TECHNICAL ANALYSIS OF ANALOGIES OF STEREO DISPLAYING TECHNIQUES WITH 3D GENERATED SCENES IN VISUALIZATION

Transcription

1 DAAAM INTERNATIONAL SCIENTIFIC BOOK 2008 pp CHAPTER 64 TECHNICAL ANALYSIS OF ANALOGIES OF STEREO DISPLAYING TECHNIQUES WITH 3D GENERATED SCENES IN VISUALIZATION SKALA, T.; TODOROVAC, M. & MRVAC, N. Abstract: Scientific visualization often requires display and perception of depth and 3D in various applications. A CGI application has been designed to allow interactive construction of stereo images from POV-Ray SDL models that has enabled comparison between two different methods of encoding stereo images. Red-cyan anaglyphs and interleave images for LCD shatter glasses have been considered in this paper. Two methods of encoding and viewing stereo images on computer monitors have been studied. Either method is not perfect: the logic cannot be cheated. Half of information in L and R pictures is lost. In interleaved image, every other line has been omitted from the resulting picture, halving the effective image resolution. Half of dots in image are going to be rendered in vain each time. In a common anaglyph, blue and green colour channels are lost from the left picture and red channel from the right. Key words: scientific visualization, POV-Ray, red-cyan anglyphe, LCD shutter glasses, 3D image Authors data: Teaching Ass. Skala, T[ibor]*; Todorovac, M[irsad]**; Prof. Mrvac, N[ikola]***, *Faculty of Graphic Arts, Laginjina 6, 10000, Zagreb, HR, **Faculty of Graphic Arts, Getaldiceva 2, 10000, Zagreb, HR, ***Faculty of Graphic Arts, Desno Sredicko 2a, Lasinja, HR, tibor.skala@grf.hr, mtodorov@grf.hr, nikola.mrvac@grf.hr This Publication has to be referred as: Skala, T[ibor]; Todorovac, M[irsad] & Mrvac, N[ikola] (2008). Technical Analysis of Analogies of Stereo Displaying Techniques with 3D Generated Scenes in Visualization, Chapter 63 in DAAAM International Scientific Book 2008, pp , B. Katalinic (Ed.), Published by DAAAM International, ISBN , ISSN , Vienna, Austria DOI: /daaam.scibook

2 Skala, T.; Todorovac, M. & Mrvac, N.: Technical Analysis of Analogies of Stereo 1. Introduction It is often required in scientific visualization and engineering to perceive 3D structure of an object, model, landscape or other entity. Scientists have studied binocular sight and its role in perception of depth in vision since Leonardo da Vinci. Since Wheatstone's paper (Wheatstone, 1838) on stereopsis in 1838 and his stereoscope, various methods have been attempted to offer space perception to viewer. Nowadays modern learning system needs new dynamic interactivity while all other used technologies are too static. Also, all 2D displays are not adequate anymore, so it is necessity to use modern technology and make changes, not just in education but also in business sector. Using visualisation, new products can be seen in the virtual space before it is produced. Changes on that can be made just by simple software modification. Students and business partners can better understand what wants if they can be generated in 3D structure and make interactivity with them when new ideas are presented. Most common experiments were carried out in 2D images with stereo information encoded using one of available methods. These methods include using anaglyph images which require coloured glasses for viewing and interleaved images for LCD shutter glasses. The former method typically involves two monochromatic filters with contrasted colours for each eye. Common examples are blue-red, green-red, blue-yellow and red-cyan glasses, the last patented by Steven Gibson in 1970s. Application of each of these filter techniques distorts the perception of colour. Whole of the colour information is not passed to both eyes. Resulting 3D image is constructed in brain. Research in this work compares two methods of 3D visualisation which will be discussed. These new technologies of generating virtual environment will be used in education. 2. Producing Stereo Image Authors used POV-Ray (Persistence of Vision Ray tracer) to produce images. POV- Ray is an open-source program, often used on GRID because of the fact that it is freely accessible, highly portable and easily compiled and optimizes for various platforms, including multiprocessor, multi core and hyper-threaded systems. POV- Ray on clusters and grids offers such rendering power and quality previously only imaginable on SGI clusters running expensive commercial software. Grid enable for tens of nodes to be used while rendering animations, ideally cutting rendering time by a factor of several tens of times (grid overhead doesn t allow for such optimization at the time of writing). This is even more important for stereo encoded animations, requiring each scene and animation sequence to be rendered twice ( In everyday life, mind perceives object in three dimensions by means of two dissimilar pictures projected from it on two retina (Brinkmann, 2008). The question arose what would happen if presenting to eyes is not the object, but two planar 790

