Understanding LW/PH values

[Callorenzo]

Good afternoon everyone,

My question relates to LW/PH values, and specifically APSC vs Full Frame.

The values I am using all come from the MTF charts of Ephotozine. Yes, I do realise that these are all different values and they are taken using different cameras, however, what I would like to know is if there is a different between a score of 3000 taken on an APSC lens and 3000 on a full frame lens. I understand that the picture size itself is also different, but would you expect the lens to perform the same, picking up as much details, if they were both scoring 3000?

I am actively comparing the image quality of the Pentax K1/K1 mk 2 and the Fuji XT5. I am seeing higher scores for some of the APSC lenses, but I know full frame has better quality in general than APSC. Scoring between these two cameras is very similar overall in various lab tests (of different areas within the cameras), but when I look at sample photos- I tend to see that the images have much higher quality on the K1. Removing other factors such as dynamic range, would you expect to see similar image quality between these two cameras if the lens were resolving the same LW/PH?

Thank you!

 

[a_c_skinner]

"I know full frame has better quality in general than APSC".

I think that assumption is a problem. In truth APSC isn't that much smaller* and the main determinants are number of pixels (same in your comparison) and lens performance (some Fuji lenses are very good indeed) and technique (always very variable outside the laboritory).

Either camera will be excellent and going from one to the other for IQ would be a disappointment.

*What seems like a big difference on paper is rarely as obvious in the final images through all photography.

 

[D Cox]

What are LW/PH values ?

Don

 

[GaryW]

Good afternoon everyone,

My question relates to LW/PH values, and specifically APSC vs Full Frame.

The values I am using all come from the MTF charts of Ephotozine. Yes, I do realise that these are all different values and they are taken using different cameras, however, what I would like to know is if there is a different between a score of 3000 taken on an APSC lens and 3000 on a full frame lens. I understand that the picture size itself is also different, but would you expect the lens to perform the same, picking up as much details, if they were both scoring 3000?

Since the PH is different between the two sensors, I would think that if you used the same lens, with 3000 on a FF camera, you'd get a lot less on the APS-C camera. The smaller sensor is a crop of the larger one. I'm sure in reality, it's more complicated than that, but I'm pretty sure you need more resolution in the APS-C lens to keep up. But if you can do that, then sure, you could have APS-C having similar resolution to a FF setup.

I am actively comparing the image quality of the Pentax K1/K1 mk 2 and the Fuji XT5. I am seeing higher scores for some of the APSC lenses, but I know full frame has better quality in general than APSC. Scoring between these two cameras is very similar overall in various lab tests (of different areas within the cameras), but when I look at sample photos- I tend to see that the images have much higher quality on the K1. Removing other factors such as dynamic range, would you expect to see similar image quality between these two cameras if the lens were resolving the same LW/PH?

I think most lens reviews say something like "don't compare between cameras", much less different sensor sizes. So many variables. Like, to do a closer comparison, you probably want a lot more pixels in the FF sensor, so that the actual sensor pixel pitch is the same on both. Then the APS-C should be the same as the crop of the FF. Meaning the FF image has more resolution. But if both cameras are 25mp, and the APS-C lens is sharper than what's needed for FF, APS-C might score as well.



You've mentioned "the lens" a couple of times, but I think the key is that it's not the same lens,which is a big difference. I'm not sure what you mean as to a big IQ difference. Do you mean overall, or are you comparing 100% crops for detail? Could be that the larger sensor can also resolve better color too? Or you prefer the way it maps the tones. Noise will be better on FF. And depending on the scene, dynamic range might matter. If both have the same 3000 LW/PH, then detail should be very similar. Ideally, I'd want to take a photo of the same scene, at the same time, with the two cameras and compare.

Thank you!

 

[FingerPainter]

Good afternoon everyone,

My question relates to LW/PH values, and specifically APSC vs Full Frame.

The values I am using all come from the MTF charts of Ephotozine. Yes, I do realise that these are all different values and they are taken using different cameras, however, what I would like to know is if there is a different between a score of 3000 taken on an APSC lens and 3000 on a full frame lens. I understand that the picture size itself is also different, but would you expect the lens to perform the same, picking up as much details, if they were both scoring 3000?

