Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Photoshop display anomaly  (Read 1005 times)
Jim Kasson
Sr. Member
****
Offline Offline

Posts: 744


WWW
« on: March 27, 2014, 11:36:48 AM »
ReplyReply

Theres an old technique in chemical photography for making etching-like images. You sandwich a negative and a positive of the same image together just slightly out of register, and print the whole thing on lith film, then you make your print from the lith.

I thought Id see what happened to an image of mine with that treatment. With my darkroom turned into a storage area, I turned to Photoshop. I converted the image to B&W, duplicated the result as another layer, and set the blend mode to difference. I zoomed in, and moved the top image a tiny bit. Then I added a Levels layer on top and moved the white point way down. That got fiddly, so I added another Levels layer and did the same thing. Then I zoomed out to see what the whole image looked like.

Blackness.

I zoomed in. The image came back. I zoomed out again. Nothing but black. I wrote out the file, brought it into Lightroom, and it looked fine. Closing Ps and reopening the image showed approximately the right thing, but much darker than the Lr presentation, and darker than a JPEG file exported from Lr.

Photoshop version 14.2.1 x64, if anyone wants to try and duplicate my experience. File size is 6764x4407. Display res is 2560x1600. Ps window was maximized.

A small JPEG of the image is here.

Jim
« Last Edit: March 27, 2014, 11:51:29 AM by Jim Kasson » Logged

Redcrown
Full Member
***
Offline Offline

Posts: 126


« Reply #1 on: March 27, 2014, 12:50:25 PM »
ReplyReply

I see the "anomaly" you describe often when I compare two pixel layers in Photoshop using Difference mode. Instead of using Levels to accentuate the difference I use a Threshold layer and adjust the Threshold Level thru low values (1 to 10) to more easily see the difference.

The number of white pixels that display will vary a lot depending on the image magnification. At 100% you see them all. As the magnification is reduced, the number of white "difference" pixels reduces. Eventually, at low magnification the image goes all black on the screen.

Consider that when any software displays an image at reduced magnification it has to temporairly resize the image. Resizing methods will vary depending on the software. Photoshop uses a method that favors speed over accuracy. Your Photoshop performance settings (cache levels and cache tile size) and your GPU are variables. "Dithering" is another variable that can effect the result. Thus two different users might see different results on the same image.

I don't know how Lightroom resizes for display, but I suspect it's different than Photoshop. Thus the anomaly might appear at a different degree in Lightroom vs. Photoshop.

When an image contains very small white pixels against a black background, it should be expected that some of them will disappear when the image is reduced in size.
Logged
TonyW
Full Member
***
Offline Offline

Posts: 120


« Reply #2 on: March 27, 2014, 12:54:46 PM »
ReplyReply

I have observed the same effect in PS using difference mode.  The complete loss of the image seems to depend on the degree of reduction and occurs around 30% and less. I assume that this is because PS cannot correctly interpret the small changes involved in a layered image

However, if you flatten your image layers (or stamp layers) then you should not see any changes while zooming in or out

« Last Edit: March 27, 2014, 12:57:29 PM by TonyW » Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3461


« Reply #3 on: March 27, 2014, 01:22:49 PM »
ReplyReply

I thought Id see what happened to an image of mine with that treatment. With my darkroom turned into a storage area, I turned to Photoshop. I converted the image to B&W, duplicated the result as another layer, and set the blend mode to difference. I zoomed in, and moved the top image a tiny bit. Then I added a Levels layer on top and moved the white point way down. That got fiddly, so I added another Levels layer and did the same thing. Then I zoomed out to see what the whole image looked like.

Blackness.

I zoomed in. The image came back. I zoomed out again. Nothing but black. I wrote out the file, brought it into Lightroom, and it looked fine. Closing Ps and reopening the image showed approximately the right thing, but much darker than the Lr presentation, and darker than a JPEG file exported from Lr.

Hi Jim,

I don't have the time to do an exhaustive test right now, but what gamma setting do you use for Layer blending in the preferences setup? I've got mine set to 1.00 by default.

Cheers,
Bart
Logged
Jim Kasson
Sr. Member
****
Offline Offline

Posts: 744


WWW
« Reply #4 on: March 27, 2014, 01:33:22 PM »
ReplyReply

...what gamma setting do you use for Layer blending in the preferences setup? I've got mine set to 1.00 by default.

Bart,

I've got mine set the same way.

Jim
Logged

Jim Kasson
Sr. Member
****
Offline Offline

Posts: 744


WWW
« Reply #5 on: March 27, 2014, 01:36:08 PM »
ReplyReply

However, if you flatten your image layers (or stamp layers) then you should not see any changes while zooming in or out.

Thanks. Also, flattening the image results in a displayed image that approximates that of Lr.

Jim
Logged

Jim Kasson
Sr. Member
****
Offline Offline

Posts: 744


WWW
« Reply #6 on: March 27, 2014, 01:44:19 PM »
ReplyReply

Your Photoshop performance settings (cache levels and cache tile size) and your GPU are variables.

Thanks. Changing the cache levels and turning GPU acceleration on and off don't change what I'm seeing.

Jim
« Last Edit: March 27, 2014, 01:50:25 PM by Jim Kasson » Logged

Jim Kasson
Sr. Member
****
Offline Offline

Posts: 744


WWW
« Reply #7 on: March 27, 2014, 01:49:33 PM »
ReplyReply

The number of white pixels that display will vary a lot depending on the image magnification. At 100% you see them all. As the magnification is reduced, the number of white "difference" pixels reduces. Eventually, at low magnification the image goes all black on the screen.

You are right. This is true even when the number of white and black pixels is approximately the same. There is apparently a bias in favor of showing black pixels over white ones at low magnification.

Jim
Logged

Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad