Recommended hardware for the taraVRbuilder

1. Basics
Basically the taraVRbuilder requires a 3D graphics card and the CPU of the computer in equal measure for the construction phase and animation of very large 3D scenes. Therefore it is meaningful to adapt these both components with each other. E.g. an exchange of a very fast 3D graphics card into an older system with slower CPU and vice versa the exchange of an updated CPU by re-utilization of an old graphics card wouldn’t be very suggestive.

2. CPU
The taraVRbuilder supports Dual Core CPUs. The CPU is responsible for the reaction rate of the software in the editing mode (e.g. loading time of the project), but also for the calculation of the animations in the 3D model. Basically you can say: The higher the CPU cycle, the faster the system.

3. 3D graphics card
The taraVRbuilder also runs with so-called onboard graphics cards. But the capability of these integrated graphics cards is not enough to present large 3D scenes flowingly. Therefore we recommend “real” 3D graphics cards of the producer nVidia or ATI. The capability of both (in different price ranges) is similar, as far as our experience goes the software of the operating company of graphics cards from nVidia is better. taraVRbuilder supports both graphics modes: Direct 9.0c and OpenGL 2.0.

The expensive OpenGL CAD graphics cards are not mandatory, the cheaper consumer cards e.g. with the GeForce chip set have, referring to taraVRbuilder, the same capability!

4. Main memory
Very large 3D scenes require not only the graphics card memory but also the main memory of the computer. Therefore we recommend the most usable amount of main memory that is supported by the system software. Upper limit is ca. 4 GB for 32 bit system software, and 8 GB for 64 bit system software (e.g. Windows Vista Ultimate 64 bit).

Tips und tricks for handling with very large taraVRbuilder projects

1. Basics
The taraVRbuilder can get „slower“ by using suboptimal hardware (see document „taraVRbuilder - recommended hardware“) and by building very large projects (amount of 3D objects > 1000) That appears in a slow and jolting navigation or animation in the 3D scene and in a slowed reaction rate of the product line (e.g. span of time between mouse click and interaction > 1 second). In extreme cases the available main memory respectively graphics card memory won’t be enough and error messages follow.

Main cause for these effects is the administration and interaction of the 3D scene. Two reasons can be located:

• The amount of polygons in the 3D scene is too high
• The amount of separate objects is too high

Reduction of the amount of polygons
The 3D objects in the libraries of taraVRbuilder are already simplified in its level of detail to keep the amount of polygons as low as possible. Therefore the amount of polygons is first problematic, when the user imports many very detailed external 3D objects in the 3D scene. To avoid this problem we recommend when importing external objects to use our software taraVRoptimizer or different programs to reduce polygons like e.g. Deep Exploration of the company Right Hemisphere.

Reduction of the amount of objects
Is the amount of 3D objects in the taraVRbuilder scene too high, the program needs too much time to analyze this 3D structure. Because this analysis is necessary at every interaction with the 3D scene, slower reaction rates and jolting animations appear. In the following special methods and program adjustments will be explained, which preferably reduce this problem.

2. Hardware
Of course performance problems with the taraVRbuilder appear earlier by using suboptimal hardware. Therefore CPU, graphics card and main memory of the computer should not meet the minimum requirement but our hardware recommendations (see document “taraVRbuilder – recommended hardware”).

3. Program adjustments
If your taraVRbuilder scene runs noticeable slow, you can increase the performance with the help of different program adjustments. Following adjustments are explained:

• Using dynamic level of detail (LOD)
• Switching the detailed supporting means to textures
• Deactivating shadows
• Fading out partitions of the 3D scene
• Working in partitions and merging partitions

Using dynamic level of detail (LOD)
In the taraVRbuilder every 3D object of the library has at least 2 levels of detail. External imported 3D objects automatically get an abstract level of display for the import. The user can activate these levels of detail against the distance of the viewer and the amount of objects. At the menu File – Settings … - Details

In the input field „View distance of details” the user can define, at what distance from the viewer 3D models change its detailing automatically. In the upper example all 3D objects that are more than 100 m away from the viewer will be displayed as abstract automatically e.g. at the flying through the entire scene.

