Geometric correction via image warping is our process of digitally manipulating the image to provide the smoothest possible transformation between projected images. Our advanced algorithms geometrically align pixels from adjacent projectors while minimizing distortions.
A warp mesh is an "instruction" describing how to manipulate pixels to achieve a desired geometric result.
In the case of EasyBlend the warp mesh instructions are the output of our calibration process. These instruction must be "executed" on a piece of hardware as describe below.
There are two approaches to applying warps to images:
1) External Image Processors
2) Warping on the Graphics Processing Unit (GPU)
External Image Processors
External image processors are high performance scalers which deliver geometric warping capability. These devices, which range in size from 1U to 4U, support a wide range of input signals and high quality scaling and de-interlacing. Generally speaking these boxes offer a great degree of flexibility because they meet the extreme warping requirements of domes and toroid screens.
Scalable has invested significant resources over the past several years to develop elegant interface to several of the leading boxes on the market. The Image AnyPlace is one such example.
Warping on the Graphics Processing Unit (GPU)
When OpenGL or DirectX applications have been modified with our SDK the warping instructions or "warp mesh" are executed by the GPU.
Precise Geometric Accuracy
When warping images, video or computer generated as in a flight simulator, the objective is to minimize the geometric distortion experienced as the screen transitions from projector to projector. Lack of uniformity in a grid pattern is a classic example of a poorly executed geometric warp. In military environments very precise measurements are taken with a laser emitting theodolite to ensure precision in the geometric warp.
EasyBlend's geometric accuracy has been tested with a theodolite on a dome section and meet the required specifications for accuracy.