What freaks me out a bit is how bad the 50mm smearing is. The Canon LTM is atrocious, and the Summilux is just bad -- looks like a $20 Sears lens on that camera. The Canon 35mm LTM is also terrible, and it's a very sharp lens.
I'm very concerned but still hopeful the A7R addresses smearing, even at the expense of color shift...
michaelwatkins wrote:
I just can't see Sony going through heroic measures to tune the A7r for such lenses when they have to support their own lens designs first and foremost.
Whatever the lens construction of the FE line up, it apparently does not have the high incident light angle problem many wide angle RF lenses have, so Sony doesn't have to optimise the sensor for this.
If the A7r does treat these problematic lenses better then it's more by happy accident or coincidence (merely the removal of the AA filter?), because FE mount lens performance must always come first.
Whatever the case it is impossible to guess - the A7r sensor is completely different to the A7....Show more →
I don't think the removal of the AA filter alone is likely to be responsible for any color shift improvement caused by high incident angle. I think that the offset sensor ... IS ... designed to work in concert with the Sony glass exit pupil location, and any potential benefit to rangefinder glass is likely going to only be "happy accident" for those lenses with similar angles that can benefit from the amount of offset.
Do we know anything about the new Sony lenses regarding their exit pupil position, size and FL for their widest forthcoming lenses. If so, that could give a clue to the limits of the trig/vector forces involved that they've determined at optimal or limit. We certainly are going to be in a "wait & see" mode, but I think we can look to what Sony engineers are doing on their new glass to garner a clue. Sony's marketing department has likely harnessed this a bit more aggressively by stating "some" gain for rangefinder, but I've got little hope for WA/UWA rangefinder @ too close, too steep for digital FF on micro-lens, Bayer array. I think we are largely fooling ourselves if we think otherwise.
Noting Shelt's comment below that the 90mm isn't too bad ... it's a far cry from 90mm (or even 50) to UWA and I doubt the "offset sensor" can make up that much ground if the A7 can't handle normal rangefinder glass very well.
I'm not strongly familiar with iconic color rangefinder images ... my memory is largely @ B&W rather than color. Just wondering about historic UWA rangefinder images on color film. Anybody with their favs to share (new thread maybe).
Thanks for the the test Ron. The results from the A7 are pretty atrocious. Not just wide angles, but it might not work well with any rangefinder lens. The shots wide the 50 lux are almost as bad as those with the 28 cron. Let's hope the A7r is better, but slightly less optimistic after seeing these.
Michael Reichmann wrote this about the A7r. He's generally pretty picky about quality, so I hope there's a big difference between the a7 and a7R, because I don't see any rangefinders in Ron's test that seem to do well (except, perhaps the 90mm).
"For some photographers (myself included) putting Leica M mount lenses on the Sony A7R may be reason enough to break out a bottle of Veuve Cliquot champagne and celebrate. Using an M to E adapter (I have a Metabones), virtually every Leica M lenses that I own works well on the A7R. Some of the ultra-wide and very wides do vignette, so be aware of this. There is no software correction for this, because we're mixing and matching. With other systems where the lens and camera are from the same company there's a lot of magic that can be done in firmware."
'If the A7r does treat these problematic lenses better then it's more by happy accident or coincidence (merely the removal of the AA filter?), because FE mount lens performance must always come first.'
+1, and +1 to MR's comments.
There is a reason Sony and Zeiss are rushing hard to cover their native lens needs in the next year, by which time there will be around 20 FE plus whaever Zeiss do with manual focus. I have little doubt they will both do a fine job of this, ad it's more important than the M users desires by a country mile.
Even more than before lens-sensor-processing is key to the end result.
"Sidebar: My Leica M lens collection, none of which have problems on the NEX-7, includes the 24mm f/1.4, 35mm f/1.4, 50mm f/1.4, 90mm f/2, 135mm f/3.4, 28-35-50mm Tri-Elmar and 16-18-21mm Tri-Elmar.