Switching the detailed supporting means to textures
A huge handicap for the hardware is the display of detailed supporting means of the materials handling (here especially rolls, minirollers etc.). It is recommended for large projects to switch these detailed 3D supporting means (if texture equivalents are available) to textures. With the button “Use textures supportings means on/off” the user can switch all 3D supporting means to textures with one click and vice versa. (The global dispatcher is available from version 7.0.12)

Deactivating shadows
In the taraVRbuilder 3D objects and the animated transported materials cast shadows at the floorage. For the calculation of shadows the CPU and the graphics card need processing power, that can be used for the display of large scenes if shadows are deactivated. The shadow can be deactivated at the File – Settings – Shadows.

With the buttons „Calculate transparency“ and „Analyse animation“ sub-functions of the shadow can be deactivated, which also release processing power. With the button “Show shadow” the complete shadow can be deactivated.

Fading out partitions of the 3D scene
All 3D objects and also entire (particular) projects can be made invisible in the taraVRbuilder using its visibility in the project structure. Invisible 3D models or even entire subprojects don’t load the graphics card. This is recommended e.g. for partitions of a 3D scene, those already have a certain planning status and the user doesn’t need to work on for the moment. Animations of goods also run on invisible 3D objects, but won’t be shown.

Working in partitions and merging partitions
Since faded out 3D objects still perform animations even if invisible, you still load the CPU of the computer even if you discharge the graphics card. So the next consistent step is a splitting of the entire scene in partitions and a working with these individual subareas. The resulting TPD-files can be handled simpler and faster and can be merged to an entire scene again at the end of the planning process. While merging e.g. transitions of the materials handling will be identified automatically and the animation of the entire scene will run automatically. With the following dialogue you can insert single TPD-files to a (master) project. (File – Insert project):

4. taraVRoptimizer
The taraVRoptimizer can simplify complex structures in 3D objects and discharge graphics cards considerably this way. A CAD export e.g. can be made up of thousands of individually geometries, who charge the graphics card with the analyzing of these geometries very much. Long loading times, very jolting navigation up to the crash of the program with storage overflow failures are the result.

Within a 3D object the thousand individually geometries can be merged to a few large geometry blocks automatically with the taraVRoptimizer. The graphics card can cope with this kind of simplified 3D object very good.

The taraVRoptimizer comes with a wizard that is very easy to handle, and that makes a preparing of 3D objects for an optimal demonstration without any deeper basic knowledge possible in a short time.

So it is recommended for complex CAD data to simplify these data with the taraVRoptimizer before importing in the taraVRbuilder. This is also essential for large 2D layouts. Even here the taraVRoptimizer can simplify the line geometries without any loss of data!

A further use case for the taraVRoptimizer in combination with the taraVRbuilder is the merging of finished partitions to one geometry.

The idea here is also the reducing of the amount of the 3D geometries again. You can export e.g. a storage with 10 shelves and a hundred goods each as statical export out of the taraVRbuilder, optimize it in the taraVRoptimizer then and import the entire storage afterwards as a 3D object in the taraVRbuilder again.

How to create presentation videos using taraVRbuilder

1. Settings in taraVRbuilder for optimum video quality
One general remark: For the video, all features (which otherwise affect performance) should
be activated in order to ensure that the video, after rendering, will run smoothly, i.e.:
   - Activate: show shadows, evaluate animations, evaluate transparencies.
   - Furthermore, DirectX 9 should be set as renderer in taraVRbuilder (right mouse
   key menu in the 3D window, then settings, then renderer, then DirectX9 (re-start
   taraVRbuilder, if necessary!!!)
   - In the case of DirectX 9 (and only in conjunction with graphics cards which support
   DirectX 9 on the hardware level too), AntiAliasing can be activated. This mitigates
   the edge effect which causes strong crizzle (refer to Bspbilder) of overview
   images, especially if 2D layouts are integrated. In order to set this feature,
   proceed as follows: right mouse key menu in the 3D window, then settings, then
   defaults, then tab "Direct3D 9 Device Settings" then switch "Multisample Type"
combobox from None to NoneMaskable, and additionally set "Multisample Quality"
combobox at 3.

