It might be a great camera with native lenses (I still want to see more city scapes and distant scenics to see how it performs in those situations) but doesn't look like a compelling camera for my purposes.
kosmoskatten wrote:
Corner smear seems evident, what are the odds that ALL his lenses are not resolving the corners well, on a crop sensor? None, I'd say.
I think there is a fairly good chance that user error/ inexperience is to blame. If that is the case, then it would make sense that ALL lenses are showing soft focus. This user did not even know how to magnify the EVF view early on to focus the adapted lenses. Who knows what other errors he might be making. The fact that there is no color vignetting is encouraging. I would think that detail smearing would be accompanied with major vignetting and or color issues in the corners and I don't see that in these shots. I think we need more examples from more users.
Tariq Gibran wrote:
I think there is a fairly good chance that user error/ inexperience is to blame. If that is the case, then it would make sense that ALL lenses are showing soft focus. This user did not even know how to magnify the EVF view early on to focus the adapted lenses. Who knows what other errors he might be making. The fact that there is no color vignetting is encouraging. I would think that detail smearing would be accompanied with major vignetting and or color issues in the corners and I don't see that in these shots. I think we need more examples from more users. ...Show more →
i suspect that the randomization of the bayer pattern has the added benefit of drastically reducing color issues in the corners (the regular pattern is why we see the colors in the corners that we do) so we may see corner smearing without color shift on this sensor.
Tariq Gibran wrote:
I think there is a fairly good chance that user error/ inexperience is to blame. If that is the case, then it would make sense that ALL lenses are showing soft focus. This user did not even know how to magnify the EVF view early on to focus the adapted lenses. Who knows what other errors he might be making. The fact that there is no color vignetting is encouraging. I would think that detail smearing would be accompanied with major vignetting and or color issues in the corners and I don't see that in these shots. I think we need more examples from more users. ...Show more →
I am sorry Tariq, but I think that is just wishful thinking.
Both the CV15 and the CV12 are near impossible to miss focus with at those kind of shots.
I have the CV15, so I know that racking focus between just short of 3m and infinity makes very little difference IRL on both the GXR and NEX5N and the lens simply does not perform like that.
Racking the lens out to infinity or 3m on the GXR gives much better results even at f4.5. Especially the stopped down shots show no substantial improvement on the X Pro 1 shots. One does not even have to resort to magnified view focusing with that lens for that type of shot.
One recent afternoon I brought the GXR and 15mm in lieu of my AF A12/18 (28) to work for job site documentation and being very tired I forgot to adjust focus on the first three/four shots which were at wide open and infinity, subject was around 4m away, all looked good in spite of that. I adjusted focus to around 2.5m for some shots and actually forgot to refocus for some more distant shots and shot at wide open (colleague was at the door wanting to call it a day) and even those shots turned out just fine. So, I bracketed very amateurishly by placing focus all over the place and all shots were in the range of very good to excellent with better corners than what I have seen so far with the same lens in plain day and stopped down.(EDIT: on the X Pro 1.)
A 15mm lens is that easy to make do with. With a 12mm f5.6? Haven't tried but it can't be worse.
If he has focused by simply racking out to infinity, the shots should look better. If he has racked the lens out to 2m the shots would look better. I am sure he did not focus at 1m, at which point some of the shots would still look better / more uniform stopped down that much.
Again, not nailing any of those shots, and not even the stopped down ones, points to the sensor not being adequate for that lens and sensor combination. (If he could not focus the 35/1.2 at such an object, he needs an AF camera, but he would be more forgiven for that.) The more samples I see the easier it is to see the real deal with the X Pro 1, which is; excellent camera within its own lens range, possibly excellent with normal and telephoto M lenses.
EDIT to keep page from turning:
Sorry for the lengthy rant Tariq, I am just off to bed and brain is shutting down making keep short sentence very hard for brain.
If you want to persist on thinking the X Pro 1 handles wide RF glass well, I will not persist in telling you otherwise. Based on Fuji documentation and real life shooting samples with known lenses I am quite sure it doesn't.
