Siesta
by Phostructor 2020-08-27 14:41
Site specific assemblage: I found the metal object and thought it was sculptural, so I put it on the boulder. iPhone cha-cha; slightly 'toed in'. Works well as parallel. I wonder if there is a way to improve the anaglyph rendition?
Comments
JackDesBwa 2020-08-31 15:17
Okay. I barely saw the rivalries you talked about on my screen, but I remembered that its gamut is large and it was calibrated some years ago. Apparently the aging shift of colors is less important than I imagined.
I moved your image in anaglyph on my TV which has no color profile, and they are very visible on it. It is clearly a crosstalk due to a poor separation of the two images (because of slightly wrong pixels' colors), and it looks so much like when I am in an angle where the light is badly polarized in 3D polarized mode (same crosstalk phenomenon) that I moved naturally to the sweet spot of the TV (obviously without gain in this case).
People who work with anaglyph developed strategies to minimize ghosting when the colors are poorly separated, but I do not know them much. Here I notice that the rivalries are very visible (on screens with color deviance) in the zones of high contrast and high parallax, which by the way are also sensible zones with some other viewing methods (very different contents on the left and right eyes are the most visible zones when one side leaks to the other)
About keystone correction, the tool you used to align your image should do it under the hood (SPM does it for example), otherwise I doubt you could achieve something as good, especially with cha-cha taking method.
Phostructor 2020-08-31 13:21
When I view the anaglyph I see some rivalry in the shadow area of the large bolt on the right, and also the edge of the shadow on the far left.
Maybe the convergence was not too strong, as I did not correct any keystone effect....
Thanks for your comments. Merci.
JackDesBwa 2020-08-28 15:14
What kind of improvement do you have in mind for the anaglyph version?
By the way, since you corrected the keystone created by the convergence pretty well, there is theoretically [and verified experimentally on my side] no difference with a strictly parallel version with the same framing.