2. About the target system
The first question you should ask yourself when creating video streamings is: "What do I need my video streaming for, what is my target medium / target system?

2.1 VIDEO DVD: This medium can be used for very large video files which are then converted by an external tool to a video DVD. We hence recommend noncompressed AVI with a resolution of 720 x 576 pixels (PAL). The resultant AVI file becomes very large (several GBs!!! IMPORTANT: Make sure that there is enough harddisk capacity!!!), but this system also avoids quality losses due to compression. The non-compressed AVI file can then be used to produce a video DVD using a DVD burner and suitable DVD burning software (which is usually included). This DVD can then also be played on external DVD players.
2.2 Video sequence for use on a PC/notebook (for example, in PowerPoint presentations): A compromise must be developed here between the file size of the video clip and the presentation quality. Various compression techniques are available which reduce the file size, but also affect the rendering quality. We suggest selecting the compression method also as a function of the video contents:
In the case of significant changes from each individual video frame to the next (for example, in the case of a round flight through a plant), too high a compression factor (for example, DivX) is not recommended because the rendering quality is adversely affected by serious interference. If the essential video contents remain unchanged (for example, stationary camera position, with moving goods only moving through the picture), AVI in DivX or MS MPEG 4 V2 compression are recommended. This then results in very small files (for example).

3. Video size and AVI compression (from non-compressed to DivX)
1. The video resolution should always be selected in such a manner that the video is presented in 100% (no size scaling) of its size. High video resolution (for example, 800x600) is recommended for beamer presentations.
2. The higher the resolution the larger the data volume which the harddisk must read during playback. In the case of non-compressed AVI, the harddisk is sometimes unable (especially via the network drives) to read the data at a sufficiently high speed and the playback becomes unsteady (even though a rate of 25 frames per second was chosen).
3. The frame rate should always be set at 25 frames per second (at a rate of 15 and above, the eye sees a smooth movement).
4. High-resolution video streamings should be compressed because the data volume would otherwise become enormous and point 2 become a problem.
5. With regard to video compression (in taraVRbuilder, the compression method is selected after the individual frames have been generated), a compromise must once again be found between rendering quality and file size, depending on the particular application. - If, for instance, a "real" video DVD is the target medium, the video should be generated without compression because compression will take place anyway during conversion to the video DVD format. - If the video size is to be kept as small as possible, we recommend DivX (www.divx.com). - If the video is to be streaming-enabled (i.e. for presentation on the Internet, for example), we recommend converting AVIs to WMV. (http://www.microsoft.com/windows/windowsmedia/de/9series/encoder/default.aspx) - I always use the "Microsoft MPEG-4 Video Codec V2" compression method for compressing AVIs. This provides good quality and a small file size. - Video Codecs for AVI are provided when the latest version of Windows Media Player is installed.

4. General statements
- The loss of quality due to compression is particularly striking in the case of round flights. Videos with a stationary view of the scene show much fewer annoying artifacts due to compression in which case DivX is a good choice in the case of a stationary view. - Round flights should not be set at too high a speed. - If possible, non-compressed single frames (BMP) can be generated from within taraVRbuilder which can then be compiled by an external tool (such as Adobe Premiere or, as a simpler method, VideoMach) to AVI. It is, for example, then also possible to resample single frames with a higher resolution (1024x768) into a video with the result of eliminating crizzle (http://www.gromada.com/videomach.html). - The screen saver must be deactivated when the single frames are rendered for the video!