A novel method for hidden surface removal is proposed in full parallax CGHs. This method is an improvement of the conventional silhouette method and calculates the object field faster than the conventional method in cases of creation of large-scale CGHs. The conventional method requires the same number of operations of total field propagation with the number of objects, whereas this method shields light based on the Babinet's principle and partial field propagation and does not use total field propagation at all. Therefore, in creation of large-scale CGHs that needs segmented frame buffers, the proposed method calculates the object field of occluded scenes much faster than the conventional method.
Techniques for numerical simulation of CGH-reconstruction have been reported in some literatures. However, the reported image is not the retinal image formed by lens of human eyes, and it is not presented to confirm the influence of non-diffraction light and the conjugate image. Recently, we proposed a novel technique for simulated reconstruction. This new method simulates the retinal image of reconstruction of CGHs by using methods of the shifted Fresnel diffraction and rotational transformation. We report the influence in reconstruction in cases that the illumination wave used in reconstruction is different from the reference wave used in CGH calculation.
This paper presents a new method for calculating Computer Generated Hologram (CGH) for 3-dimentional (3D) display. A "Ray-Sampling (RS) plane" is defined near by the object, and the wavefront at the RS plane is calculated from the light rays that can be obtained from multi-view images. Then the Fresnel diffraction from the RS plane to the CGH plane is calculated. Since the CGH is calculated from the light-ray information, the image data generated by artificial computer graphics or captured by camera array can be used. It is possible to reproduce the angular reflection properties which is important for highly realistic 3D image. In the experiment, high-resolution 3D image with gloss appearance is simulated with using multi-view images data generated by commercial rendering software.
In this paper, we have investigated the real image reconstruction from computer-generated hologram (CGH). The reconstructed real image is excellent as a display to give a strong three-dimensional appearance to an observer. However, it is difficult to get enough diffraction angles so that high space resolution is demanded for the real image reconstruction of the CGH. Therefore the reconstructed image usually does not have enough viewing area and size. In this report, we get real image that has wide viewing area which is using conjugate source as reconstruct source of the CGH. However this method doesn't give an accurate reconstructed image from the CGH. Therefore, we changed object data to use for a calculation to suitable form. We made CGH with these data and output a high-resolution hologram with a fringe printer.
We have already proposed the holographic display technique that sequentially generates a number of elementary holograms in order to enlarge the image size and the viewing zone angle. The elementary holograms displayed by a high-speed spatial light modulator are scanned horizontally by a galvano mirror. In this paper, a calculation method of the elementary holograms is described. Under the actual situations, the inconsistency in reconstructed images of the elementary holograms causes blurs in reconstructed images and the image quality is deteriorated. Horizontal positions and pixel pitches of the elementary holograms are measured in advance, and are used for the calculation of elementary holograms. The distance between a horizontal light focusing point to a screen is experimentally determined. Consequently, image blurs are reduced and the image quality is improved.
In this paper, we report the performance and scalability of fast generation of Fresnel computer-generated-hologram by use of a GPU cluster and the wavefront-recording method. The wavefronf-recording method is two-steps algorithm, in this paper, we implemented the second-step (diffraction calculation) onto GPU. The GPU cluster consists of three PCs and two GPU boards per the node.
When shooting a hologram, it requires special optical equipments and a darkroom. If the darkroom is not large enough, it is not easy for many people to enter in the dark room. In addition, it is not easy to handle special and expensive optical equipments for beginners. When teaching holography in a class room, it is not convenient to bring these equipments. Therefore, we use Augmented Reality technology in this study to build a virtual optical system on the desk. The image of the real scene is acquired with a video camera. Then virtual optical elements are superposed through markers corresponding to each optical element. The user can change the position of the virtual optical element to build correct optical setup without the dark room and any special optical elements. With this system, people can simulate the assembling procedure of the optical system and we can expect to learn the basics to make a hologram for anyone easily.