussion about moire problems here, his images seems to show that the NEX-7 isn't that much different from the M9 for that scene (spiral staircase railing).
As per my disclaimer, sharpening for these was equalized rather than optimized - also, these are handheld in flat light at ISO 200 - both cameras can certainly 'do better'.
The 5N sharpens best from ISO 100 (where its smoother than the GXR) and in more contrasty light does a great job:
100% crop of Summaron at 5.6 with more aggressive sharpening:
My main interest in the Ricoh is performance at the periphery and the handling - the 5N would look just as good in sharpened prints over 90% of the frame.
Douglas, Ok, I got you now.
Personally I always prefer files that need less work in post pro. With the NEX5n and the GXR M I think the final output is so close that it is more or less a toss up between the two. The only real merit for the GXR M is the package it comes in and the better handling of wide angles and that it comes out requiring a lot less sharpening, which I find helps with small detail rendition and easier to fine tune. The smearing is what takes the NEX5n down a notch for me, and the handling.
Optimized sharpening is always a preference for me as there are three things that interest me: out of camera RAW quality, how much tweaking is needed (and can be done) in post pro, and final output. If two solutions produce a somewhat equally excellent output but one gets there faster I would prefer it if all else was equal.
For both cameras I think that the wide angles are the crunch factor; it is hard getting wide angles/sensors to resolve minute landscape detail as comfortably or nicely as full frame digital / medium format does. I get the feeling details seem a little compressed/crude.
For close up work with wide angles I seldom get the same feeling, usually they work great.
Ed; might be happening soon. Not too sure on the optical RF though.
kosmoskatten wrote:
Douglas, Ok, I got you now.
Personally I always prefer files that need less work in post pro. With the NEX5n and the GXR M I think the final output is so close that it is more or less a toss up between the two. The only real merit for the GXR M is the package it comes in and the better handling of wide angles and that it comes out requiring a lot less sharpening, which I find helps with small detail rendition and easier to fine tune. The smearing is what takes the NEX5n down a notch for me, and the handling.
Optimized sharpening is always a preference for me as there are three things that interest me: out of camera RAW quality, how much tweaking is needed (and can be done) in post pro, and final output. If two solutions produce a somewhat equally excellent output but one gets there faster I would prefer it if all else was equal.
For both cameras I think that the wide angles are the crunch factor; it is hard getting wide angles/sensors to resolve minutel landscape detail as comfortably or nicely as full frame digital / medium format does. I get the feeling details seem a little compressed/crude.
For close up work with wide angles I seldom get the same feeling, usually they work great.
Yeah, don't get me wrong. I think the GXR sensor certainly does a better job, and I'd swap for that sensor in a second. It's just difficult to figure out real world differences if the sharpening isn't optimized for each, and I'm just talking about simple LR3 sharpening settings, not spending a bunch of time in post with sharpening programs or anything. Since I'm considering AA filter removal, I want to get a real sense of the differences. I guess I just need some raws so I can play with them myself.
No worries at all. I think GXR files require very little sharpening in post, and that's the way I like it. I didn't spend all that much time with the NEX5n but the files seem to take to sharpening well.
Smearing aside I think the final output is very similar.
What's interesting to me is the possible Nex-5N sensor in the Ricoh camera. If they then on top of that develop a viewfinder similar to the Sony one and perhaps even a new body for the lensors - well, it wouldn't hurt to switch then.
I think that with sensor in the NEX 7, the APS-C sensor/system may soon reach it maturity. The next step will a FF sensor in a small package. We will see what Leica will bring out next year, a new FF system or a FF compact X2.
Jonas B wrote:
What's interesting to me is the possible Nex-5N sensor in the Ricoh camera. If they then on top of that develop a viewfinder similar to the Sony one and perhaps even a new body for the lensors - well, it wouldn't hurt to switch then.
Yeah, that's why I'd like to see some raws for both cameras, so I can weigh the options. If Ricoh brings a tilt screen and new EVF, then it is a no brainer for me, but, as it is now, the possibility of removing the AA filter from the 5N is interesting to me...although MaxMax hasn't returned my email.
ocean2059 wrote:
I think that with sensor in the NEX 7, the APS-C sensor/system may soon reach it maturity. The next step will a FF sensor in a small package. We will see what Leica will bring out next year, a new FF system or a FF compact X2.
Hmm 35/2 ff Leica x2 $3k....I'd probably buy it! Only if it was a lot faster to tolerate than the x1 and max f2
Great write up, interesting to read. I'll be watching with interest for a 16mpix mensor