-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Comparing Formats
- Thread starter Doppler9000
- Start date
How much MTF above Nyquist do you need for moire to appear? Significant is not a very scientific quantification
For moiré to appear you need MTF@Nyquist>0
Usually engineers relate to MFT50 to define resolution i.e. 50% constrast a a given cy/mm, I know of no lens can get this high at Nyquist with those sensors (P45, A7RIV), but that's a technicality
Furthermore MTF@Nyquist>0 is valid for a monochrome sensor, don't forget that we have a Bayer matrix here i.e. FNyquist will be half of the sensor's for the blue and red channel, so that color moiré will appear much sooner than when the lens will "outresolve" the sensor
Anyway, my point was that moiré is not a good indication of lens/sensor adequacy
5% to 10%.
If the sensor and the CFA are considered together, the Nyquist frequency for the combination is less than that of a monochrome sensor of the same pitch.
It’s not a bad indicator of the lens having too much resolution for the sensor, at least with the subject that provoked the moiré. Of course you can get aliasing without moiré.
Do you know about image system Q?
--
https://blog.kasson.com
Last edited:
Erik Kaffehr
Veteran Member
Something like 10% I would say
Right
No, engineers don't define resolution as 50% contrast at a given cy/mm. The mostly used 'cut off' for resolution is 20% MTF, often used in machine vision. It may be dubious to say that a lens resolves say 50 cy/mm at 50% MTF, but it may be technically adequate.
System resolution is always a combination of lens resolution and sensor resolution as the pixel opening blurs the image.
I see your reasoning, but that may still not be correct.
My point is that you are wrong.
Best regards
Erik
Are you assuming that many who shoot 44x33 do not also shoot FF, etc.?
Obviously FF has changed a lot over the last decade. 44x33 has not. In almost the same amount of time, the only notable difference on the 44x33 front has been the 100mp GFX. The Sony sensor in cameras like the Hasselblad 50c, Pentax 645z and the first Fuji GFX aren't quite a decade old, but practically close to it. What was a notable performance/resolution gulf between FF and 44x33 in 2014 is not indicative of the same permanence difference today.
You can't compare a 44X33 dinosaur (wonderfully competent dinosaurs) to a modern "full frame" camera and reasonably expect the same performance gap that was experienced nearly 10 years ago. That would be like comparing your grandmother's professional sewing machine, to a modern Juki offered today... or comparing a Ferrari 328 GTSI over the latest Honda Civic R and wondering why someone isn't still singing performance-based praises for the old Ferrari...
--
Teila K. Day
http://teiladay.com
Last edited:
Erik Kaffehr
Veteran Member
These MTF plots compare Planar 100 at f/5.6 on P45+ with the Voigtlander 65 on the A7rIV, I use 20% as 'cut off' for resolution. The Planar keeps up well with the Voigtlander on the A7rIV. The A7rIV has 60 MP while the P45+ has only 39 MP. That may be the major difference.
Comparing the Voigtlander on the A7rIV to GFX 50S with the 63/2.8 lens. The two are very at 50% MTF, but the GFX 50S holds a small advantage below and above. The GFX 50 has undersize microlenses and will have more aliasing at Nyquist (the vertical lines). The comparison is calculated on the short edge, the A7rIV has more pixels on the horisontal side, therefor it has higher MP count above 20%.
In practice, I would say that the systems would be very close.
I would be pretty sure that the data are pretty comparable. The slanted edge I have used in my test and from DPR's test also have an MTF and that may cause a difference.
What I think sets the Hasselblad behind in real life are two factors:
- Difficult to achieve perfect focus on the ground glass.
- The lens tested here is sharp across the field, but many Hasselblad lenses vary quite a lot over the field.
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
Last edited:
Erik Kaffehr
Veteran Member
It has been suggested that it is incorrect to compare 'film era gear' with modern gear. There is something to it.
Some Hasselblad lenses were quite good, even with modern standards, one of the better ones is the Planar 100/3.5. It was intended for long distance work and it was one of the sharpest lenses for the Hasselblad V-series.
These MTF numbers are good, but not spectacular by today's standards.
Now, if we put that lens on 2007 era P45+ sensor and compare it to DPreview's test of the GFX 50S, what do we get?
We can see that the GFX 50S is a bit sharper across the frequency range. At lower frequencies the difference seems to be small. At higher frequencies the higher MP count paired with the undersized pixel aperture benefits the GFX. The GFX 50S has pretty high MTF at Nyquist that will yield aliasing artifacts on any subject with high frequency detail.
