Hi, rehess, In another thread you asked:
rehess wrote:
What characteristics are you looking for in an image - what makes it a good candidate for 3D conversion?
This is a chapter to itself in the book I am writing on 2D to 3D conversion.
Basically, the same characteristics Ilook for are those that usually make one say: âÂÂSure wish I could see THAT in 3Dâ
They include:
good foreground object/detail
a sense of depth, even in 2D, that âÂÂpulls you inâ
linear perspective
receding perspective lines...straight or winding road or river...
distant detail that you woulÂÂd like to see in contour with depth separation, rather than the usual dimensionless Ââpainted stage backdropâ due to fall-off of the effect of horizontal parallax somewhere in the middle ground.
A dramatic sky above interesting terrain with a relatively clean horizon
internal framing
An image with an obvious vanishing point within ...or close to....or at the edge ...or just beyond the edge of the image.
In close-ups and macros:
obvious different levels of detail.
No image possesses all these features, but the more there are, the better the chances for a good 3D conversion.
And I obviously havenâÂÂt figured out all the indicators of success, because I sometimes receive an image for conversion that I am convinced lacks any promise, but every once in a while what I consider to be an unlikely or hopeless candidate can turn out to provide a stunning 3D conversion. Those are ones I marvel over and study most.
The two 3D conversions illustrated below (the two I most recently posted and which my solitary, frustratedly angry critic claims reveal no 3D effects...?) represent two extremes of degrees of likelihood to be successful candidates for 3D conversion., but both worked well.
The âÂÂRed Chairsâ is an example of uncomplicated âÂÂsweepâ depth recession from the bottom (the plane of closest perception) to the top (infinity in the sky). It includes distant, background hills across the water that are crying out for revelation of the range of their depths and of their contours unrevealed in the original 2D image (the left eye image). A basic transformation of the original rectangle that exaggeratedly maximizes the important monocular cues at the bottom, while minimizing them at the top fills the bill of optimal vertical spatial disparity between the base (left eye) image and the disparate ( right eye) image that allows our cortical visual center to provide our consciousness with the desired 3D effect, revealing depths and contours all the way to the horizon.
The âÂÂCyclamenâ image, submitted by a photographer well savvy in viewing stereo pairs, was a more complex challenge. The saving graces in the offered image are the obvious multiple depths within the low foliage and the beautifully and gently rendered white blossoms with contours of practically each petal available to be accentuated against the stark, dead black background. Deciding on the optimal direction of depth recession was aided by using the direction toward which the lighting was directed (brighter toward darker) as revealed by the brighter, slightly higher green foliage on the right and the darker, lower foliage on the left. So, of the forty-odd specific transformations that I have devised as my basic tool kit (each of which can be imposed in an infinite range of strengths of action), one that directs depth recession to the lower left at a lower strength was chosen to impose the necessary exaggerated direction of depth recession. And the result, I feel, is commensurate with the beautifully gentle presentation of this floral arrangement. And, of course, when the base imageâÂÂs original photographer, one well capable in stereo image viewing, explicitly states that he easily saw my intended 3D effect, that is also rewarding!
ThatâÂÂs a pretty good summary of the sorts of characteristics and features I look for in candidates for 2D-to-3D conversion.
Thanks, rehess, for the good question.
Note that these images are not to be downloaded or modified for any purpose without my permission tand permission of the original poster’s.
Dave