La sua efficacia e' assolutamente fuori discussione, ovviamente come tutte le cose segue i gusti personali.
Se qualcuno vuole questa e' la procedura alla base dell'algoritmo del Darbee, da riprodurre in Photoshop:
Using Adobe Photoshop™ to Implement the Darbee Transform
The Darbee Transform basically involves blurring and subtracting one image of a stereo pair from the other image. Because image subtraction is not a commonly available image-processing option, a negative (inverted) image is instead added. This simple procedure is easily accomplished in Adobe Photoshop™ by following the steps below. Movies can also be processed similarly frame-by-frame using Adobe Premiere™ or similar programs.
1. One first opens the image file desired to be processed using the DarbeeVision (DVn) procedure. One can use CtrI-O to do this. We will assume that the image file contains a stereo pair of images arranged side-by-side for “cross-eyed viewing,” i.e. with the right-eye image on the left and the left-eye image on the right.
2. Using the Rectangular Marquee Tool, outline the image on the right.
3. Cut out the outlined image using Ctrl-X.
4. Create a new layer either by using the Layer menu or by typing Alt-L Alt-W Alt-L. A new layer will appear on the Layer menu, but one will not see anything else appear on the screen.
5. Paste the cut image onto the new layer using Ctrl-V.
6. Invert the pasted image (make it a negative image) using Ctrl-I, or by using the Image>Adjust>Invert menu command.
7. Set the opacity of the inverted image to 50% using the Opacity control in the Layer menu.
8. Select the Move Tool and slide the inverted image to the left over the other image. One can use the arrow keys to fine-tune the placement of the image so that the object of most interest has the best convergence.
9. Using the Filter>Blur>Gaussian Blur . . . menu command, call up the Gaussian Blur menu. Set the blur radius to a number of pixels that is approximately one-hundredth the width of the image in pixels. One can experiment with different radii to achieve a pleasing “glow” around features in the image. Images that have a lot of disparity (areas that are misconverged) generally will need a larger blur radius.
10. Use the Crop Tool to select the borders of the overlaid image. Sometimes one will have to crop part of the borders of the image if it was necessary to misregister the two images in order to achieve the convergence wanted. One also might want to crop borders where the Gaussian blur shows up due to boundary conditions.
11. Press the Enter key to crop the image.
12. Set the Opacity to 25%.
13. If it is desired to save the original right image, turn off Layer 1 by clicking on its eye icon.
14. One can now save the original unaltered image using Ctrl-Shift-S. Rename the image with a “_R” suffix to show that it is the right image of the pair, and change its format to jpg using the dropdown format menu. One can accept the default jpg options when the menu appears.
15. Now turn Layer 1 back on by clicking on its eye icon.
16. Flatten the image down to one layer (the background) by using Alt-L Alt-F, or by using the Layer>Flatten Image menu command.
17. Increase the image contrast by 50% using Alt-I Alt-A Alt-C or by using the Image>Adjust>Brightness/Contrast menu command. Leave the brightness at 0.
18. One can now save the processed image using Ctrl-Shift-S. One can add the suffix “_O25_R08_C50” to designate it as using an Opacity of 25%, a Gaussian blur radius of 8 and a final contrast increase of 50%. One can accept the default jpg options when the menu appears.
19. The procedure is now complete. The image has been enhanced using the DarbeeVision algorithm. Three-dimensional information has been added to a two-dimensional image in such a way that objectionable double-image artifacts do not appear. There is also a contrast-stretching effect that makes the image appear more vibrant, along with an image-sharpening effect that makes the image appear clearer.
One can experiment with other opacity values for adding the blurred-inverted image in Step 12. Higher values add more of the blurred-inverted image. If one changes the opacity in Step 12, one will have to compensate by varying the final contrast of the flattened image in Step 17. One can also experiment with varying the brightness in Step 18. When one does such experiments, it is useful to compare the results to the original right image that was saved in Step 14. Simply open the original image using Ctrl-O and place it on the screen next to the DVn image.