Stitch/merge musings and questions

JimKasson said:

DMillier said:

The proof of the pudding is in the eating as they say.

This image is an attempt to simulate a 36x36mm sq 60MP medium format image. It was made from 10 frames combined to produce a HDR pano in LR, saved as a dng. The dng was imported into ubuntu/darktable, edited and output as a 85% jpeg.

View attachment 340911358e8a452e8af177c7d7a2f098.jpg

Not the greatest weather for doing testing (threatening rain, overcast, dull). Colour is a bit odd as well on my screen. But the question is in terms of resolution, dynamic range, shadow noise etc, is this remotely medium format like or is this all a lot of effort for no gain?

Click to expand...

This is a credible image, albeit not one with a high subject DR -- it's actually the opposite. The lens is not particularly sharp, which served you well here in preventing aliasing in the tree branches. I'm guessing that in this cas, you could have done about as well by shooting full frame and upsizing in a good resampling program. Give that a try and see if the pano is enough better to make it worth your while.

Click to expand...

Jim commented that my lens isn't very sharp. This surprised me as it is a Contax G Carl Zeiss Planar 35mm f/2 which had a good reputation in its day. It's not supposedly quite as good as the 45mm, but still highly rated.

I decided to see whether it was possible to come to a more objective sense of how much my lens contributed to a lack of sharpness by testing it.

I followed Erik's suggested method here: https://www.dpreview.com/forums/post/65122260

I'm completely new to this so I have probably made loads of methodological blunders that completely invalidates the results but everyone has to start somewhere!

What I did:

1. Downloaded Erik's tilted square target.

2. Printed it out on A4 glossy paper

3. Taped it to a window

4. Set up my camera on tripod about 4 metres away

5. Photographed the target (which was very small in the frame) in the centre of the frame

6. Loaded the raw file into LR. Turned off sharpening and noise reduction

7. Cropped target to about a 100x200 px crop

8. Adjusted sliders until the slant edge crop had about the same tonal values as Erik's test crop (mine seemed noisier - does this matter?)

9. exported as a tiff

10. Loaded into mtf mapper

11. I got a MTF50 score of 0.27

What, if anything, that does this tell me?







--
DPReview gallery: https://www.dpreview.com/galleries/0286305481
Website: http://www.whisperingcat.co.uk/ (2018 - website revived!)
Flickr: http://www.flickr.com/photos/davidmillier/ (very old!)

DMillier said:

JimKasson said:

DMillier said:

The proof of the pudding is in the eating as they say.

This image is an attempt to simulate a 36x36mm sq 60MP medium format image. It was made from 10 frames combined to produce a HDR pano in LR, saved as a dng. The dng was imported into ubuntu/darktable, edited and output as a 85% jpeg.

View attachment 340911358e8a452e8af177c7d7a2f098.jpg

Not the greatest weather for doing testing (threatening rain, overcast, dull). Colour is a bit odd as well on my screen. But the question is in terms of resolution, dynamic range, shadow noise etc, is this remotely medium format like or is this all a lot of effort for no gain?

Click to expand...

This is a credible image, albeit not one with a high subject DR -- it's actually the opposite. The lens is not particularly sharp, which served you well here in preventing aliasing in the tree branches. I'm guessing that in this cas, you could have done about as well by shooting full frame and upsizing in a good resampling program. Give that a try and see if the pano is enough better to make it worth your while.

Click to expand...

Jim commented that my lens isn't very sharp. This surprised me as it is a Contax G Carl Zeiss Planar 35mm f/2 which had a good reputation in its day. It's not supposedly quite as good as the 45mm, but still highly rated.

I decided to see whether it was possible to come to a more objective sense of how much my lens contributed to a lack of sharpness by testing it.

I followed Erik's suggested method here: https://www.dpreview.com/forums/post/65122260

I'm completely new to this so I have probably made loads of methodological blunders that completely invalidates the results but everyone has to start somewhere!

What I did:

1. Downloaded Erik's tilted square target.

2. Printed it out on A4 glossy paper

3. Taped it to a window

4. Set up my camera on tripod about 4 metres away

5. Photographed the target (which was very small in the frame) in the centre of the frame

6. Loaded the raw file into LR. Turned off sharpening and noise reduction

7. Cropped target to about a 100x200 px crop

8. Adjusted sliders until the slant edge crop had about the same tonal values as Erik's test crop (mine seemed noisier - does this matter?)

9. exported as a tiff

10. Loaded into mtf mapper

11. I got a MTF50 score of 0.27

Click to expand...

What are the dimensions of that? Cycles per pixel? If so, what is the pixel pitch of your camera?

It is better to look at the raw planes directly and avoid Lr entirely.

DavidMillier said:

What, if anything, that does this tell me?

Click to expand...

--

JimKasson said:

DMillier said:

JimKasson said:

DMillier said:

The proof of the pudding is in the eating as they say.

This image is an attempt to simulate a 36x36mm sq 60MP medium format image. It was made from 10 frames combined to produce a HDR pano in LR, saved as a dng. The dng was imported into ubuntu/darktable, edited and output as a 85% jpeg.

View attachment 340911358e8a452e8af177c7d7a2f098.jpg

Not the greatest weather for doing testing (threatening rain, overcast, dull). Colour is a bit odd as well on my screen. But the question is in terms of resolution, dynamic range, shadow noise etc, is this remotely medium format like or is this all a lot of effort for no gain?

Click to expand...

This is a credible image, albeit not one with a high subject DR -- it's actually the opposite. The lens is not particularly sharp, which served you well here in preventing aliasing in the tree branches. I'm guessing that in this cas, you could have done about as well by shooting full frame and upsizing in a good resampling program. Give that a try and see if the pano is enough better to make it worth your while.

Click to expand...

Jim commented that my lens isn't very sharp. This surprised me as it is a Contax G Carl Zeiss Planar 35mm f/2 which had a good reputation in its day. It's not supposedly quite as good as the 45mm, but still highly rated.

I decided to see whether it was possible to come to a more objective sense of how much my lens contributed to a lack of sharpness by testing it.

I followed Erik's suggested method here: https://www.dpreview.com/forums/post/65122260

I'm completely new to this so I have probably made loads of methodological blunders that completely invalidates the results but everyone has to start somewhere!

What I did:

1. Downloaded Erik's tilted square target.

2. Printed it out on A4 glossy paper

3. Taped it to a window

4. Set up my camera on tripod about 4 metres away

5. Photographed the target (which was very small in the frame) in the centre of the frame

6. Loaded the raw file into LR. Turned off sharpening and noise reduction

7. Cropped target to about a 100x200 px crop

8. Adjusted sliders until the slant edge crop had about the same tonal values as Erik's test crop (mine seemed noisier - does this matter?)

9. exported as a tiff

10. Loaded into mtf mapper

11. I got a MTF50 score of 0.27

Click to expand...

What are the dimensions of that? Cycles per pixel?

Click to expand...

According to Erik's tutorial post 50% MTF cy/px

JimKasson said:

If so, what is the pixel pitch of your camera?

Click to expand...

Apparently it's 4.5 µm

JimKasson said:

It is better to look at the raw planes directly and avoid Lr entirely.

Click to expand...

I found this method described by Jack Hogan

http://www.strollswithmydog.com/how-to-get-mtf-curves-for-your-camera-and-lens/

Would this be a way to avoid LR?

JimKasson said:

JimKasson said:

What, if anything, that does this tell me?

Click to expand...

Click to expand...

--
DPReview gallery: https://www.dpreview.com/galleries/0286305481
Website: http://www.whisperingcat.co.uk/ (2018 - website revived!)
Flickr: http://www.flickr.com/photos/davidmillier/ (very old!)

DMillier said:

JimKasson said:

DMillier said:

The proof of the pudding is in the eating as they say.

This image is an attempt to simulate a 36x36mm sq 60MP medium format image. It was made from 10 frames combined to produce a HDR pano in LR, saved as a dng. The dng was imported into ubuntu/darktable, edited and output as a 85% jpeg.