3 DAAAM INTERNATIONAL SCIENTIFIC BOOK 2008 pp CHAPTER 64 images as they appear to each eye? This question had lead Wheatstone to invent stereoscopy. Fig. 1. Stereo projection Not different from original stereoscope, stereo vision has been achieved by simulating two cameras, looking in direction of the same POV-Ray SDL s look_at point, much like binocular vision of the human vision system. Point a has different projections for the left and for the right eye in the projection plane b b. The same applies for point a. Mind is able to reconstruct depth from positions of a and a. In the same manner, any number of objects can be represented by their left and right image (Birn, 2006). Fig. 2. A CGI application for interactive construction of stereo images 791

4 Skala, T.; Todorovac, M. & Mrvac, N.: Technical Analysis of Analogies of Stereo The cameras have been separated by a parallax, an empirically determined distance. Too small of a parallax distance, and the 3D effect will not be perceived. Too great parallax, and the eyes and the brain will not be able to assemble a perception of a 3D model, but they will perceive two distinct objects. Because of difficulties in using a stereoscope device with computer screen, a number of methods have been developed to encode L and R images into one, allowing three dimensional perceptions to appear only while viewing through appropriate filters, different for each eye. Additionally, web application has been made to enable engineers to practice making of stereo images. Stereo encoding method and encoding parameters are chosen onthe-fly. The left and the right image are rendered simultaneously if more than one processor core is available. The povdriver CGI (Common Gateway Interface) program accepts input form variables selecting model and the parameters. The data are then fed into the parametric model. Povdriver starts two independent POV-Ray processes and feeds the results into stereo combine program. Stereo combine is a GNU GPL licensed open-source program, allowing for red-cyan anaglyph or interleaved shutter glasses stereo encoding. It uses libgd library to manipulate PNG pictures. The closer look at both techniques will be discussed in the next chapter. 3. Comparison of Techniques 3.1 LCD shutter glasses LCD shutter glasses are in the PC area, certainly the best known and most widely used tool to the 3D representation. The functioning of the LCD shutter glasses is basically quite simple. To create the 3D depth pressure on a screen to obtain, each eye makes own image. That makes a shutter glasses in conjunction with a customized graphics card driver. The screen alternately shows the image or the prospect for the right and the left eye. Running a monitor with 100 Hz means that the picture 100 times per second makes new view. A shutter glasses halved this refresh, as an image now consists of two images - just one for each eye. The monitor should therefore have at least 120Hz vertical frequency so it may be in connection with the glasses still effective at 60 Hz. Since the reply to the refresh rate is linked, the graphics card needs to connect to the glasses in order to ensure synchronization. This is done by cable or by IR transmitter and receiver to the appropriate glasses. Shutter glasses work only in conjunction with a monitor tubes (CRT), as it can interchange image sufficiently fast. Flat panel displays (LCDs or TFTs) are too slow. But even with a CRT, there are many problems. Since the phosphorus coating of the screens light after a certain time, each eye sees always a kind of ghost image of the other, especially in contrast scenes, this "ghosting" increase. 3.2 Red cyan glasses Depending on the application, there are glasses with different filter combinations. Proven have red/green, red/blue and red/cyan (blue-green). In principle, any 792

5 DAAAM INTERNATIONAL SCIENTIFIC BOOK 2008 pp CHAPTER 64 combination works with any application, but wins the representation in the right eye to quality. Red/green and red/blue commonly found in books, comics and computer applications. Red/cyan is optimized for colour anaglyph stereo picture. Despite some drawbacks include the anaglyph red/cyan process, it is the most popular 3D technique and production of images and filter glasses is inexpensive. The view is simple and leads for most users immediately to success. The distance and the viewing angle are critical for image and the effect of the depth. An anaglyph stereo picture is provided of the 3D model of Tesla s three phase AC electric motor. For comparison, the following is the interleave-encoded stereo image for LCD shutter glasses of the same model. Note the colour deficiency in the anaglyph colours. Watching screen through red-cyan glasses in our experiment also caused appearance of ghosts : apparently the visual cortex attempted to compensate for the different colours in the same picture seen with the left and the right eye, and it caused strain and involuntary movements of the eyeballs. It is important to consider that the left eye sees the red image, and the right one cyan image impression of other colours is created in the mind. Watching anaglyph images with poorly calculated stereographic parallax and perspective may also cause significant eye strain, just like with LCD shutter glasses. Sometimes it required two consecutive passes through filter to completely eliminate cyan image from red lens and vice versa. LCD shutter glasses require a special program (or driver) that will cause graphic adapter and computer monitor to start interleaving frames of display. All information on the display that was not interleave-encoded (like menus etc.) will be blurred and unreadable. Fig. 3. An anaglyph image for red-cyan filters Fig. 4. Interleaved stereo picture for LCD shutter glasses 793