Now, lets compare the GFX 50S to high resolution 24x36 mm camera:
Here, blue line is the Voigtlander 65/2 APO Lanthar on the A7rIV compared to the Fujinon 63/2.8 GF on GFX 50S. The GFX 50S has a very thin advantage over smaller format camera.
Finally comparing the A7rIV to the Phase One IQ 4150, here the IQ 4150 is the blue line. At high frequencies the 54x41 mm having 150 MP of resolution has much higher contrast. That combo, as tested in the DPReview studio shot, seems to be a bit limited by the lens. Having low contrast at Nyquist (the blue vertical line) will reduce aliasing artifacts to a minimum.
There is more to image quality than sharpness and resolution.
A larger sensor may have benefits for signal noise ratio. Assuming the same underlying sensor technology, doubling the sensor area improves SNR by 41%.
It may be that older sensor have lower SNR for the same sensor area. Need to do more experimentation to look into that.
An area where my P45+ is left behind is DR (Dynamic Range). Modern sensors have much reduced readout noise.
But, I think that DR is not very relevant in many cases, and there are also a range of misconceptions about DR.
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
Some Hasselblad lenses were quite good, even with modern standards, one of the better ones is the Planar 100/3.5. It was intended for long distance work and it was one of the sharpest lenses for the Hasselblad V-series.
These MTF numbers are good, but not spectacular by today's standards.
Now, if we put that lens on 2007 era P45+ sensor and compare it to DPreview's test of the GFX 50S, what do we get?
We can see that the GFX 50S is a bit sharper across the frequency range. At lower frequencies the difference seems to be small. At higher frequencies the higher MP count paired with the undersized pixel aperture benefits the GFX. The GFX 50S has pretty high MTF at Nyquist that will yield aliasing artifacts on any subject with high frequency detail.
Now, lets compare the GFX 50S to high resolution 24x36 mm camera:
Here, blue line is the Voigtlander 65/2 APO Lanthar on the A7rIV compared to the Fujinon 63/2.8 GF on GFX 50S. The GFX 50S has a very thin advantage over smaller format camera.
Finally comparing the A7rIV to the Phase One IQ 4150, here the IQ 4150 is the blue line. At high frequencies the 54x41 mm having 150 MP of resolution has much higher contrast. That combo, as tested in the DPReview studio shot, seems to be a bit limited by the lens. Having low contrast at Nyquist (the blue vertical line) will reduce aliasing artifacts to a minimum.
There is more to image quality than sharpness and resolution.
A larger sensor may have benefits for signal noise ratio. Assuming the same underlying sensor technology, doubling the sensor area improves SNR by 41%.
It may be that older sensor have lower SNR for the same sensor area. Need to do more experimentation to look into that.
An area where my P45+ is left behind is DR (Dynamic Range). Modern sensors have much reduced readout noise.
But, I think that DR is not very relevant in many cases, and there are also a range of misconceptions about DR.
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
Last edited:
Erik Kaffehr
Veteran Member
We may ask, how reliable are these data?
Slanted edge measurements are quite reliable, my guess would be within 1% or so.
The final factor is diffraction. On high end lenses, diffraction comes into play at f/8 or even f/5.6. In a way, the better corrected a lens is, the more we lose stopping down.
I have used f/4 on 24x36 mm and f/5.6 on my P45+.
In the end, I found that in most cases I get better results from my Sony gear than from my Hasselblad gear. I would think the major factors are:
Making best use of gear may matter more than having the best gear.
Best regards
Erik
Slanted edge measurements are quite reliable, my guess would be within 1% or so.
- But, much depends on the input data. I am using Lightroom for demosaic, with no sharpening applied. But, Lightroom seems to do some kind of sharpening under the skin.
- The next question is focusing accuracy. On the Sony A7rIV I would use motor controlled macro rail and typically shoot with 2.5 mm intervals at 50X focal distance and take the next image.
- With the Hasselblad, I take fewer images.
- I don't know how DPReviews optimizes focusing on their test images.
- Anyway, focusing here is much more accurate that what we may achieve in real world shooting situations. Although, on the A7rIV I almost always use peaking at maginified live view at shooting aperture, that is about as accurate we can get.
- There are quite a lot of variations between lenses.
- I think he uses undemosaiced data, while I use demosaiced data using Lightroom.
- Jim uses a weighted average of the channel data, while I only look at the green channel.
The final factor is diffraction. On high end lenses, diffraction comes into play at f/8 or even f/5.6. In a way, the better corrected a lens is, the more we lose stopping down.
I have used f/4 on 24x36 mm and f/5.6 on my P45+.