View attachment 340911358e8a452e8af177c7d7a2f098.jpg

Not the greatest weather for doing testing (threatening rain, overcast, dull). Colour is a bit odd as well on my screen. But the question is in terms of resolution, dynamic range, shadow noise etc, is this remotely medium format like or is this all a lot of effort for no gain?

Click to expand...

This is a credible image, albeit not one with a high subject DR -- it's actually the opposite. The lens is not particularly sharp, which served you well here in preventing aliasing in the tree branches. I'm guessing that in this cas, you could have done about as well by shooting full frame and upsizing in a good resampling program. Give that a try and see if the pano is enough better to make it worth your while.

Click to expand...

Jim commented that my lens isn't very sharp. This surprised me as it is a Contax G Carl Zeiss Planar 35mm f/2 which had a good reputation in its day. It's not supposedly quite as good as the 45mm, but still highly rated.

I decided to see whether it was possible to come to a more objective sense of how much my lens contributed to a lack of sharpness by testing it.

I followed Erik's suggested method here: https://www.dpreview.com/forums/post/65122260

I'm completely new to this so I have probably made loads of methodological blunders that completely invalidates the results but everyone has to start somewhere!

What I did:

1. Downloaded Erik's tilted square target.

2. Printed it out on A4 glossy paper

3. Taped it to a window

Click to expand...

Not intended to be used with backlight, but no reason it would not work.

DavidMillier said:

4. Set up my camera on tripod about 4 metres away

5. Photographed the target (which was very small in the frame) in the centre of the frame

6. Loaded the raw file into LR. Turned off sharpening and noise reduction

Click to expand...

OK!

DavidMillier said:

7. Cropped target to about a 100x200 px crop

Click to expand...

OK!

DavidMillier said:

8. Adjusted sliders until the slant edge crop had about the same tonal values as Erik's test crop (mine seemed noisier - does this matter?)

Click to expand...

Does not matter a lot, but adjust as little as possible.

DavidMillier said:

9. exported as a tiff

10. Loaded into mtf mapper

11. I got a MTF50 score of 0.27

Click to expand...

Pretty good. Excellent lenses can get to about 0.3.

The figure is not a score. You can multiply the vertical resolution of the sensor by the 0.27 value, that gives you lp/PH value which is comparable between systems.

DavidMillier said:

What, if anything, that does this tell me?

Click to expand...

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

Thank you!

Presumably, that number can only represent the performance of the lens at the exact point in the field of view where the target sits i.e. it is not generic for the whole lens circle?

This suggests that to characterise the lens properly means setting up multiple targets at centre, mid-frame, edges, extreme corners and measuring them all separately?

DMillier said:

JimKasson said:

DMillier said:

JimKasson said:

DMillier said:

The proof of the pudding is in the eating as they say.

This image is an attempt to simulate a 36x36mm sq 60MP medium format image. It was made from 10 frames combined to produce a HDR pano in LR, saved as a dng. The dng was imported into ubuntu/darktable, edited and output as a 85% jpeg.

View attachment 340911358e8a452e8af177c7d7a2f098.jpg

Not the greatest weather for doing testing (threatening rain, overcast, dull). Colour is a bit odd as well on my screen. But the question is in terms of resolution, dynamic range, shadow noise etc, is this remotely medium format like or is this all a lot of effort for no gain?

Click to expand...

This is a credible image, albeit not one with a high subject DR -- it's actually the opposite. The lens is not particularly sharp, which served you well here in preventing aliasing in the tree branches. I'm guessing that in this cas, you could have done about as well by shooting full frame and upsizing in a good resampling program. Give that a try and see if the pano is enough better to make it worth your while.

Click to expand...

Jim commented that my lens isn't very sharp. This surprised me as it is a Contax G Carl Zeiss Planar 35mm f/2 which had a good reputation in its day. It's not supposedly quite as good as the 45mm, but still highly rated.

I decided to see whether it was possible to come to a more objective sense of how much my lens contributed to a lack of sharpness by testing it.

