If Im correct, that color shift problem is caused by thick IR/UV/glass cover.. not AA filter. If you remove whole assembly except CFA, and replace with reasonably thick IR/UV.. then in theory it might be gone. But I wonder what will it do with AF..
Tho as its focus based on what sensor see.. maybe nothing.
Mescalamba wrote:
If Im correct, that color shift problem is caused by thick IR/UV/glass cover.. not AA filter. If you remove whole assembly except CFA, and replace with reasonably thick IR/UV.. then in theory it might be gone. But I wonder what will it do with AF..
Tho as its focus based on what sensor see.. maybe nothing.
yup shouldn't effect contrast detect AF. although it wouldn't matter to me either way. i'd want to see some proof that it gets rid of color shift and corner smearing before i ask to get it done to my camera. i don't have any need for extra resolution from my NEX-7.
Sorry i meant the same thing. I don't think you see smearing without shift but you will see shift without smearing. It appears smearing is a bad case of shift where data is lost at that point, not just altered.
FlyPenFly wrote:
Sorry i meant the same thing. I don't think you see smearing without shift but you will see shift without smearing. It appears smearing is a bad case of shift where data is lost at that point, not just altered.
the Fuji shows an awful lot of smearing without shift. it seems to me the smearing is due to astigmatism induced by cover glass while the shift is due to how oblique rays interact with the cfa.
sebboh wrote:
on my NEX-7 pretty much all my lenses produce moire if I shoot a cityscape.
I have done quite a bit of city shooting with the 7 and very sharp lenses (G 45, Summilux 50, Elmar 24), and not noticed moire. Though I have seen it on close-ups of textiles, so I know it can happen. Just not on my cityscape shots, it seems. I'll be very interested to see the results from the procedure. Apparently in a few months the successor to the 7 will appear, which will make my 7 pretty near worthless. So having it AA-less as a backup body makes sense IMHO.
philber wrote:
I have done quite a bit of city shooting with the 7 and very sharp lenses (G 45, Summilux 50, Elmar 24), and not noticed moire. Though I have seen it on close-ups of textiles, so I know it can happen. Just not on my cityscape shots, it seems. I'll be very interested to see the results from the procedure. Apparently in a few months the successor to the 7 will appear, which will make my 7 pretty near worthless. So having it AA-less as a backup body makes sense IMHO.
the high res of the 7 means you need to find pretty high frequency patterns to get it but i posted a really obvious example previously in the thread: https://www.fredmiranda.com/forum/topic/1162576/0#11088909
look at the corrugated aluminum on the warehouse in the lower half. i got noticeable moire on that scene from f/2 to f/8 after which diffraction acted as an AA filter.
Corner hue shift is an effect that has its' main causes below the CFA. There's not much on top of it that can make angle dependent hue shift better or worse except a seriously miscalculated microlens array (well - the used lens of course...), and that's rare now.
The cell height from the middle of the CFA layer thickness down to the active parts of the pixel embedded in the silicon underneath should preferably not be taller than the pixel is wide. That ratio is usually around 1:1 in large pixels and about 1:2 in small pixels, but since the sensor surface has a really, really high index of refraction this isn't a problem as long as the passivation layer and the bonding layer between the surface proper and the CFA bottom isn't to thick.
Lightpipes and other similar concepts all minimize cross-pixel bleed and sensor-level contaminations, but they unfortunately don't do much about flat-stack optics effects above the surface.
.............
Smearing on the other hand can be one of many things. When the toppings approach 1.5mm or more as they do in many cameras, the spherical aberrations are magnified (all over the surface) and astigmatism increases (or decreases if the value was negative to start with...).
This is one of the things that should make a lens "made for digital", the construction is pre-planned to compensate for this.
What one would ideally have is a filter that spreads oblique rays less than it spreads perpendicular rays. That would add in AA for medium apertures and telecentric lenses, and less AA for large apertures and WA symmetrical lenses.
Unfortunately I know of no method to do that, and even less how to do that in mass production...
Removing the AA filter completely adds in more pixel level contrast in the image before demosaic, and that gives aliasing to a sparsely populated matrix. In laymans terms - you might think you see more detail, but almost all of it is false detail - stuff that wasn't there in reality in front of the lens. The only place where AA-less cameras give more real detail than a camera with a good AA filter is on flat surfaces with a high green image content. In those you get more surface texture, and the texture is rendered with good reliability since the interpolation algorithm can suppose that the entire surface is uni-colored.
There are only two ways to get more REAL detail in an image. One is more base resolution (more MP), the other is stacked cells. Unfortunately the only stacked cell that is practically viable in mass production right now is Foveon, and that system has it's own pitfalls - that makes it totally unsuitable for anything requiring accurate color or high ISO performance. Which unfortunately is quite a lot of normal photography.
The X-Pro 1 has a normal density IRUV. I doubt that causes any problems in a converter good for the CFA pattern they use - unfortunately that means Fuji's own, and nothing else AFAIK right now.
