Is it my monitor or my technique ?

[buso23]

I have this image that I'm trying to add a drop shadow underneath and although I have blurred it and softened it, I can still see a sharp distinction where the white background meets the shadow.

Is it my monitor that is showing it or there's an error in my technique ?

I have tried blurring the line, painting with a white brush, etc. but no matter what I do, I can still see a sharp divide between the white background and the drop shadow.

Thanks, Ljupco

[John Wheeler]

Hi Ljupco

Neither I believe. I think this is an optical illusion. On my monitor it looked pretty good then I saw the edge lightness. It reminded me of another image that I have. The image below is a linear gradient from black to white hear the middle. It continues on at the same white level all the way to the right. For some people it will look like there is a brighter band right in the middle of the screen. Of course, there is always the possibility that the monitor is doing something funny. An easy check for that on my image is just to move a piece of paper over the left side and move it the the right blocking the gray. Also do it from the right side. When you get close to the middle, that bright edge disappears and it looks more uniform. Not sure if everyone sees it or not, I certainly do.

Click on Thumbnail
Screen shot 2011-12-09 at 8.16.42 PM.png

For another related hoot, check out this web site with other optical illusions.
http://www.scientificpsychic.com/graphics/

The brain-eye combination does some strange things

[buso23]

Hmmmm.... I get what you are saying John and after thinking for a while I guess might be right about being an illusion. But the thing that bothers me is why other not computer generated reflections/dropshadows do not show this ? For example the wheel in the same photo. I don't know if you can see it from the small picture, but the natural dropshadow does not show a sharp line, it's more blurred and realistic.

I'm away from the PC right now but when I get back home I'll upload a larger version of the wheel so that you see what I'm talking about.

Thanks and regards, Ljupco

[shift studio]

without addressing your second post, I think its very likely your display - some less-capable displays simply cannot display a smooth transition to pure white. the values that can't be displayed are dropped off creating a harsh line on your display.
I have two displays side by side, on one (a 2 year old NEC) your shadow is very smooth. I slide the browser window to the other display (a 7 year old Benq) and I see the line as you describe...

This can be frustrating, and as far as I know your only solution is a better display.
--shift studio.

[Flashtones]

Sometimes it's the display profile. You might try a new calibration/profile with different parameters (for instance, native white point rather than D65, as one example) and see if that helps.

(the shadow looks okay on my monitor, btw)

[John Wheeler]

Hmmmm

While the issue can certainly be the display quality or calibration where it causes banding or bright edges, I am not sure I am getting folks to believe that the eye-brain combination can be totally fooled by simple gradients (and such gradients often occur at the edges of shadows).

Let me provide one more example and for this case I challenge any of you to say that this is a monitor calibration issue.

Step 1 - Just click open the thumbnail and take a look at the image with a lighter gray on the left and a darker gray on the right. Carefully inspect it

CornSweetSFW.jpg

Step 2 - Open up the same two tone image and put a pen vertically right over the middle boundary where the two tones meet. The two tones are now gone and both sides have the same gray level. If you don't believe me, download it to Photoshop and check it out.


This image has all the same tone gray level except right in the middle where there is a little gradient going from left to right making it slightly brighter and another gradient right near the middle going from right to left that goes slightly darker. The effect is better known as the Cornsweet illusion.

The phenomena in my previous post is called a Mach Band Illusion.

These are not just for fun, it turns out that this is an issue that has to be dealt with in the Radiology field when trying to interpret X-Rays. The eye-brain combination will create whiter or darker areas just because of an adjacent gradient. Here is a link from a radiology site:
http://www.wikiradiography.com/page/Mach+bands+and+other+Optical+Illusions

Here is a second link that focuses just on Mach Bands while other links on the left go to other perceptual issues. Many do show up in regular photos:
http://perceptualstuff.org/machbandsbasic.html

So even with the best monitor and best calibration, that accuracy and certainty ends when the light enters your eyeball. All IMHO of course.

Of course these perceptual issues may not be what Ljupco is seeing for his image. Just wanted to raise the issue that such perceptual issues are real.

[Flashtones]

John, lets say for the moment that the issue is an optical illusion. Why is it that some monitors/profiles cause one to suffer the illusion and others don't? The answer probably gets at the cause of the problem, and I suspect its display gamma related.

[John Wheeler]

Originally Posted by Flashtones

John, lets say for the moment that the issue is an optical illusion. Why is it that some monitors/profiles cause one to suffer the illusion and others don't? The answer probably gets at the cause of the problem, and I suspect its display gamma related.

Hi Flashtone, good questions and like I indicated in my last post it very well may not be an illusion.

Originally Posted by John Wheeler

Of course these perceptual issues may not be what Ljupco is seeing for his image. Just wanted to raise the issue that such perceptual issues are real.

I actually could not see the illusion on either of my dual monitors until I looked a little more carefully (again this may not be Ljupco's issue and could just as easily be non linear LUTs in the graphics card).

As far as the differences between two adjacent monitors by the same observer is a good question. Yes, it is possible to have an illusion that shows on one and not the other side by side monitor. Here are the reasons and not all may apply to your situation.