WHAM (smack on the back of my head from GF is signal to brain; shut off, go to bed.)
michaelwatkins wrote:
Of all his images, the CV15 shots of the apartment buildings with the green building in front look the most reminiscent of how a bad lens/sensor matchup (like the NEX-7 with some lenses) will destroy details in the edges. Centre third sharp (not that sharp in these samples), outer thirds progressively clubbed to death by FilterPack The Destroyer.
It might be a great camera with native lenses (I still want to see more city scapes and distant scenics to see how it performs in those situations) but doesn't look like a compelling camera for my purposes.
To be honest, I'm relieved. If the XP1 delivered stunning edge to edge image quality with RF lenses I own or might own, I might have been tempted to look at it. In person. In secret.
The standing joke around here goes something like this:
Wife: I've not seen your small camera in the open for a few hours | days. Is it still the same one or have you exchanged it for another?
Me: Occasional red face and explanation that starts of with "But..."
Probably not all the image quality issues in that series are due to operational error. Edge detail on the wider wides does seem to be an issue with this camera and at least some RF lenses. Looking at this image again:
The centre seems in focus enough. Which building was his target? It almost doesn't matter because the smearing shows up in the foreground and background plane. Probably the lens was focused at the infinity stop - granted that could be a problem depending on the adapter. At f/5.6 the edge detail smearing is fairly evident, in both the trees and building in the background on the left, and in the foreground building on the right. On the right hand side of the image
At f/8 the effect is there but substantially diminished. Probably for many users and purposes they'd be willing to live with it, especially since colour cast and vignetting do not seem to be issues with this lens/camera combo. I bet an in-camera jpeg of this scene would have looked better too.
I can't remember how the CV15 works out on the NEX-7 but would not be surprised if the XP1 is working with this lens a little better, user operational issues aside, with respect to edge detail destruction.
Edit: Just saw Henrik's note. He's cut to the point and I agree. I was going to comment on how hard it is to badly miss focus with these lenses but I only go as wide as the ZM18. Even that one... difficult to screw up. It might crop like 28mm but it still focuses like an 18. It can be difficult to nail exactly where critical focus is, there is so much area in acceptable focus.
By f/8 or f/11 a 12 or 15mm is an infinity camera machine.
Anyway.. the image above shows acceptable focus on both near and far buildings. Probably Silkypix (or the Silkypix user) is muddying the waters - no sharpening or some such thing. The edges are mutilated, but I think perhaps not as bad as we've seen with some combos.
kosmoskatten wrote:
I am sorry Tariq, but I think that is just wishful thinking.
Both the CV15 and the CV12 are near impossible to miss focus with at those kind of shots.
I have the CV15, so I know that racking focus between just short of 3m and infinity makes very little difference IRL on both the GXR and NEX5N and the lens simply does not perform like that.
Racking the lens out to infinity or 3m on the GXR gives much better results even at f4.5. Especially the stopped down shots show no substantial improvement on the X Pro 1 shots. One does not even have to resort to magnified view focusing with that lens for that type of shot.
One recent afternoon I brought the GXR and 15mm in lieu of my AF A12/18 (28) to work for job site documentation and being very tired I forgot to adjust focus on the first three/four shots which were at wide open and infinity, subject was around 4m away, all looked good in spite of that. I adjusted focus to around 2.5m for some shots and actually forgot to refocus for some more distant shots and shot at wide open (colleague was at the door wanting to call it a day) and even those shots turned out just fine. So, I bracketed very amateurishly by placing focus all over the place and all shots were in the range of very good to excellent with better corners than what I have seen so far with the same lens in plain day and stopped down.(EDIT: on the X Pro 1.)
A 15mm lens is that easy to make do with. With a 12mm f5.6? Haven't tried but it can't be worse.
If he has focused by simply racking out to infinity, the shots should look better. If he has racked the lens out to 2m the shots would look better. I am sure he did not focus at 1m, at which point some of the shots would still look better / more uniform stopped down that much.