In the end, I found that in most cases I get better results from my Sony gear than from my Hasselblad gear. I would think the major factors are:
- Focusing accuracy, magnified live view combined with peaking at shooting aperture, it is as good we get. My Hasselblad/P45+ has no functional live view.
- Non optimal aperture. I would normally stop down the Hasselblad to f/11.
- Some of the Hasselblad lenses are not so great. The Sonnars 180/4 and 150/4 I found to be excellent and that also applies to the Planar 100/3.5 at long distance.
- The other lenses I have owned, like 40/4 CF, 50/4 CF, 60/3.5CF, 80/2.8 CFE are not so great off axis.
- We also have the Macro Planar 120/4 CFi. At infinity it has a lot of field curvature. It seems that it is quite OK for close up work.
Making best use of gear may matter more than having the best gear.
Best regards
Erik
Erik Kaffehr
Veteran Member
I wouldn't mind finer pitches... Or, OLP filters with the pitches we have.
Best regards
Erik
Eric if you compared the latest HB XCDII and lenses to your Sony would you think it may change your view of your results? More modern lens and sensor. Maybe you have answered this in your discussion. I have to admit all the technical stuff is above my learning and I know simplification of information is nuanced and sometimes difficult. Thanks for you reviews though.
Erik Kaffehr
Veteran Member
Well, it seems coming from DPReview Studio Scene images that the GFX 50S has a very small advantage over the A7rIV, when the latter is using a top notch lens.
Comparing the GFX 50S to my 'blad' had some advantage, but I don't know if I would call that significant. But, I will check back on that tomorrow.
Would i print large, like 80x120 cm, it may make some sense to go with the GFX 100S, I do actually think that Megapixels can matter.
I do think that Hasselblad needs to update their sensor.
No doubt, it can deliver excellent images. But I am in some doubt that the sensor is a good match for the lenses.
There are of course other factors that may be in favor of the XD1II, like the leaf shutter.
But, for me lens choice and flexibility matters more than small advantages in image quality.
Best regards
Erik
Erik Kaffehr
Veteran Member
Seems that my Sony A7rIV with the Voigtlander 65 is a dead ringer for the X1D as tested by DPReview.
But, DPR tested at f/8 and I tested at f/4. I would guess the X1D would perform higher at f/5.6 or f/4. Diffraction plays a role.
Looking at DPReviews test of the Sony A7rIV we get similar results:
Here the X1D seems to be a tiny bit sharper, but I don't think that would be visible in prints.
I may look into SNR, too. DxO-mark reports 0.6 DB advantage to the X1D over A7rIV.
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
Last edited:
Erik Kaffehr
Veteran Member
I have calculated MTF using 'slanted edge' measurements for different systems, mostly based on DPReview's Studio Scene.
The results may be a bit surprising.
Regarding the GFX 100, Jim Kasson did a very good comparison between the GFX 50S and the GFX 100, with a somewhat surprising result:
blog.kasson.com
Jim's article indicates that the GFX 50S and the GFX 100 would yield the same sharpness at the picture level, due to differences in the microlenses over each sensor.
The data I calculated shows much better sharpness for the GFX 100 compared to the GFX 50S.
The GFX 100 was tested at f/5.6 while the GFX 50 was tested at f/8, but I don't think that diffraction alone can explain the difference between Jim's data and what I can calculate from the DPR test images.
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
The results may be a bit surprising.
- There is not a lot of difference between the 50 MP systems, the A7rIV seems to be in the same group.
- The Phase One P45+, paired with the Zeiss 100/3.5 CF is a bit below, while the 100 MP GFX 100 and 150 MP Phase One IQ 4150 performs higher.
- The GFX 100 with the 120/4 macro at f/5.6 seems to be sharper than the Phase One IQ 4150 with the 150/2.8 at f/8.
Regarding the GFX 100, Jim Kasson did a very good comparison between the GFX 50S and the GFX 100, with a somewhat surprising result:
Fujifilm GFX 100 sharpness compared to GFX 50s - the last word
This is one in a series of posts on the Fujifilm GFX 100. You should be able to find all the posts about that camera in the Category List on the right sidebar, below the Articles widget. There’s a drop-down menu there that you can use to get to all the posts in this series;
blog.kasson.com
Jim's article indicates that the GFX 50S and the GFX 100 would yield the same sharpness at the picture level, due to differences in the microlenses over each sensor.
The data I calculated shows much better sharpness for the GFX 100 compared to the GFX 50S.
The GFX 100 was tested at f/5.6 while the GFX 50 was tested at f/8, but I don't think that diffraction alone can explain the difference between Jim's data and what I can calculate from the DPR test images.
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
I don't trust the focusing in the DPR test images for slanted edge testing.I have calculated MTF using 'slanted edge' measurements for different systems, mostly based on DPReview's Studio Scene.