1) Depends on the actual physical size on your monitor of the object (screen resolution and image magnification). Physically larger or smaller gradients make illusions go away.
2) Depends on the luminosity of the monitors and where the peak brightness is set (cd/m^2) Subtle differences can make it come and go
3) Differing angle of view specs of the monitors
4) Viewing angle of the monitor (tilt) which is tied to #3
5) Angle of view of you eye to the screen relative to the direction your head is facing (your eye-brain combination sees it different believe it or not)
6) Gamma or LUT values not identical creating still linear brightness yet different brightness on each screen.
7 Difference of white point (color) for each monitor makes a difference

I tested out each of the above (except #8) and could make the illusions come and go. The actual test to see if it is the issue (assuming you can see it to begin with) is easy to do. You just mask out the gradient physically on the monitor and see if the bright anomaly is still there.

I am not saying Ljupco's issue is an illusion. Certainly could be a calibration issue (I seen that as well on my own system). And you could certainly say this is "all in my head", and actually, you would be right

[Tony W]

I would say that it is an optical illusion (may well be worsened by how monitor is set up) and I would say that it may be improved by increasing the shadow density and lessening the blur.

IMO the shadow is far too light, even though the boxes are above the floor by a good margin they should be darker than they are - or the wheel shadow should be much lighter.

The wheel shadow actually looks like it was a cut, vertical flip and paste job and I think this could be improved which hopefully will give a little more realism to the scene

[Repairman]

I always add a bit of noise/grain to airbrushed shadows Buso. Try and match whatever grain is in the photo and it will look less 'painted'. As mentioned already the edge banding shows up on some monitors and not on others; might be interesting to output a decent print and take it from there.

[shift studio]

Interesting topic, and I've been in this situation myself. I believe John's mind-eye/ perceptual issues are real and are intruiging.

I want to point out however, regardless of the cause of the harsh transition on Buso's display ( or in Buso's eye-mind interoperation), IF the transition appears smooth on at least one device, then that is not an optical illusion or display issue, its real, that transition IS smooth

If I'm wrong, please let me know.
--Shift Studio.

[Tony W]

shift studio I would not presume to say that you are wrong about viewing device differences however it could also be argued that if one device the transition appears smooth then that device could be incorrect and the others correct, but how can you tell?

My gut feeling is that this is as John said an optical illusion (similar in a way to the illusion one gets when sharpening - but on a larger scale).

I still feel that one of the problems here is too light a shadow and too large a blur. If you look at my attachment and wait a few seconds you can see just how large the blur is (applied curves to highlight). It is also quite noticeable the edge effect of the blur that is not as noticeable on the normal image but IMO still there and causing issues. Of course this might just be artefacting due to image size restrictions

As the gif quality poor have posted another image to compare

[shift studio]

Originally Posted by Tony W

shift studio I would not presume to say that you are wrong about viewing device differences however it could also be argued that if one device the transition appears smooth then that device could be incorrect and the others correct, but how can you tell?

I don't have any documented proof, just in my experience - in instances like this, displays tend towards 'dropping' off the near whites creating an abrupt line and not the reverse. i.e. the don't take a harsh transition and smooth it out...

yes, the OP's illustrated shadow is not dark enough, it should be crisper, it should have grain/noise matching the photographic element...

--shift studio.

[John Wheeler]

Originally Posted by shift studio

IF the transition appears smooth on at least one device, then that is not an optical illusion or display issue, its real, that transition IS smooth

If I'm wrong, please let me know.
--Shift Studio.

That is a good question and I do not know the answer. For any image in question I would view it at 100% to make sure there are no anomalies and also at the approximate print size on the screen. If it doesn't look like a problem I wouldn't worry about it.

Your question about if it looks good on the screen (at least one screen) are you good to go. Not sure. That might work for gradients. The trick that I have seen to not have this issue is to have a smooth tail on the gradient. The eye is good at seeing edges which from a math standpoint is the rate of change of the gradient. If the gradient has a sudden change (e.g. black to white gradient right into a white field) then the illusions will be there. Using gradients set with the control at a high level of smoothness helps and also as others already suggested adding a blur seems to make a big difference.

However, if you are asking "can I make adjustments on my calibrated monitor and it looks good then I am good to go?" I would say most of the time yes. In some cases possibly not. Here is an example leveraged from the CornSweet Illusion to demonstrate. The top half of the image shown below is the same as what I had in a previous posted. Both the left side and the right side have a gray level of 212 through it does not appear that way. In the lower half, I added a Levels Adjustment Layer and masked it to adjust the left side. I shifted the white point down until I thought I had a visual grayscale match. This was when the white point was brought down to 226 or 10% down from the peak. Note that once I thought I had a visual match, the gray value was 188 of the left and 212 on the right.

So even when visually matching with your eyes, the actual luminosity is different. I wonder if this matters though. Would the average person see the same thing anyways? This may make a difference in the Radiology field in making a diagnosis, yet in the photography field it may be just a curiosity with no real consequence.

Screen shot 2011-12-11 at 5.15.23 PM.png

I am attaching a PNG file of the original CornSweet image if you want to try this yourself.

CornSweetSFW.png

Continued comments welcome of course.