I'll add that I have a couple of M to E lens adaptors but my preferred brand is Novaflex. First rate fit and finish."
philip_pj wrote:
'If the A7r does treat these problematic lenses better then it's more by happy accident or coincidence (merely the removal of the AA filter?), because FE mount lens performance must always come first.'
+1, and +1 to MR's comments.
There is a reason Sony and Zeiss are rushing hard to cover their native lens needs in the next year, by which time there will be around 20 FE plus whaever Zeiss do with manual focus. I have little doubt they will both do a fine job of this, ad it's more important than the M users desires by a country mile.
Even more than before lens-sensor-processing is key to the end result. ...Show more →
This is fine with me. All that matters to me is the availability of reasonably priced high quality lenses. In general, I just assume modern lenses designed for a specific camera will likely be better than adapted lenses. Very good wide angle lenses are expensive and coveted for a reason. They're just so difficult to design and build. For these cameras, I'm ok with waiting for a Zeiss wide angle zoom and maybe a few Leica R lenses for telephoto. I'll also will no longer be hindered by Nikon's lack of quality TSE lenses. I can now adapt Canon's TSE lenses.
Yeah, it'll suck for those who own a lot of RF lenses, but that group is very small. There is a bigger pool of people with Canon and Nikon lenses
shelt wrote:
What freaks me out a bit is how bad the 50mm smearing is. The Canon LTM is atrocious, and the Summilux is just bad -- looks like a $20 Sears lens on that camera. The Canon 35mm LTM is also terrible, and it's a very sharp lens.
Haha. Both of those are some of my favorite lenses, and I will even admit they are not really as good of lenses as you probably think they are... They are 50 year old canon lenses. they are great on film and passable on M9.
Awesome for the price, when you can get then for a reasonable price.
I'm more than happy with how they look on the Sony.
philip_pj wrote:
Even more than before lens-sensor-processing is key to the end result.
+1 @ system designed glass will likely be stellar to play very nice with the offset sensor ... the rangefinder UWA, I don't think that's really why Sony offset the sensor, despite any hype/hope that marketing might try to piggyback onto engineering for more sales to a larger audience.
RustyBug wrote:
After seeing the magenta on the sides ...
The optical projection of the lens will always be the same, whether you put it on an APS-C or FF ... all the APS-C has done is crop away the bad stuff. Thus, the whole APS-C equivalency comparison @ fair is a moot point at whether or not the A7R with it's offset sensor will address the color shift that occurs from the additional angles of refraction imparted by the microlens redirecting the light onto an RGB bayer array.
The theoretical hope that an offset sensor would result in a lesser amount or refraction required from the microlens to direct the light onto the sensor appears to have little impact (from the magenta sides image and from the comments regarding by others).
The optical projection is designed to have the light converge to align wavelengths at a given point. Having those projections converge (from differing angles) to a point that is then subsequently being refracted is always going to have a detrimental effect on the aligment/divergence. I think that the preliminary results of that magenta shot coupled with the comments by others early experience with a given lens likely tells the story that the offset sensor is no "magic beans" to address the microlens refraction.
While we can clamor over this lens vs. that lens and discount the negative results from a given lens in hope that it will work better for a different lens to our liking, it may be a bit folly.
Light is an energy, a force that is being redirected onto our sensor/film. As such it is a vector quantity. The vignetting that we see from those steep angles is a product of the trigonometry from the projected image. As long as the microlens exists in the same film plane, the angles from the projected image to the microlens will remain the same for a given distance from the optical axis.
On film, will don't have the bayer array to contend with. While the "offset sensor" sent hope that it might have some ability to contend with the color shift imparted by the microlens refracting the converged light, these early images and comments at the A7R make me think that as long as we are using a bayer array in the same film plane, we'll get the same results. The subsequent diffraction of an otherwise properly converged light wavelengths only serves to separate those wavelengths from proper alignment. As that angle of refraction increases (corresponding to the steeper angles of incidence), the degree of separation likewise increases.