Again, not nailing any of those shots, and not even the stopped down ones, points to the sensor not being adequate for that lens and sensor combination. (If he could not focus the 35/1.2 at such an object, he needs an AF camera, but he would be more forgiven for that.) The more samples I see the easier it is to see the real deal with the X Pro 1, which is; excellent camera within its own lens range, possibly excellent with normal and telephoto M lenses.
EDIT to keep page from turning:
Sorry for the lengthy rant Tariq, I am just off to bed and brain is shutting down making keep short sentence very hard for brain.
If you want to persist on thinking the X Pro 1 handles wide RF glass well, I will not persist in telling you otherwise. Based on Fuji documentation and real life shooting samples with known lenses I am quite sure it doesn't.
WHAM (smack on the back of my head from GF is signal to brain; shut off, go to bed.)
If the adapter allows the lenses to focus past infinity and the user simply set the lens to the infinity stop, all bets are off! That could easily explain the situation. I'm not saying that the X-Pro1 is going to perform well with wide rangefinder lenses, only that we need more samples by more experienced users before any conclusion can be reached.
sebboh wrote:
i suspect that the randomization of the bayer pattern has the added benefit of drastically reducing color issues in the corners (the regular pattern is why we see the colors in the corners that we do) so we may see corner smearing without color shift on this sensor.
The lack of the AA filter could explain it (which is also why we might not expect smearing) but I highly doubt the non-bayer filter array has anything to do with it.
Tariq Gibran wrote:
The lack of the AA filter could explain it (which is also why we might not expect smearing) but I highly doubt the non-bayer filter array has anything to do with it.
um, why? the color shift is caused by light that passes through a given color filter not hitting the correct photosite isn't it? i've forgotten.
sebboh wrote:
um, why? the color shift is caused by light that passes through a given color filter not hitting the correct photosite isn't it? i've forgotten.
I believe the problem is due to the oblique angle of light at the periphery of the image being further distorted by the filter plate (which normally contains the IR and AA filter). The area in question is so large that the difference between a 6x6 (xtrans) and a 2x2 (bayer) pixel array pattern is not going to matter - I suspect.
Tariq Gibran wrote:
I believe the problem is due to the oblique angle of light at the periphery of the image being further distorted by the filter plate (which normally contains the IR and AA filter). The area in question is so large that the difference between a 6x6 (xtrans) and a 2x2 (bayer) pixel array is not going to matter.
that is what causes the astigmatism yes. is the fuji color filter a regular pattern - all 6x6 grids are the same? i thought it was more randomized, e.g. same number of each color in a 6x6 grid but the location of each color varies. it is the regular pattern that causes the color shift. if the filter locations were randomized i would think you would get more color noise in the corners but not uniform colors for each corner as the NEX cameras do.
sebboh wrote:
that is what causes the astigmatism yes. is the fuji color filter a regular pattern - all 6x6 grids are the same? i thought it was more randomized, e.g. same number of each color in a 6x6 grid but the location of each color varies. it is the regular pattern that causes the color shift. if the filter locations were randomized i would think you would get more color noise in the corners but not uniform colors for each corner as the NEX cameras do.
Two 3x3 patterns inside the regular 6x6 pattern IIRC. The 6x6 patter is repeated in a regular fashion.
Tariq Gibran wrote:
I think there is a fairly good chance that user error/ inexperience is to blame. If that is the case, then it would make sense that ALL lenses are showing soft focus. This user did not even know how to magnify the EVF view early on to focus the adapted lenses. Who knows what other errors he might be making. The fact that there is no color vignetting is encouraging. I would think that detail smearing would be accompanied with major vignetting and or color issues in the corners and I don't see that in these shots. I think we need more examples from more users. ...Show more →
I agree we need to see more samples. Drawing conclusions from one user is insufficient, especially when he admits his copy of the CV15 might be a poor one, based on his results with the M9 (assuming he means it's soft in the corners, in addition to the strong colour shift that lens produces on the M9).
