p.16 #1 · First Sighting: A7rii Thin filter in action
A7 NKIR improvement over A7R 2 + PCX 2.5m
Lens: Zeiss ZM 25mm 2.8
Post Processing: Auto Tone, Sharpening (no WB adjustment)
Improvements are clearly seen all around at wider f-stop. NKIR UT images are even sharper at extreme edge areas although they have purple color cast due to deeper pixel wells of non-BSI Sony A7.
Steve Spencer Online • • • • • • • Upload & Sell: On
p.16 #2 · First Sighting: A7rii Thin filter in action
teeraash wrote:
@artur5@ are you sure that pcx2000 is better than pcx2500 on ZM 25mm? I didn't follow that threat lately. But I recall that the consensus is that pcx2500 is the best. Color pof NKIR (0.75) is much better than what I got with Karali v2. Probably you can adjust in-camera WB tint b1g2 to get nice jpeg color straight out of the camera or simply use LR color picker. It takes lot more time to adjust the Kolari color which deviates a lot from stock Sony.
Oh I browsed that thread again and saw the simulatiin on p67 which indicates that pcx2000 is better than 2500. Probably I go by the testing results of Guy and Fred? which show remarkable corners sharpness of pcx2500 on this lens.
I have Optosigma pcx 1500, 2500, 5000 and Eksma pcx 10000. All of them performs worse than NKIR UT on all lenses that I've tried so far (ZM 35/1.4, 25/2.8, 50/1.5 Leica 50/1.4 asph, 50/2 IV).
Up until now I've tested only near distance but my preliminary finding is that m-mount lenses on NKIR UT are even better than on M240. Only extreme edges and corners are slightly worse due to vignette! I also see that Loxia 50mm performs better on A7 NKIR UT than stock A7R 2 even in the peripherals in terms of sharpness and contrast. I have to redo the test as I myself find that these are hard to beleive!
I'm not associated with NKIR in anyway and would like to point some apparent negative points. NKIR takes out the sensor cover (noone has done this) and he said around 5% may have scratched as the process is complicate and not risk-free. He will mitigate this damage for you (I don't know how, cmos surface repair?). Secondly, moire can be severe with some lenses particularly on the A7 as the mod makes the mirror stack ultra thin with no AA. Thirdly, we may run into infrared leak for certain lenses as it is only 0.75mm thick. He sent me the full spectral uv-ir cut analysis. The 0.75mm is slightly worse than the mediocre M9 above 700nm. If I do it again, I'll choose 0.85mm over the 0.75mm.
Lot of hurdle and headache. Is it worth to do it? I don't know but I don't like what they say "m-lenses are best used on m-cameras" ...Show more →
That Loxia 50 test is very interesting, but if you redo it make sure you downsample the A7rII files to match the resolution of the A7 NKIR files. Higher resolution (i.e., higher magnification) can very much affect appearance of sharpness so downsampling is probably the best way to try to equalize that. If you can do that test it would be very interesting, and if you have a ZM 50 f/2 to throw into the mix that would be interesting as well. I see the Loxia 50 and 35 as a lot like the front end filter solution for the ZM lenses built into the lens. I wouldn't be surprised if the ZM 50 f/2 on the A7 NKIR beat the Loxia 50 even on the A7r II, but I would be surprised if the Loxia 50 on the A7 NKIR was better than the ZM 50 f/2 on that body or even the Loxia 50 on the A7r II once the files have been down sampled.
p.16 #3 · First Sighting: A7rii Thin filter in action
Steve, what downsample algorithm would be appropriate for this purpose or I can use LR and export a7r2 images at reduced size?
May 18, 2017 at 08:23 AM
Steve Spencer Online • • • • • • • Upload & Sell: On
p.16 #4 · First Sighting: A7rii Thin filter in action
teeraash wrote:
Steve, what downsample algorithm would be appropriate for this purpose or I can use LR and export a7r2 images at reduced size?
Use can just use lightroom and reduce the image size using bicubic sharper. There may be better algorithms, but the LR one works quite well in my experience.
p.16 #5 · First Sighting: A7rii Thin filter in action
Steve, I respectfully disagree with the downsampling system.
if you want to compare sensors of different Mpix count, just upsample the 24mp file of the A7 to 42mp. Downsampling throws away the extra resolution of the A7rII, while upsampling the A7 to 42mp doesn't add any real resolution to the file, just more pixels.