The results may be a bit surprising.
I am a bit surprised that the around 50 MP systems bundle up as much as they do, but the real surprise is the GFX 100.
- There is not a lot of difference between the 50 MP systems, the A7rIV seems to be in the same group.
- The Phase One P45+, paired with the Zeiss 100/3.5 CF is a bit below, while the 100 MP GFX 100 and 150 MP Phase One IQ 4150 performs higher.
- The GFX 100 with the 120/4 macro at f/5.6 seems to be sharper than the Phase One IQ 4150 with the 150/2.8 at f/8.
Regarding the GFX 100, Jim Kasson did a very good comparison between the GFX 50S and the GFX 100, with a somewhat surprising result:
https://blog.kasson.com/gfx-100/fujifilm-gfx-100-sharpness-compared-to-gfx-50s/
Jim's article indicates that the GFX 50S and the GFX 100 would yield the same sharpness at the picture level, due to differences in the microlenses over each sensor.
The data I calculated shows much better sharpness for the GFX 100 compared to the GFX 50S.
The GFX 100 was tested at f/5.6 while the GFX 50 was tested at f/8, but I don't think that diffraction alone can explain the difference between Jim's data and what I can calculate from the DPR test images.
Best regards
Erik
--
https://blog.kasson.com
Last edited:
Erik Kaffehr
Veteran Member
I played with that explanation, too...I don't trust the focusing in the DPR test images for slanted edge testing.I have calculated MTF using 'slanted edge' measurements for different systems, mostly based on DPReview's Studio Scene.
The results may be a bit surprising.
I am a bit surprised that the around 50 MP systems bundle up as much as they do, but the real surprise is the GFX 100.
- There is not a lot of difference between the 50 MP systems, the A7rIV seems to be in the same group.
- The Phase One P45+, paired with the Zeiss 100/3.5 CF is a bit below, while the 100 MP GFX 100 and 150 MP Phase One IQ 4150 performs higher.
- The GFX 100 with the 120/4 macro at f/5.6 seems to be sharper than the Phase One IQ 4150 with the 150/2.8 at f/8.
Regarding the GFX 100, Jim Kasson did a very good comparison between the GFX 50S and the GFX 100, with a somewhat surprising result:
https://blog.kasson.com/gfx-100/fujifilm-gfx-100-sharpness-compared-to-gfx-50s/
Jim's article indicates that the GFX 50S and the GFX 100 would yield the same sharpness at the picture level, due to differences in the microlenses over each sensor.
The data I calculated shows much better sharpness for the GFX 100 compared to the GFX 50S.
The GFX 100 was tested at f/5.6 while the GFX 50 was tested at f/8, but I don't think that diffraction alone can explain the difference between Jim's data and what I can calculate from the DPR test images.
Best regards
Erik
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
Erik Kaffehr
Veteran Member
The data I plotted is not in agreement with measurements by Jim Kasson.
Jim has found that MTF for the GF 50S and GFX 100 is pretty similar, due to the differences in microlens design between the two systems.
The painful conclusion may be that the DPReview Studio Test image is not good enough for slanted edge evaluation, making my comparison pretty much irrelevant.
A couple of possible issues may be that slanted edge has to low resolution or that optimal focus may have not been achieved.
Best regards
Erik
Jim has found that MTF for the GF 50S and GFX 100 is pretty similar, due to the differences in microlens design between the two systems.
The painful conclusion may be that the DPReview Studio Test image is not good enough for slanted edge evaluation, making my comparison pretty much irrelevant.
A couple of possible issues may be that slanted edge has to low resolution or that optimal focus may have not been achieved.
Best regards
Erik
Both are possible. The edge resolution can be compensated for if it's constant, but focus variation cannot. I find it impossible to get consistent slanted edge results without some form of focus bracketing.
Erik Kaffehr
Veteran Member
Same here.
What confuses me a bit is that I got very consistent results for X1D, GFX 50S, Pentax 645Z. A significant variation in focusing, I would expect that to result in a significant variation in MTF.
Also, it seems that DPReview learned a lesson on focusing accuracy when testing the GFX 100 and I would hoped that they have not forgotten about that.
Their result for the GFX 50SII is quite low.
As a side note, I have found that the Planar 100/3.5 on the Hasselblad P45+ is a good match for the Voigtlander 65/2 APO on the A7rII, when shooting slanted edge and doing some bracketing.
But, I cannot reach the same results in the real world. I think that the Blad is just to difficult to focus exactly.
Best regards
Erik
Similar threads
Keyboard shortcuts
- f
- Forum