I wouldn't expect "the lens" meaning the same single lens used on both cameras, to get the same LW/PH score on both cameras. A lens will cast a number of LW/mm on the sensor. Multiply that by the sensor height and you get LW/PH cast onto the sensor. Digitize that with pixels and you get the LW/PH measured in the digital image. Digitization inherently reduces the LW/PH achievable, but the greater the pixel count the smaller the loss, but with diminishing returns. It would take a very large advantage in pixel resolution for the smaller sensor to get the same LW/PH as the larger sensor from the same lens - probably not achievable with the range of pixel counts on current sensors..

I am actively comparing the image quality of the Pentax K1/K1 mk 2 and the Fuji XT5.

The difference in lineal pixel resolution between a K1 II and a X-T5 is negligible, so any significant difference in score is due almost entirely to differences in sensor sizes difference (in favour of the K1) and in lens performance (could be in favour of either).

I am seeing higher scores for some of the APSC lenses, but I know full frame has better quality in general than APSC.

But not all FF lenses have better quality that all APS-C lenses. It is easier to make a sharper lens if it has a smaller image circle. In general, lenses made for smaller formats are shaper (in terms of LW/mm cast onto the sensor), but also, in general, not so much sharper as to overcome the effect the difference in sensor size makes to the calculation of LW/PH from LW/mm.

Very good lenses on smaller formats may provide a higher LW/PH score than poor lenses on larger formats, if the sensor pixel counts are similar. But in general, a low -quality lens for a smaller format will score less well in LW/PH than a low-quality lens on a larger format, and a high-qualify tens for a larger format will score higher than a high-quality lens for a smaller format

Scoring between these two cameras is very similar overall in various lab tests (of different areas within the cameras), but when I look at sample photos- I tend to see that the images have much higher quality on the K1.

The are several different aspects of image quality. LW/PH rates only sharpness, and only under ideal light.

Removing other factors such as dynamic range, would you expect to see similar image quality between these two cameras if the lens were resolving the same LW/PH?

No, you also have to remove all other aspects of IQ except LW/PH (including colour gradation and accuracy and noise) to expect the same IQ given the same LW/PH scores.

Then also consider the non-optical factors that influence the quality of the image, such a a focus accuracy, timing of the phot, etc.

Thank you!

 

[mailham]

What's LW/PH?

 

[bastibe]

What are LW/PH values ?

Don

Line Widths per Picture Height. To test resolution, they photograph a dense pattern of alternating black and white lines. As the lines get closer and closer, at some point they start to blur together. At the point where the contrast between black and white has diminished to 50%, we mark the line density as the resolution of the lens.

Sensor resolution is a limiting factor. A 24 MP 2x3 sensor has 4000x6000 pixels. That's 4000 pixels vertically. So at most, it can resolve 4000 LW/PH, exactly alternating black and white. In practice, demosaicing will reduce achievable resolution by about 30%, so 3000 is a more realistic maximum. A perfect lens will resolve just that. Obviously, cameras with more megapixels allow more resolution to be measured.

Theoretically, a lens could resolve many more lines than what is measurable like this, but measuring that would require specialized equipment (as opposed to merely a regular camera body), and is generally too expensive for normal review websites.

 

[bodeswell]

What are LW/PH values ?

Don

Line Widths per Picture Height. To test resolution, they photograph a dense pattern of alternating black and white lines. As the lines get closer and closer, at some point they start to blur together. At the point where the contrast between black and white has diminished to 50%, we mark the line density as the resolution of the lens.

Sensor resolution is a limiting factor. A 24 MP 2x3 sensor has 4000x6000 pixels. That's 4000 pixels vertically. So at most, it can resolve 4000 LW/PH, exactly alternating black and white. In practice, demosaicing will reduce achievable resolution by about 30%, so 3000 is a more realistic maximum. A perfect lens will resolve just that. Obviously, cameras with more megapixels allow more resolution to be measured.

Theoretically, a lens could resolve many more lines than what is measurable like this, but measuring that would require specialized equipment (as opposed to merely a regular camera body), and is generally too expensive for normal review websites.

But if a lens could resolve many more lines than common review website tools can measure, doesn’t that call the significance of those reviews into question? Are all those pretty charts and graphs just window dressing?

 

[chrisfisheye]

Good afternoon everyone,

My question relates to LW/PH values, and specifically APSC vs Full Frame.

The values I am using all come from the MTF charts of Ephotozine. Yes, I do realise that these are all different values and they are taken using different cameras, however, what I would like to know is if there is a different between a score of 3000 taken on an APSC lens and 3000 on a full frame lens. I understand that the picture size itself is also different, but would you expect the lens to perform the same, picking up as much details, if they were both scoring 3000?