I followed Erik's suggested method here: https://www.dpreview.com/forums/post/65122260

I'm completely new to this so I have probably made loads of methodological blunders that completely invalidates the results but everyone has to start somewhere!

What I did:

1. Downloaded Erik's tilted square target.

2. Printed it out on A4 glossy paper

3. Taped it to a window

4. Set up my camera on tripod about 4 metres away

5. Photographed the target (which was very small in the frame) in the centre of the frame

6. Loaded the raw file into LR. Turned off sharpening and noise reduction

7. Cropped target to about a 100x200 px crop

8. Adjusted sliders until the slant edge crop had about the same tonal values as Erik's test crop (mine seemed noisier - does this matter?)

9. exported as a tiff

10. Loaded into mtf mapper

11. I got a MTF50 score of 0.27

Click to expand...

What are the dimensions of that? Cycles per pixel?

Click to expand...

According to Erik's tutorial post 50% MTF cy/px

DavidMillier said:

If so, what is the pixel pitch of your camera?

Click to expand...

Apparently it's 4.5 µm

Click to expand...

If there is no LR sharpening going on, you are looking at respectable sharpness.

DavidMillier said:

DavidMillier said:

It is better to look at the raw planes directly and avoid Lr entirely.

Click to expand...

I found this method described by Jack Hogan

http://www.strollswithmydog.com/how-to-get-mtf-curves-for-your-camera-and-lens/

Would this be a way to avoid LR?

Click to expand...

Yes.

DavidMillier said:

DavidMillier said:

DavidMillier said:

What, if anything, that does this tell me?

Click to expand...

Click to expand...

Click to expand...

That the lack of aliasing came from off-axis performance, or some other smoothing effect.

--

DMillier said:

Thank you!

Presumably, that number can only represent the performance of the lens at the exact point in the field of view where the target sits i.e. it is not generic for the whole lens circle?

Click to expand...

True.

DavidMillier said:

This suggests that to characterise the lens properly means setting up multiple targets at centre, mid-frame, edges, extreme corners and measuring them all separately?

Click to expand...

You can use one target for multiple framings. That's what I do.

DMillier said:

Thank you!

Presumably, that number can only represent the performance of the lens at the exact point in the field of view where the target sits i.e. it is not generic for the whole lens circle?

Click to expand...

Yes.

DavidMillier said:

This suggests that to characterise the lens properly means setting up multiple targets at centre, mid-frame, edges, extreme corners and measuring them all separately?

Click to expand...

MTF measurements are very sensitive to alignment, so measuring MTF across the field is no easy task.

You could use a set of patches, covering a larger area and measure each individually.

MTFMapper has tools to generate area targets that can measure MTF over the whole image, but you need a large printout, flat mounting and near perfect alignment, Doing that, MTFMapper can give accurate results.

You can also use a single target and move it to different points in the field of view. Best way is to refocus for each placement.

That gives info on performance, but leaves out field curvature.

Best regards

Erik

Erik Kaffehr said:

DMillier said:

Thank you!

Presumably, that number can only represent the performance of the lens at the exact point in the field of view where the target sits i.e. it is not generic for the whole lens circle?

Click to expand...

Yes.

Erik Kaffehr said:

This suggests that to characterise the lens properly means setting up multiple targets at centre, mid-frame, edges, extreme corners and measuring them all separately?

Click to expand...

MTF measurements are very sensitive to alignment, so measuring MTF across the field is no easy task.

You could use a set of patches, covering a larger area and measure each individually.

MTFMapper has tools to generate area targets that can measure MTF over the whole image, but you need a large printout, flat mounting and near perfect alignment, Doing that, MTFMapper can give accurate results.

You can also use a single target and move it to different points in the field of view. Best way is to refocus for each placement.

That gives info on performance, but leaves out field curvature.

Best regards

Erik

Click to expand...

How sensitive to errors is this process eg does a small mistake generate a correspondingly small incorrect result or is it really sensitive and the slightest mistake makes the whole exercise pointless?

I'm trying to judge whether you have to be as meticulous as Jim or whether you can get away with a rough and ready approach if you just want to get an idea of what your lenses are doing and aren't too fussy about absolute accuracy...

I'm primarily interested in stopped down performance at f/8 and f/11. I'm not expecting miracles from my 1970s budget lens collection! I've already weeded out the ones that provided visually blurred/smeared edges.

--
DPReview gallery: https://www.dpreview.com/galleries/0286305481
Website: http://www.whisperingcat.co.uk/ (2018 - website revived!)
Flickr: http://www.flickr.com/photos/davidmillier/ (very old!)

DMillier said:

Erik Kaffehr said:

DMillier said:

Thank you!

Presumably, that number can only represent the performance of the lens at the exact point in the field of view where the target sits i.e. it is not generic for the whole lens circle?

Click to expand...

Yes.

DavidMillier said:

This suggests that to characterise the lens properly means setting up multiple targets at centre, mid-frame, edges, extreme corners and measuring them all separately?

Click to expand...

MTF measurements are very sensitive to alignment, so measuring MTF across the field is no easy task.

You could use a set of patches, covering a larger area and measure each individually.

MTFMapper has tools to generate area targets that can measure MTF over the whole image, but you need a large printout, flat mounting and near perfect alignment, Doing that, MTFMapper can give accurate results.

You can also use a single target and move it to different points in the field of view. Best way is to refocus for each placement.

That gives info on performance, but leaves out field curvature.

Best regards

Erik

Click to expand...

How sensitive to errors is this process eg does a small mistake generate a correspondingly small incorrect result or is it really sensitive and the slightest mistake makes the whole exercise pointless?

I'm trying to judge whether you have to be as meticulous as Jim or whether you can get away with a rough and ready approach if you just want to get an idea of what your lenses are doing and aren't too fussy about absolute accuracy...

I'm primarily interested in stopped down performance at f/8 and f/11.

Click to expand...

The results are accurate, but would be affected by manipulation of the tone curve.

Focusing accuracy plays a major role:


As you can see on the two first groups, manually focusing the pretty sharp Sonnar 180/4 CFi on the Hasselblad yields considerable variation.

The f/5.6, f/8 and f/11 data for the Sonnar 180/4 CFi are best from focus bracketing with 40 exposures, similar to Jim Kasson's testing.

Best regards

Erik

Erik Kaffehr said:

DMillier said:

Erik Kaffehr said:

DMillier said:

Thank you!

Presumably, that number can only represent the performance of the lens at the exact point in the field of view where the target sits i.e. it is not generic for the whole lens circle?

Click to expand...

Yes.

Erik Kaffehr said:

This suggests that to characterise the lens properly means setting up multiple targets at centre, mid-frame, edges, extreme corners and measuring them all separately?

Click to expand...

MTF measurements are very sensitive to alignment, so measuring MTF across the field is no easy task.

You could use a set of patches, covering a larger area and measure each individually.

MTFMapper has tools to generate area targets that can measure MTF over the whole image, but you need a large printout, flat mounting and near perfect alignment, Doing that, MTFMapper can give accurate results.

You can also use a single target and move it to different points in the field of view. Best way is to refocus for each placement.

That gives info on performance, but leaves out field curvature.

Best regards

Erik

Click to expand...

How sensitive to errors is this process eg does a small mistake generate a correspondingly small incorrect result or is it really sensitive and the slightest mistake makes the whole exercise pointless?

I'm trying to judge whether you have to be as meticulous as Jim or whether you can get away with a rough and ready approach if you just want to get an idea of what your lenses are doing and aren't too fussy about absolute accuracy...

I'm primarily interested in stopped down performance at f/8 and f/11.

Click to expand...

The results are accurate, but would be affected by manipulation of the tone curve.

Focusing accuracy plays a major role:

https://www.dpreview.com/forums/post/62311443

As you can see on the two first groups, manually focusing the pretty sharp Sonnar 180/4 CFi on the Hasselblad yields considerable variation.

The f/5.6, f/8 and f/11 data for the Sonnar 180/4 CFi are best from focus bracketing with 40 exposures, similar to Jim Kasson's testing.

Best regards

Erik

Click to expand...

I tested using a 35mm lens focused with 10x magnification and peeking enabled. It seemed quite easy to focus but I've only taken one measurement, beginner's luck perhaps Long lenses must be more difficult to measure.

--
DPReview gallery: https://www.dpreview.com/galleries/0286305481
Website: http://www.whisperingcat.co.uk/ (2018 - website revived!)
Flickr: http://www.flickr.com/photos/davidmillier/ (very old!)

DMillier said:

Erik Kaffehr said:

DMillier said:

Erik Kaffehr said:

DMillier said:

Thank you!

Presumably, that number can only represent the performance of the lens at the exact point in the field of view where the target sits i.e. it is not generic for the whole lens circle?

Click to expand...

Yes.

DavidMillier said:

This suggests that to characterise the lens properly means setting up multiple targets at centre, mid-frame, edges, extreme corners and measuring them all separately?

Click to expand...

MTF measurements are very sensitive to alignment, so measuring MTF across the field is no easy task.

You could use a set of patches, covering a larger area and measure each individually.

MTFMapper has tools to generate area targets that can measure MTF over the whole image, but you need a large printout, flat mounting and near perfect alignment, Doing that, MTFMapper can give accurate results.

You can also use a single target and move it to different points in the field of view. Best way is to refocus for each placement.

That gives info on performance, but leaves out field curvature.

Best regards

Erik

Click to expand...

How sensitive to errors is this process eg does a small mistake generate a correspondingly small incorrect result or is it really sensitive and the slightest mistake makes the whole exercise pointless?

I'm trying to judge whether you have to be as meticulous as Jim or whether you can get away with a rough and ready approach if you just want to get an idea of what your lenses are doing and aren't too fussy about absolute accuracy...

I'm primarily interested in stopped down performance at f/8 and f/11.

Click to expand...

The results are accurate, but would be affected by manipulation of the tone curve.

Focusing accuracy plays a major role:

https://www.dpreview.com/forums/post/62311443

As you can see on the two first groups, manually focusing the pretty sharp Sonnar 180/4 CFi on the Hasselblad yields considerable variation.

The f/5.6, f/8 and f/11 data for the Sonnar 180/4 CFi are best from focus bracketing with 40 exposures, similar to Jim Kasson's testing.

Best regards

Erik

Click to expand...

I tested using a 35mm lens focused with 10x magnification and peeking enabled. It seemed quite easy to focus but I've only taken one measurement, beginner's luck perhaps Long lenses must be more difficult to measure.

Click to expand...

10X with peaking should be pretty accurate. Try preferably at shooting aperture, if possible.

Best regards

Erik

DMillier said:

I tested using a 35mm lens focused with 10x magnification and peeking enabled. It seemed quite easy to focus but I've only taken one measurement, beginner's luck perhaps Long lenses must be more difficult to measure.

Click to expand...

Make 16 images with independent focusing, and then compare the results. You may be surprised.

Jim

I did a 4 shot run, refocusing for each measurement.

I changed things around a bit as well.

I moved the target to a glass door so it wasn't backlit and I turned on every light I could find.

I used a 5 sec self timer and an IR remote release.

I put a black sock over the lens barrel to prevent any light leaks through the adaptor.

Results @f/8 no sharpening or noise reduction, but exposure, blacks and shadows sliders used to brighten the crop.

#1 0.25 cy/px

#2 0.16 cy/px

#3 0.11 cy/px

#4 0.17 cy/px

Erik said that 0.3cy/px represents an excellent result. My results seem too good to be true, every run beats that handily. I kind of expect any mistakes to make things worse, not better, so is there anything I could be doing that is inflating the performance?

here's #1:


Converted to png from tiff

Edit: Photodo have some optical bench scores for my lens

Weighted MTF 35 mm: f2 0.68 | f2.8 0.77 | f4 0.82 | f8 0.83 Average Weighted MTF 0.82

Weighted MTF 10 lp/mm 0.90

Weighted MTF 20 lp/mm 0.80

Weighted MTF 40 lp/mm 0.58

--
DPReview gallery: https://www.dpreview.com/galleries/0286305481
Website: http://www.whisperingcat.co.uk/ (2018 - website revived!)
Flickr: http://www.flickr.com/photos/davidmillier/ (very old!)