With your method you could downsample a 50Mpx file from a Canon 5DSR to 4Mpix and conclude that isn't any better than a native 4Mp. file from a 'prehistoric' Canon 1D.
p.16 #6 · First Sighting: A7rii Thin filter in action
OK I'll do something like this. Say up the A7 to 42mp, up the A7R 2 to 80mp and the down to 42mp to match the upped A7. This would be fair as both will be resample not only one of them. Or I can up the A7 to say 36mp and down the A7R 2 to 36mp to match it?
p.16 #8 · First Sighting: A7rii Thin filter in action
I don't think you should up-res the A7 at all. Just down-res the A7r and compare them at 24MP. The A7r has the advantage since it starts with more MP, which especially will reduce false-color aliasing that these mods on the A7 suffer from to some extent, but that's a reasonable comparison. Trying to make more pixels serves no realistic purpose since you are not going to do that in real life.
Michael teeraash wrote:
OK I'll do something like this. Say up the A7 to 42mp, up the A7R 2 to 80mp and the down to 42mp to match the upped A7. This would be fair as both will be resample not only one of them. Or I can up the A7 to say 36mp and down the A7R 2 to 36mp to match it?
p.16 #9 · First Sighting: A7rii Thin filter in action
But on a second thought, what I want to do here is to compare front-end filter with NKIR UT mod. If I up the A7 to A7R 2 that will give advantage to the latter due to its higher& superia sensor. No?
Arthur, my A7R 2 is still a virgin but may be not for long.
p.16 #11 · First Sighting: A7rii Thin filter in action
As promised, a comparison of Loxia 50mm f2 on Stock Sony camera (A7R II) VS mod A7 NKIR UT. Ideally I should shoot images with stock A7 against A7 NKIR UT but since I don't have a stock A7 so I used stock A7R II instead. I resized A7R II images to match those of the A7 to proximate the image quality of the stock A7. All in all, the downsizing of the A7R II images should have quality equal or even higher than images straIght out of the A7.
Lens: Loxia 50mm f2
Cameras: A7R II VS A7 NKIR UT
Post Processing: Auto tone, WB with color picker, equal sharpening
Exporting: LR, A7 NKIR at actual 100% size, A7R2 reduced size to match A7
1. I can see improvements in terms of sharpness, contrast and acuity of the NKIR UT mod. Even in the center area where one can see the improvements up to f5.6. In mid-frame and corner areas the improvements are even more pronounced. Texture of pink/orange shirt can be clearly seen from the mod but not from the stock camera. Corner areas are improved and noticeable. This is probably due to the much thinner stack 0.75mm vs around 2mm of the stock (which has sensor cover, and another 3 layers of filters on top while the NKIR only has only one 0.75mm filter layer).
2. If images don't have IR leakage it is fairly easy to adjust color to match the stock. But in this extreme test of wardrobe with plenty of black and dark fabrics, it is impossible to match color without resorting to time consuming localize adjustment. Look like Leica M8 once again where one needs uv/ir cut filter in front of lenses. I will order one and see how much improvement it will bring.
3. I was afraid of pattern moire which is also very difficult to correct but it seems that the mod is only slightly worse than the stock. Probably this problem may make images unuseable in certain scenarios. I don't know.
4. NKIR sent me spectrum analysis of uv/ir cut of varios mirror stacks and Kolari v2 is superb. But as of now I think NKIR extreme mod is very very good in over all sharpness, corner sharpness, contrast and acuity.
p.16 #12 · First Sighting: A7rii Thin filter in action
Thanks for the pictures, teeraash. Very interesting and not what, in theory, we might expect beforehand.
First thing is that I’d like to ask Zeiss how it’s possible that a lens specifically designed for the Sony sensor stack performs better with a significantly thinner cover glass. We take for granted that Zeiss optic engineers are second to none, so maybe the naked truth is that there’s a limit to what one can do to counteract the damaging effects of a thick slab of glass between sensor and lens ?
I’d love to see this test with the Loxias 21 and 35. The Loxia 21 enjoys a great reputation; not so much the Loxia 35.
Best thing would be of course if we were able to make the comparison using two copies of the same Sony model. i.e. a stock camera vs a UT modded unit. Upsampling or downsampling adds another unwanted variable to the equation.
As a side note, I don't see overall IR contamination.on the MKIR-UT pictures. A Leica M8 would have turned garishly purple the synthetic blacks of the mid-frame pictures. Not this case here. Those UT images are contrastier and tend to orange-yellow if compared to the stock camera but I dare say that a custom WB and a camera profile would take care of the color balance, although more testing is needed. Apart from synthetic black fabrics, skin tones are specially revealing if there's IR leakage
Steve Spencer Online • • • • • • • Upload & Sell: On
p.16 #13 · First Sighting: A7rii Thin filter in action
teeraash wrote:
As promised, a comparison of Loxia 50mm f2 on Stock Sony camera (A7R II) VS mod A7 NKIR UT. Ideally I should shoot images with stock A7 against A7 NKIR UT but since I don't have a stock A7 so I used stock A7R II instead. I resized A7R II images to match those of the A7 to proximate the image quality of the stock A7. All in all, the downsizing of the A7R II images should have quality equal or even higher than images straIght out of the A7.
Lens: Loxia 50mm f2
Cameras: A7R II VS A7 NKIR UT
Post Processing: Auto tone, WB with color picker, equal sharpening
Exporting: LR, A7 NKIR at actual 100% size, A7R2 reduced size to match A7
1. I can see improvements in terms of sharpness, contrast and acuity of the NKIR UT mod. Even in the center area where one can see the improvements up to f5.6. In mid-frame and corner areas the improvements are even more pronounced. Texture of pink/orange shirt can be clearly seen from the mod but not from the stock camera. Corner areas are improved and noticeable. This is probably due to the much thinner stack 0.75mm vs around 2mm of the stock (which has sensor cover, and another 3 layers of filters on top while the NKIR only has only one 0.75mm filter layer).
2. If images don't have IR leakage it is fairly easy to adjust color to match the stock. But in this extreme test of wardrobe with plenty of black and dark fabrics, it is impossible to match color without resorting to time consuming localize adjustment. Look like Leica M8 once again where one needs uv/ir cut filter in front of lenses. I will order one and see how much improvement it will bring.
3. I was afraid of pattern moire which is also very difficult to correct but it seems that the mod is only slightly worse than the stock. Probably this problem may make images unuseable in certain scenarios. I don't know.
4. NKIR sent me spectrum analysis of uv/ir cut of varios mirror stacks and Kolari v2 is superb. But as of now I think NKIR extreme mod is very very good in over all sharpness, corner sharpness, contrast and acuity.
Thanks for this comparison. It is very interesting. What I see is not much difference in sharpness and I think that what you are seeing and interpreting as greater sharpness with the NKIR is false sharpness created be moire. For my eye, I would be much happier with the A7rII stock camera in these samples. The colour and dynamic range are just much better on the stock camera. Notice in the first set how the detail in the blacks are squashed. For my taste the moire is also a pretty big negative in these shots. I don't like the false detail it provides and it doesn't look real to me having an effect a bit like over sharpening. I hate to be a downer, but that is what I see.
p.16 #14 · First Sighting: A7rii Thin filter in action
I should clarify a little on my PP. I clicked auto tone in one or two pictures and then hit the sync button in LR development module to sync setting for all images. Thus, some images are over exposed a little and black and shadow levels of many images are not properly set. My bad but DR is not my concern here. Micro contrast, acuity and sharpness of the mod are higher for sure. Look at threats of clothes, texture. Yes, the AA less mod has more moire compares to the much higher MP A7R II. It is responsible for false color, repetitive weird patterns. This is the price we pay for having sharper image.
May 19, 2017 at 10:34 AM
Steve Spencer Online • • • • • • • Upload & Sell: On
p.16 #15 · First Sighting: A7rii Thin filter in action
teeraash wrote:
I should clarify a little on my PP. I clicked auto tone in one or two pictures and then hit the sync button in LR development module to sync setting for all images. Thus, some images are over exposed a little and black and shadow levels of many images are not properly set. My bad but DR is not my concern here. Micro contrast, acuity and sharpness of the mod are higher for sure. Look at threats of clothes, texture. Yes, the AA less mod has more moire compares to the much higher MP A7R II. It is responsible for false color, repetitive weird patterns. This is the price we pay for having sharper image....Show more →
In this case it is also the cause of false sharpness which occurs when the moire makes it look like patterns are there that are not and those patterns look sharp. It doesn't look natural and I don't think there is higher micro contrast here. The ultra thin mods are very nice, but they do have their drawbacks and you see that in this example, IMO. I am not a fan of moire, false colours, and false sharpness (over sharpness).
