We audiophiles are not the only hobbyists with issues......

[accwai]

On 4/26/2021 at 5:51 AM, yamamoto2002 said:

I use ColorChecker Classic instead of brick wall and Matlab Image Processing Toolbox measureColor() and imapplymatrix() to correct color appearance of photos, when it is necessary.

 

The freely available X-Rite ColorChecker Camera Calibration is supposed to be able to generate DNG/ICC camera profiles from many ColorChecker variants. Wouldn't that do? Back in the DNG Profile Manager days, one can give it two calibration photos at different light source color temperature and it's said that Lightroom can then interpolate across the whole input image color temperature range based on the resulting DNG profile. The same should be in the current generation.

 

Quote

To get lens distortion parameters and undistort photos, I take multiple images of DIY A4 size Checkerboard and Single Camera Calibrator app of Matlab Computer Vision Toolbox. There is OpenCV alternative and it is free.

 

Wouldn't lens distortion depend on focus distance? PTLens recommends at least 25ft from a building facade with perfect straight lines. And for zooms, 10-12 images with denser distribution on the short side where distortion characteristic changes more quickly.

 

By the way, any before/after demo photos of your process?

[accwai]

7 hours ago, bluesman said:

So no one has a “brick wall test” for SQ???

 

How about Radiohead's Packt Like Sardines In a Crushd Tin Box? By the way, how's the brick wall test below? Is it acceptable by stringent standard of the internet experts?

 

[accwai]

On 4/27/2021 at 5:12 PM, yamamoto2002 said:

Matlab solution works well for me. I think X-Rite ColorChecker app is not easy to use, ColorChecker chart detection function sometimes does not work for no reason.😀

 

Looks like the X-Rite app is quite finicky about data in the patches. And when it runs into problems, it just tells you something generic like capture size not big enough. The chart function in the Adobe DNG Profile Editor has much more specific error message, down to which patch is causing problem. So running the captures through that could provide some insight. How easy/painful the process is probably depends on the color engine in the specific camera. The before/after DNG profiling below is from a phone:

 

 

 

DNG file from the phone is pretty much useless without camera profiling. But the amount of twisting the profile does is quite significant. So even with it, strange behavior could be lurking in the process somewhere. We will see...

[accwai]

On 5/26/2021 at 6:07 AM, NoNaim said:

There are other more stringent tests done but when buying a £1k-£4k lens you want it to be right.

 

What kind of stringent tests need to be done on those lenses? Actually I'm really curious what kind of lens could be worth that much. Any example of the kind of images those lenses would provide in real use?

[accwai]

22 hours ago, NoNaim said:

[...] Zeiss & Sony GM lenses. Most of these vary from about £1k - £12k for the longer lenses. [...] Thankfully most lenses I need normally top out at the £4K mark.

 

I'm quite curious the kind of image that needs £2.5k-£4K lens to produce. Any examples to share?

[accwai]

On 5/29/2021 at 7:06 AM, yamamoto2002 said:

I'm also curious, large aperture prime lenses always disappoint me for poor corner sharpness even on F5.6 ～ F8.

For example, my Nikon AF-S 24mm F1.8 is better than AF-S 24mm F1.4

 

24/1.4 is a strange case. On the Canon side for example, the 24/1.4 has better construction quality than the 24/2.8. So the max aperture got mixed into product tiering consideration. But once one has decided on the bigger one, how would the f/1.4 be used for real? Most of the 24mm f/1.4 images floating around on the net seem rather contrived. Actually it's already bad enough to find good use case for 35mm @ f/1.4, the significantly wider 24mm would be even more difficult. Alternatively, architecture lenses are specifically designed to have uniform resolution across the field and minimal rectilinear distortion. The smaller max aperture could well be a worthwhile tradeoff for some:

 

 

Quote

And wide aperture of lenses cannot be used anyway.  It is waste of sensor resolution

 

Actually fast portrait lens often has a specific look that makes its own kind of sense:

 

 

Probably less than 10% of the area above is actually sharp. In this case, the sensor resolution is used to describe how blazing sharpness melts away. So it's not like resolution is completely wasted. Similar situation with going longer:

 

 

Only part of the head is absolutely sharp, but the image as a whole doesn't seem questionable. Similar for going even longer:

 

 

And believe it or now, the place where one really want to keep aperture big is high macro:

 

 

If one tries get to more depth by stopping down, after certain point everything turns into mush real fast due to diffraction...

[accwai]

On 6/1/2021 at 2:28 PM, NOMBEDES said:

The only wall on this site should be by PINK FLOYD

 

Would the illustrious members of AS be so kind as to accept this wall then?