Parallax or 2.5D is a technique which consists in giving a 3D like effect to a painting, distinguishing various visual elements into various layers in this present case.
Within a 3D environment, camera effects give an impression of depth as some elements are animated at a different pace than others, as if you moved your eyes and noticed a tree in the foreground moving faster whereas the background sky is scrolling slower.
Because the various elements are placed at different depths, when the camera moves, a 3D animation-like effect is perceived.
Not all paintings or images are appropriate for parallax/2.5D. This may be due to the fact that the image’s elements are so closely interrelated (Bacon) or that creating different layers for different visual elements in a painting with no perspective (Servranckx) could be considered as altering the painting, hence requiring approval from the artist or her/his rights holders.
Here are examples of artworks which are appropriate and not:
If producing an HD (1920 x 1080 pixels) augmentation using the 2.5D technique, the original image should be 25% to 50% more detailed. If producing a UHD or 4K (3840 x 2160 pixels) augmentation, the original image should again be 25% to 50% more detailed.
As an example, a 4K image requiring a x2 close up in its middle would require an original digitization of 7680 x 4320 pixels.
Close up levels, if exaggerated, might end up producing animations where pixels are blurred, as the source will be scattered among various pixels, with insufficient resolution.
If a certain element of a painting needs more focus, the original digitization size should have the capacity of that given focus, depending on the close-up level desired.
In other words, when the camera zooms in, the pixels get bigger and some blur happens if the resolution of the original image is not big enough. The original image must have a resolution big enough so that at the maximum zoom level of the movie, the pixels of the movie are still bigger than the ones of the original image. For an HD movie of 1920×1080 and a zoom factor of x2, the original image must be at least 3840×2160.
In our case, production and broadcasts were done in portrait mode, which is the normal size of a modern screen. Briefs augmenting artworks should be indicated with this format in mind.
It is also possible to work in portrait mode with this format.
The image sources must be in RGB and not CMYK. If in the later, which is used for printing, the image should be transformed into RGB.
Creating the layers
The first step for 2.5D production is to separate the artwork in distinctive layers. This is done in an image manipulation program. The objectives of the augmentation should remain clear. It is not possible to separate all elements. Produced elements should be logical so as to represent their natural depth. For instance, shadows should be associated with their main element. There can be one or eventually two elements within a group (i.e. statue + shadow composed of statue and shadow layers).
WallMuse Production Services uses Adobe Photoshop, which has a number of interesting tools.
The process consists in separating the foreground elements individually into a separate layer. It is a detailed exercise and “intelligent scissors” may be helpful, although their precision must be adequate.
All layers behind a given layer must be completed so that when an element in front is visually moved by the camera, it does not film empty parts. In a simple example with a statue on a red soil with surrounding buildings and sky, the camera movement will perceive at moments parts that do not exist. We used a Photoshop “Patch Tool” to add new parts, based on other surrounding existing parts.
This is a trial and error approach. Generally, 5% of the inner parts of the following layer needs to be patched. It is a very sensitive exercise, so we must improve or extend some “patches” as and when needed.
Avoid altering the original painting when motion making
Animating a painting can bring it to light and make it more dynamic, but this must be in line with the original artwork, unless it is a choice made by the artist him or herself.
WallMuse will integrate these into its Player directly for SHAREX members as a service.
The alteration of the image achieved thanks to the layers and the camera can be checked against the original at various stages of the animation process. Both the beginning and the end of the animation will concur with the original image, so that no major differences can be noted during the camera movement.
As developed but not tested yet, we would assist first productions.