See this link for a description of what I mean by false sharpness:
p.16 #16 · First Sighting: A7rii Thin filter in action
Steve, may be my sharpening routine is too high for the mod and it might be more useful if I can upload 1-2 raw files of each set for you guys to look and play with.
p.16 #17 · First Sighting: A7rii Thin filter in action
Now let's compare A7 NKIR UT with Leica M240
Lens: Zeiss ZM 25mm 2.8
PP: adjust exposure, color picker at the same spot for WB, adjust tint of A7 slightly to match color, sharpening 31 with radius 0.8 (on the indoor wardrobe I used 51, 1 which may be a bit too much), no auto tone as done previously.
The images are at 100%. It was cloudy and pretty muddy weather and thus images apperar to be less contrast and less sharp than normal. It should be added that my 25mm infinity focus hard stop is accurate but for this test, I also cross-checked the focus with evf.
Sony A7 NKIR UT tops out Leica 240 in terms of sharpness in ALL AREAS even at f11. Contrast is also better on Sony. Vignette is about the same on both (Leica's lens detection was turnoff).
p.16 #18 · First Sighting: A7rii Thin filter in action
Hi artur5, I don't think that it's worth the risk to mod A7x for native lenses. But it's a diiferent story if one uses m-mount lenses or other adapted legacy lenses as the UT mod makes them shine. My lux 50mm asph, zeiss 50 1.5, two lenses that I love so much now have no issue at all on the A7 UT while the Kolari V2 leaves lot to be desired.
I'm thinking of sending my a7r2 as well as the mod is not affected my native lenses, Loxia 50, fe macro 90. Voig 15 4.5 also appears to work very well but I need to do more test to be certain.
I've posted my tests in this threat because I feel in debt to people here particularly uhoh7 who started the thread and sebboh. They open my eyes to the UT mod.
I will ask my friend to bring his a7 II for a comparison with my modded a7 as you suggested.
p.16 #20 · First Sighting: A7rii Thin filter in action
teeraash wrote:
@artur5@ are you sure that pcx2000 is better than pcx2500 on ZM 25mm? I didn't follow that threat lately. But I recall that the consensus is that pcx2500 is the best. Color pof NKIR (0.75) is much better than what I got with Karali v2. Probably you can adjust in-camera WB tint b1g2 to get nice jpeg color straight out of the camera or simply use LR color picker. It takes lot more time to adjust the Kolari color which deviates a lot from stock Sony.
Oh I browsed that thread again and saw the simulatiin on p67 which indicates that pcx2000 is better than 2500. Probably I go by the testing results of Guy and Fred? which show remarkable corners sharpness of pcx2500 on this lens.
I have Optosigma pcx 1500, 2500, 5000 and Eksma pcx 10000. All of them performs worse than NKIR UT on all lenses that I've tried so far (ZM 35/1.4, 25/2.8, 50/1.5 Leica 50/1.4 asph, 50/2 IV).
Up until now I've tested only near distance but my preliminary finding is that m-mount lenses on NKIR UT are even better than on M240. Only extreme edges and corners are slightly worse due to vignette! I also see that Loxia 50mm performs better on A7 NKIR UT than stock A7R 2 even in the peripherals in terms of sharpness and contrast. I have to redo the test as I myself find that these are hard to beleive!
I'm not associated with NKIR in anyway and would like to point some apparent negative points. NKIR takes out the sensor cover (noone has done this) and he said around 5% may have scratched as the process is complicate and not risk-free. He will mitigate this damage for you (I don't know how, cmos surface repair?). Secondly, moire can be severe with some lenses particularly on the A7 as the mod makes the mirror stack ultra thin with no AA. Thirdly, we may run into infrared leak for certain lenses as it is only 0.75mm thick. He sent me the full spectral uv-ir cut analysis. The 0.75mm is slightly worse than the mediocre M9 above 700nm. If I do it again, I'll choose 0.85mm over the 0.75mm.
Lot of hurdle and headache. Is it worth to do it? I don't know but I don't like what they say "m-lenses are best used on m-cameras" ...Show more →
I had the PCX 2500 on the ZM 25 2.8 on a stock A7rII and got really good results with it. Perfect extreme corners at 5.6. I since sold the lens but the PCX 2500 already mounted in a filter setup is on the buy and sell.