6 Skala, T.; Todorovac, M. & Mrvac, N.: Technical Analysis of Analogies of Stereo 4. Results LCD shutter glasses sometimes poorly synchronize with monitor s vertical frequency, causing flicker and eye strain. Eye needs to adapt few seconds before it perceives depth. One of the possible causes of confusion for eye and mind may be the lack of blur for objects out of focus in the POV-Ray SDL modelling, causing eye to rapidly change focus to and fro, especially noted in perspectives with wider angle. Note that LCD shutter glasses are supreme for experimentally created and viewed MPEGs. Standard FFmpeg creation utility was used, which has no knowledge of stereo encoding, whether anaglyph or interleaved. It was hard to avoid blur of the anaglyph video, even at the best setting for MPEG quality. Separate red and cyan information appears to have confused MPEG s sampling and compression algorithms. Since MPEG is preferred to animated GIF in longer animations, red cyan anaglyph might have serious drawbacks in this context. Red-cyan glasses LCD shutter glasses Encoding easy easy Resolution complete halved, interleaving causes loss of every other line in both L and R pictures Colour fidelity anaglyph encoding complete causes loss of certain colour planes in both L and R image Subjective perception of requires visual cortex to slightly better depth learn Sharpness red known to be blurred better Use with LCD displays easy only on high vertical frequencies Compression for better 30-40% animated GIF Compression for MPEG better 20% MPEG quality sharp somewhat blurred Moiré effect greater, almost double same as original the original Flicker no considerable Eye strain some considerable Price ~ $0.50-$10 ~ $100 Tab. 1. Comparison of two stereo techniques 794

7 DAAAM INTERNATIONAL SCIENTIFIC BOOK 2008 pp CHAPTER 64 Fig. 5.Making an anaglyph image Fig. 6. Making an interleaved image for LCD shutter glasses 795

8 Skala, T.; Todorovac, M. & Mrvac, N.: Technical Analysis of Analogies of Stereo 5. Conclusion and Discusion These innovations are used on Department for Multimedia and Information system on collegiums Multimedia communications and Optoelectronics. As it is worked with the students there is great progress in learning and understanding of different fields as they see objects in 3D. Now student can by its own ideas create dynamic models and can link theory and praxis. Interaction is the key in learning while students can understand how something works as they can combine knowledge with the experience. Results made in the last 3 years show a big progress in students understanding and time spending in learning. Students do not need to leave the room or class and to go to the manufactory if they want to see some product and the way it works. For this paper we were modelling Nikola Tesla s AC motor ( and on the same way, it is possible to generate 3D object for different science projects. Two methods of encoding and viewing stereo images on computer monitors have been studied. Either method is not perfect: the logic cannot be cheated. Half of information in L and R pictures is lost. In interleaved image, every other line has been omitted from the resulting picture, halving the effective image resolution. Half of dots in image are going to be rendered in vain each time. In a common anaglyph, blue and green colour channels are lost from the left picture and red channel from the right. This will cause loss of information that mind adapts to, yet often a time on scale of few seconds is required while viewing image before brain learns to perceive depth from a pair of the filtered images with given parameters. Consequently this method of representing 3D is often not considered true 3D, but 2 ½ D imaging. In the future these technologies will be used in everyday situations. We are planning to make the application that will offer us a possibility to go around the 3D object and to make interaction with it so students will have the possibility to see and learn better about certain product. 6. References Birn, J. (2006). Digital Lighting and Rendering, New Riders Press, ISBN: , Berkeley, CA Brinkmann, R. (2008). The Art and Science of Digital Compositing, Second Edition: Techniques for Visual Effects, Animation and Motion Graphics, Morgan Kaufmann, ISBN: , Burlington, USA Wheatstone, C. (1838). Contributions to the Physiology of Vision. Part the First. On some remarkable, and hitherto unobserved, Phenomena of Binocular Vision, Available from: Accessed: Interactive construction (form) Available from: Accessed: Libgd library, Available from: Accessed: Stereo image of phase AC electric motor (shutter + anaglyph), Available from: Accessed: