Les Dittert
Well-known member
Here is a simple, basic way to evaluate an image to see if you are hitting a certain resolution.
The method relies on the principal that when you resize an image to be smaller, you are lowering it's resolution, in essence 'crushing out the details' by making it smaller.
1> Load the image in Photoshop. Lets say it is 4K ( 4096 ) for this example.
2> Make a duplicate of it.
3> Apply image resize to one of them, lets say to 3000 across.
4> Apply resize again to the now 3000 across image, but this time back to the original size , in this case 4096 across.
5> Select all on the image and paste over the other untouched image, the duplicate you made earlier.
6> change the layer properties to be 'difference' instead of a normal paste. So now the image is mostly black because it is showing you where the two images are actually different. Flat and non detailed areas will be blackish.
7> Flatten the image to bake it into a simple non layered image.
8> Use adjust levels to brighten it up a bit. You may see some differences at this point, in the edge areas. This is because it is the edges that 'go away' when you resize down.
How to interpret the results:
If you see no clear edging in this 'difference' image, then you sample did not have actual 4K resolved well.
If you see distinct edging, then you have the possibility of some real detail. The reason that I say 'possibility' is that if some big sharpening was on the incoming image it will show false edges that are representing fake edge detail that have little to do with what was real in front of the camera.
So if you already saw real edge detail when you sized your 4K down to 3K , then you have higher than 3K resolution.
If you resized all the way down to 2K , and still saw no real looking edger, well, you have lower than a real 2K image.
Again, the idea here is that if there were no details in the image to start with, you have nothing to loose if you resize it down a bunch!
I recommend only doing this test with no sharpening or very mild sharpened images so you are not fooled by fake edge data.
We all know we can always add some 'Zap' with sharpening, but that treads in the fake detail category very quickly. Looks OK if you step back a few feet from the monitor
This is not an offical MTF/line pair per millimeter/Imatest type of test, but still effective to see where you are, roughly.
I hope this method is useful to some of you.
The method relies on the principal that when you resize an image to be smaller, you are lowering it's resolution, in essence 'crushing out the details' by making it smaller.
1> Load the image in Photoshop. Lets say it is 4K ( 4096 ) for this example.
2> Make a duplicate of it.
3> Apply image resize to one of them, lets say to 3000 across.
4> Apply resize again to the now 3000 across image, but this time back to the original size , in this case 4096 across.
5> Select all on the image and paste over the other untouched image, the duplicate you made earlier.
6> change the layer properties to be 'difference' instead of a normal paste. So now the image is mostly black because it is showing you where the two images are actually different. Flat and non detailed areas will be blackish.
7> Flatten the image to bake it into a simple non layered image.
8> Use adjust levels to brighten it up a bit. You may see some differences at this point, in the edge areas. This is because it is the edges that 'go away' when you resize down.
How to interpret the results:
If you see no clear edging in this 'difference' image, then you sample did not have actual 4K resolved well.
If you see distinct edging, then you have the possibility of some real detail. The reason that I say 'possibility' is that if some big sharpening was on the incoming image it will show false edges that are representing fake edge detail that have little to do with what was real in front of the camera.
So if you already saw real edge detail when you sized your 4K down to 3K , then you have higher than 3K resolution.
If you resized all the way down to 2K , and still saw no real looking edger, well, you have lower than a real 2K image.
Again, the idea here is that if there were no details in the image to start with, you have nothing to loose if you resize it down a bunch!
I recommend only doing this test with no sharpening or very mild sharpened images so you are not fooled by fake edge data.
We all know we can always add some 'Zap' with sharpening, but that treads in the fake detail category very quickly. Looks OK if you step back a few feet from the monitor
This is not an offical MTF/line pair per millimeter/Imatest type of test, but still effective to see where you are, roughly.
I hope this method is useful to some of you.