This time I'm testing my Miranda 24mm f/2.8. This is an OM mount budget Cosina lens (from the late 1970s?). I paid about £12 for this one on ebay.

Results at f/8

#1 0.12 cy/px

#2 Failed to measure

#3 0.11 cy/px

#4 0.12 cy/px

I'm suspicious of these numbers...

--
DPReview gallery: https://www.dpreview.com/galleries/0286305481
Website: http://www.whisperingcat.co.uk/ (2018 - website revived!)
Flickr: http://www.flickr.com/photos/davidmillier/ (very old!)

This is my 50mm f/1.4 Minolta MD

Shot @ f/8

Results

#1 0.16 cy/px

#2 0.21 cy/px

#3 0.19 cy/px

#4 0.14 cy/px

Contax G Carl Zeiss Sonnar 90mm f/2.8

Tested at f/8

#1 0.12 cy/px

#2 0.11 cy/px

#3 0.10 cy/px

#4 0.11 cy/px

Photodo optical bench scores

Weighted MTF 90 mm: f2.8 0.81 | f4 0.84 | f8 0.85

Average Weighted MTF 0.84

Weighted MTF 10 lp/mm 0.92

Weighted MTF 20 lp/mm 0.81

Weighted MTF 40 lp/mm 0.62

--
DPReview gallery: https://www.dpreview.com/galleries/0286305481
Website: http://www.whisperingcat.co.uk/ (2018 - website revived!)
Flickr: http://www.flickr.com/photos/davidmillier/ (very old!)

For comparison purposes, a modern lens

Shot at 18mm (36mm equiv) and f/4.5. These were autofocused as the lens doesn't have a MF ring.

#1 0.20 cy/px

#2 0.19 cy/px

#3 0.32 cy/px

#4 0.19 cy/px

DMillier said:

For comparison purposes, a modern lens

Shot at 18mm (36mm equiv) and f/4.5. These were autofocused as the lens doesn't have a MF ring.

#1 0.20 cy/px

#2 0.19 cy/px

#3 0.32 cy/px

#4 0.19 cy/px

Click to expand...

Hi,

1, 2 and 4 are close. Number 3 stands out.

Best regards

Erik

Here are some samples from my own testing...

The protocol:

My composite test target used, including Bart van der Wolf's test target, two sides of a 1$US bill and Norman Koren's MTF chart and high contrast resolution test target from HCam.de.

• Aligned using LensAlign
• Voigtlander 65/2 Apo lanthar at f/4
• Focusing using peaking in magnified live view
• Focus are, I don't recall, probably George Washington's head
• Exported from Lightroom with sharpening set to zero in gamma 1.0 color space.
• Evaluated by SFRMAT3 from my own Matlab script.
• Cameras used Sony A7rII, Sony A7rIV

What I see is:

Sony A7rII reaches 0.30 cy/px while Sony 7rIV reaches 0.27 cy/px at 50% MTF, both rounded to two decimals.

• Results very close up to 1000 cy/PH
• A7rII has 13% MTF at NYquist (blue vertical line) while A7rIV is 14% MTF at Nyquist (red line)
• Lightroom has weak halo on the A7rII

Assume now that I print at A2-size, which is my normal print size. Looking at 180 PPI on a 24" MP 2K monitor is in my experience a good approximation of that print size. Here are crops at that size from the image:


Images downsized to 42x62 cm at 180 PPI. Sharpened 100% at 0.7 radius in smart sharpen.

Printing large, my print shop recommend 200 PPI. An 80x120 print at 200 PPI would look like this:


Images resized to 80x120 cm at 200 PPI. Sharpened with FocusMagic, radius one pixel and amount 75%.

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