I am actively comparing the image quality of the Pentax K1/K1 mk 2 and the Fuji XT5. I am seeing higher scores for some of the APSC lenses, but I know full frame has better quality in general than APSC. Scoring between these two cameras is very similar overall in various lab tests (of different areas within the cameras), but when I look at sample photos- I tend to see that the images have much higher quality on the K1. Removing other factors such as dynamic range, would you expect to see similar image quality between these two cameras if the lens were resolving the same LW/PH?

Thank you!

If both have 3000 LW/PH then this is the same sharpness. The result between APS-C and FF will be quite similar.

Be aware that you usually do not use the same f# between FF and APS-C to take a similar image. If you use f/2 with APS-C, there is a good chance you will use f/2.8 with FF to take a similar image, which is less extreme. FF might be sharper at these extremes (at equivalent aperture)

 

[bastibe]

What are LW/PH values ?

Don

Line Widths per Picture Height. To test resolution, they photograph a dense pattern of alternating black and white lines. As the lines get closer and closer, at some point they start to blur together. At the point where the contrast between black and white has diminished to 50%, we mark the line density as the resolution of the lens.

Sensor resolution is a limiting factor. A 24 MP 2x3 sensor has 4000x6000 pixels. That's 4000 pixels vertically. So at most, it can resolve 4000 LW/PH, exactly alternating black and white. In practice, demosaicing will reduce achievable resolution by about 30%, so 3000 is a more realistic maximum. A perfect lens will resolve just that. Obviously, cameras with more megapixels allow more resolution to be measured.

Theoretically, a lens could resolve many more lines than what is measurable like this, but measuring that would require specialized equipment (as opposed to merely a regular camera body), and is generally too expensive for normal review websites.

But if a lens could resolve many more lines than common review website tools can measure, doesn’t that call the significance of those reviews into question? Are all those pretty charts and graphs just window dressing?

Why? Most lenses don't achieve 3000 lw/ph. Just because some specialized lenses might, doesn't invalidate all the rest.

 

[FingerPainter]

What's LW/PH?

Line Widths per Picture Height.

It is a measure of image sharpness.

 

[alcelc]

Yes, the limitation of sensor could lead to this sort of limitation.

However, if comparing different lenses on the same camera, the result could tell the differences of lenses. It is still useful but, must be careful on how the result was produced.

 

[Thom Hogan]

My question relates to LW/PH values, and specifically APSC vs Full Frame.

There's a lot of poor information in this thread. Ephotozine uses Imatest. I suggest you read Imatest's article on Slanted Edge results (https://www.imatest.com/docs/sharpness/imatest-slanted-edge-results/).

And that's the first isssue: some of you are mixing up lppm with lp/ph. What I've found is that the quality of your slanted edge chart and lighting of it will change results, too.

As I've written before, measured results like this are only useful if there are no variable changes involved in the testing. Even small things like accuracy of focus, light, chart quality/cleanliness, perpendicularness, etc., can make results not directly comparable.

 

[GaryW]

What are LW/PH values ?

Don

Line Widths per Picture Height. To test resolution, they photograph a dense pattern of alternating black and white lines. As the lines get closer and closer, at some point they start to blur together. At the point where the contrast between black and white has diminished to 50%, we mark the line density as the resolution of the lens.

Sensor resolution is a limiting factor. A 24 MP 2x3 sensor has 4000x6000 pixels. That's 4000 pixels vertically. So at most, it can resolve 4000 LW/PH, exactly alternating black and white. In practice, demosaicing will reduce achievable resolution by about 30%, so 3000 is a more realistic maximum. A perfect lens will resolve just that. Obviously, cameras with more megapixels allow more resolution to be measured.

Theoretically, a lens could resolve many more lines than what is measurable like this, but measuring that would require specialized equipment (as opposed to merely a regular camera body), and is generally too expensive for normal review websites.

But if a lens could resolve many more lines than common review website tools can measure, doesn’t that call the significance of those reviews into question? Are all those pretty charts and graphs just window dressing?

