Available Knowhow
Enables desired elements to be snapped into place
Categories |
Computer Science and Engineering, Imaging/Computer Graphics |
Development Stage |
Development of method completed |
Knowhow |
Available Knowhow |
Highlights
-
Digital image editing often involves image composition where images from different shots are combined into a single output image that looks natural and realistic. Current tools require the image backgrounds to be aligned and for the inserted element to approximate to the place into which it is going.
Our Innovation
-
New method for composing images enables the simple insertion of different objects into an image even if the backgrounds cannot be aligned. A video describing the results may be viewed at:
Key Features
-
Objects in the different images are marked and dragged into a new location in the target canvas. Backgrounds around the dragged objects, as well as the final locations of the objects themselves, are automatically computed to give a seamless composition.
-
If the selected object is not dragged to a desired place, it will automatically snap into a suitable location.
Figure 1: Snap Composition. (a-b): Input images with regions for inclusion in composition marked in green and regions to be ignored marked in red. (c) Initial canvas with selected children in their original locations. (d) Snap Composition results. The children from (a) have been automatically spaced to allow the child from (b) to snap in between them and the background has been rearranged to give a seamless composition.
Figure 2: Comparison of example in Fig. 1 with Adobe Photomerge Group Shot. This emphasizes the importance of individual displacement of objects, which is not one of the features supported by Photomerge. (a-b) Photomerge results obtained using different ordering of input images. (c) Snap Composition results.