I'm not sure though that detail smearing and color shift is coupled. Detail smearing seems to be very much influenced by the thickness of the toppings over the sensor. Color shift seems to be more a factor of the sensor itself, perhaps the micro lenses, based on what I've been reading in some Phase One documentation on their support site (I figured I'd check there since they've had many years of experience with this problem).
sebboh wrote:
The lack of the AA filter could explain it (which is also why we might not expect smearing) but I highly doubt the non-bayer filter array has anything to do with it.
The M9 doesn't have an AA filter and still experiences edge color shift. Same with a lot of medium format backs, hence the reason Phase One includes the LCC feature in the RAW converter.
(^^^ Not to mention the GXR which despite having no AA filter still exhibits colour cast with some lenses, albeit often not objectionable at all.)
Ron, definitely we should look at others but kinson12's photos and the conclusions they invite do remind me of Rob Skeoch's series on the NEX-7. There was enough information in Rob's images to draw unhappy conclusions that would later be confirmed but only after much hope and angst (me included) has passed under the bridge.
EDIT
Then again the CV35 shots of the same towers looked at below are much better by f/8, to the point where even if smearing is an issue on this camera, it would appear that stopping down has more effect than on the NEX-7 for comparison.
I think I will stop looking at XP1 lens shots as a solution.
More samples please! Someone shoot some Zeiss glass on this puppy too. Purdy please?
Remainder left as is:
Tariq, sure, an adapter that allows the CV15 to focus past infinity would degrade the image big time, but it would tend to degrade somewhat more evenly across the field. Looking at the apartment tower series it looks like if anything the lens was focused on the near building or somewhere in between near and far. Or at a possibly accurate infinity stop.
Base image, a little contrast and sharpening applied. Didn't help a lot but made me feel better.
I took a bunch of 100% crops from the full sized image - top and lower centre, top and lower left and right - as much for myself as anything. The guard rails at the top of the towers and other fine details are recognizable and somewhat sharp.
By the time you get to the left edge of the towers, these fine details are significantly degraded but here at f/8 the result is much better than at f/5.6. (Warning, distortion alert! Grab hold of something if feeling dizzy...)
Right edge, top, same:
Centre, bottom - not as crisp as I'd hope for but looks perhaps a little sharper than centre, top:
Bottom left, hard to make out much, sorry:
Bottom right:
This wall in front of the lower green building rules out an accidental close in focus problem:
Even if the CV15 has some curvature of field, the edge details shouldn't be so mushy at f/8 (or even f/5.6!) if this sensor (and lens, yes the lens may be flawed) is a good mate for our more problematic wides.
michaelwatkins wrote:
Tariq, sure, an adapter that allows the CV15 to focus past infinity would degrade the image big time, but it would tend to degrade somewhat more evenly across the field. Looking at the apartment tower series it looks like if anything the lens was focused on the near building or somewhere in between near and far. Or at a possibly accurate infinity stop.
The CV15 looks so soft, even in the center of the frame, that something appears to be awry. Even on the original NEX-5, the CV15 I tested on it was tack sharp in the center. Anyway, his results will either be confirmed or rejected after we see more samples from more users.
I don't actually have an X-Pro 1 on pre-order myself - so I don't have a dog in this fight, though I did order the adapter to have on hand just in case . I figure I will wait and see. I crave resolution so my pre-orders are tied up with Nikon D800's. I was hoping the X-Pro 1 might serve as a one camera solution but it clearly seems to involve too many compromises for my uses, which makes the high price even tougher for me to justify.
I appreciate the study of CV15 Heliar performance on the X-Pro1, here.
In my own study of that lens on the demanding NEX7, it is capable of decent corner performance, if precautions are taken and cornerfix is used. I'm not the only one suggesting to focus and expose for the corners at around f/8.
In these tests, the lens was pointed too high and should have aimed down bringing the buildings into the corners more, for at least top corners evaluation. As is, I'm not so sure the Heliar was focused optimally, here.