They would still have value as long as you compare using the same methodology. I mean, there's a reason some lenses will be sharper and others not. It's not like your common tools are going to randomly pick a blurry lens to be sharp, and then consistently, but randomly, do it again. ;-)

I think what is meant is that you can't know the full, ultimate sharpness of a lens unless a sensor outresolves it. So, given a set of tests, if a number of lenses test at the top, you might wonder if there might be a bigger spread, if you could measure them with more detail.

But if I want to know which of my lenses are sharper than others, I can run my tests on my camera. It's fine.

 

[bodeswell]

What are LW/PH values ?

Don

Line Widths per Picture Height. To test resolution, they photograph a dense pattern of alternating black and white lines. As the lines get closer and closer, at some point they start to blur together. At the point where the contrast between black and white has diminished to 50%, we mark the line density as the resolution of the lens.

Sensor resolution is a limiting factor. A 24 MP 2x3 sensor has 4000x6000 pixels. That's 4000 pixels vertically. So at most, it can resolve 4000 LW/PH, exactly alternating black and white. In practice, demosaicing will reduce achievable resolution by about 30%, so 3000 is a more realistic maximum. A perfect lens will resolve just that. Obviously, cameras with more megapixels allow more resolution to be measured.

Theoretically, a lens could resolve many more lines than what is measurable like this, but measuring that would require specialized equipment (as opposed to merely a regular camera body), and is generally too expensive for normal review websites.

But if a lens could resolve many more lines than common review website tools can measure, doesn’t that call the significance of those reviews into question? Are all those pretty charts and graphs just window dressing?

Why? Most lenses don't achieve 3000 lw/ph. Just because some specialized lenses might, doesn't invalidate all the rest.

Not sure I follow. Your point seemed to be that “a lens” could do better than what some review site could measure, not that “specialized lenses” could do better than 3000 lw/ph. Also, lots of lenses blow right by 3000 on modest revolution cameras, less than 24mp, even inexpensive lenses. Check out the old reviews on photozone.de. Not saying their charts aren’t eye candy too, just saying I don’t follow your comment.

 

[GaryW]

What are LW/PH values ?

Don

Line Widths per Picture Height. To test resolution, they photograph a dense pattern of alternating black and white lines. As the lines get closer and closer, at some point they start to blur together. At the point where the contrast between black and white has diminished to 50%, we mark the line density as the resolution of the lens.

Sensor resolution is a limiting factor. A 24 MP 2x3 sensor has 4000x6000 pixels. That's 4000 pixels vertically. So at most, it can resolve 4000 LW/PH, exactly alternating black and white. In practice, demosaicing will reduce achievable resolution by about 30%, so 3000 is a more realistic maximum. A perfect lens will resolve just that. Obviously, cameras with more megapixels allow more resolution to be measured.

Theoretically, a lens could resolve many more lines than what is measurable like this, but measuring that would require specialized equipment (as opposed to merely a regular camera body), and is generally too expensive for normal review websites.

But if a lens could resolve many more lines than common review website tools can measure, doesn’t that call the significance of those reviews into question? Are all those pretty charts and graphs just window dressing?

Why? Most lenses don't achieve 3000 lw/ph. Just because some specialized lenses might, doesn't invalidate all the rest.

Not sure I follow. Your point seemed to be that “a lens” could do better than what some review site could measure, not that “specialized lenses” could do better than 3000 lw/ph. Also, lots of lenses blow right by 3000 on modest revolution cameras, less than 24mp, even inexpensive lenses. Check out the old reviews on photozone.de. Not saying their charts aren’t eye candy too, just saying I don’t follow your comment.

For example, here they review one of Sony's worst lenses... also one of the cheapest, it should be mentioned... and they still measure a peak of 3281. They did use a 24mp camera, and their tests correctly find the low resolution of the corners. How can most lenses (at least modern ones) NOT achieve 3000 lw/ph if this lens can?!

Sony E 16mm f/2.8 (SEL-16F28) - Review / Re-Test - Analysis

Sony E 16mm f/2.8 (Sony NEX) Review

photozone.de

Anyway, probably best to only compare Photozone/Optical Limits tests to their own other tests.

When I convert my last test values to lw/ph, the highest value I got from my sharpest lens was 2,577. I'm sure that's low, but I'm limited by my low-budget testing setup. (There are other factors too -- if I used sharpened, in-camera JPEGs, I'd get higher numbers.) Converting my results for the 16/f2.8, at f5.6, I get 2,311. While the number is way different than the Optical Limits results, both of us have consistent findings within our own tests. These peak values are at f5.6.