As such, I'm inclined to think that unless a Foveon approach (or Canon's "tiered Foveon") is utilized where there is no subsequent refraction to re-separate the properly converged light ... the color shifts will always occur for the steep angles in conjunction with a micro-lens Bayer array. Unless the film plane itself is modified, so that the amount of refraction imparted by the microlens remains below the (visible) threshold for producing that separation, it is physically destined to remain a byproduct of the micro-lens refraction imparted onto the properly converged light.
Let the bashing begin ... but I'm no longer holding my breath that the A7R's offset sensor is going be the "magic beans" that I was very much hoping. Sigma needs to bring out a FF mirrorless Foveon, or Canon needs to finish their "tiered Foveon".
Imo Bayer array and micro-lens refraction is not the answer for UWA steep angle color shifts. The only way that it could be is if a lens was designed to strike the microlens with a non-convergent light, that would subsequently be refracted by the micro-lens to proper convergence. This of course would be most suitable to a fixed mount lens ... like the RX1. As such, that would still mitigate the use of rangefinder glass. UWA, rangefinder glass on Bayer micro-lens for color isn't something I'd hold my breath any longer. Even with SLR glass, it isn't an ideal approach. Thus ... I better get more adept at software correction/profiling for my UWA's until we have a FF non-Bayer/microlens solution. That, or look UWA glass designed for APS-C for Sigma's Foveon (which is rather limited in choices).
RustyBug wrote:
I don't think the removal of the AA filter alone is likely to be responsible for any color shift improvement caused by high incident angle.
Maybe not responsible alone, but given Zeiss wrote a white paper about astigmatism a side effect of AA filters and high light angle incidence from some lens designs (their own), and given that the M9 and Ricoh GXR/M both lacked AA filters, I guess I'm expecting the A7r to be better as a result of that one difference. Better enough?
dunno.
I saw that quote from MR the other day and said to myself... I'll believe it once someone publishes a thorough round of tests. MR's own collection might be ... probably is in fact ... the less problematic M lenses given he was shooting them on NEX-7 and had if I remember correctly altered his lens line up accordingly.
The A7's result is terrible. I was hoping it can beat the M240 easily, but it's not. If A7r shared the sensor with D800e, then it does have LPF (in a different way) which may not solve the problem. I am kind of disappointed since I got my 21SEM last year, still waiting for a workable EVIL camera for it.
I wonder how difficult it would be to replace the AA and IR cut filter, with something thinner and having a constant refractive index? Could be an enterprising camera modification to eliminate the smearing.
I'm not sure if it's just the AA filter and microlenses causing troubles, there's also the coverglass, which I don't know if a sensor can be without. I mean, you have to be able to clean your sensor, right?
Sony on Sony looks likely to be smokin' hot on the A7R ... but I just don't think it is going to be the FF holy grail that rangefinder folks have been longing for with their M glass. Mountable yes ... WA/UWA IQ, we'll see.
We'll see how wide Sony can take their own designs (and whether or not they design mustache distortion into it like Zeiss does). I'm guessing that Sony goes with a mild amount of mustache to help keep the angle of incidence in the edges/corners moderated and work within the limits of their offset sensor. But will they dare go deep into UWA territory with their own ... guessing not soon.
Spyro P. wrote:
I'm not sure if it's just the AA filter and microlenses causing troubles, there's also the coverglass, which I don't know if a sensor can be without. I mean, you have to be able to clean your sensor, right?
On dpreview there was someone who made this mod for their NEX7 - replacing the cover glass, AA filter and IR cut with a thinner stack that reduced refraction/smearing with M lenses. Still reported color issues though.
I think the M wide angles are important to mainly a niche group of shooters. For landscape, I'm more interested in the Canon 17mm TSE for this camera. When I shot with Hassy before, I was looking at the HTS adapter, which costs $7k. For my Nikon, I didn't really have any choices for good wide TSE. Now with this camera, I can pick and chose which lens I want to use from both Canon and Nikon and Leica R. 14-24 from Nikon, 17mm TSE from Canon, 85L from Canon, etc.
Before the camera was announced, I thought I'd keep my Nikon D800E, but now that I think of all the possibilities, keeping the Nikon body makes less sense.
