3D Modeling - Horse

Last night I started modeling the horse in Cinema 4D. Tonight I spoke with the softimage instructor (A. Cho) about boning the horse. He pointed out some strategies for modeling so that around places on the model that are meant to move (e.g. joints), there will be enough polygons of the right structure to accomodate the animation. He also suggested that we storyboard out every scene of the animation so that we don't wind up modeling things like Apollo's feet which may never be seen if he is standing in the chariot. Other suggestions he had were to simplify some aspects of Apollo and and exagerrate others. Since he will be a small part of the screen, it may be moe efficient to simplfy his "headdress" and emphasize the four thundering horses.

Shopping

Visited Urban Ore looking for any equipment possibilities. Found an octagonal poker table--a little beaten up but workable for our TUI table. It was a little pricey though so we decided to do a little bargain hunting online only to find that it was already a bargain at $85. It fit within our budget, its design fit our needs, and it fit into the back of my truck. We made plans to go back and pick it up the next day.

We also visited Al Lasher's looking for LEDs (for lighting the inside of our cubes) and possibly micro video cameras. We left a little disappointed since they had nothing in stock that we needed nor did they know if they would have anything in soon. Next we visited Mike Quinn's. We were also disappointed here since they're closed on Mondays. A side trek to a Radio Shack revealed that they didn't have the kinds of LEDs that we'd probably need.

To get even more detailed specs for our cameras and LEDs, Stephen dusted off his high school trig and figured out the optimal FOV (field of view) requirements for the cameras and LEDs. Turns out that our initial webcam purchase has too narrow an FOV to be used under the TUI table. To get the full table within the field of view, the camera would have to be mounted on the floor and the table would have to be 4.5 feet tall!!! The under table camera will need to have an FOV of between 71-81 degrees for a table just under 4 feet in diameter and 30" tall. To get a good "magic window" effect for our AR Viewer optimized for virtual objects and markers approximately 1-2 feet in front of the viewer, an FOV of around 56 degrees will be required. Our LEDs within the small little cube will need to be really, really bright to pass illumination through markers printed on paper and have an extra wide viewing angle of between 90 and 100 degrees.

Shockwave Export and Lighting

I tried exporting the model of Apollo's chariot into the shockwave 3D format. However, when I viewed the model in Director, it only could be viewed properly from certain angles. The bottom and one side of the model appeared to be there but in darkness so it could not be seen. I added lights around all sides of the model and exported it again. The model can now be seen properly from all angles. I am not sure if this matters in DART's 3D world. It is possible that 3D world has its own lights. I will need to test this. In the future, if the model is meant to be manipulated by the user like in the case of the moonshot rather than a preset animation with the only possible angles being preset, we may have to make sure that models are properly lit so they will not be in darkness at some angles.

Camera Specs

Researched camera options more and made a list of specific requirements for each of our cameras (AR Viewer and Under-Table TUI). Got directions and made plans to visit some electronic stores on Monday. We decided that the camera for the AR viewer should be a pinhole type to make focus a non-issue and that a resolution of 380 TVL or better would be necessary to ensure we can correctly identify AR markers.

Apollo's Head

Worked on reducing the number of polygons on for the model of Apollos head. Surprisingly, if done right the model can look very similar with a drastic reduction in the number of polygons.

Lighting, Cameras, and Cubes

Researched camera parameters such as Lux, FOV, and resolution. Met with Scott who got us talking with James from the Photography department. Together we did some lighting calculations and set up some lighting examples for us to have a frame of reference with which to make better camera purchasing decisions. Worked on the pipeline for displaying AR markers inside our see-through acrylic cubes. ID'd the patterns for each of the sides of a test cube and was able to track it with DART. Song suggested printing the graphics inside in a shape allowing one to cut and tape it into a cube shape that could be inserted into the acrylic cube.

Equipment

Spent the day checking competing products for our AR Viewer display. Found a nice 8" TFT display that was inexpensive and light weight with a good resolution and minimal power draw. Ran out of time and wanted to get exact dimensions before purchasing. Will likely do this tomorrow along with purchasing other needed equipment for the viewer.

Acrylic cubes were purchased to be used as our Age Phicons as well as designs worked out to light them from within and skin them with AR markers.

Loose Ends

Today, we checked out Song's Apollo character face in real-time AR. It loaded pretty quickly and tracked well. We did a live demonstration of this to some visiting CSU administration officials.

We also heard back from Gavin regarding our software request--he doesn't know if it's come in yet, and we'll be responsible for installing it when it does.

We emailed the folks at the DART development group to get the answers to some lingering questions we have concerning some functions and their use (especially incorporation of ARTag markers for TUI tracking).

More on 3D

After exporting Apollos head to the shockwave 3D format required for DART, there were many artifacts that did not translate well from the original model. Apparently shockwave 3D format does not support spline/nurbs objects so I altered the headdress of sun rays to regular polygon objects. They are essentially the same, it just takes longer to model them. Also I am working on reducing the number of polygons which make up Apollos head so that it will take less time to load/render him over the live video. Also as far a texturing, it seems I will have to rely on simple texture mapping and colors, because materials like glow are not exporting properly into the shockwave 3D format.

More on Software Request

Emailed Gavin regarding our software order status.

3D Modeling

Worked on 3d model of Apollo's head and his chariot.

Chabot thoughts

Just had one thought on the moon landing simulation/video game at Chabot. Only one user can use it at a time. Many kids crowded around the user and simply had to wait their turn. I am hopeful that how we design our table and AR viewer will allow more interaction/collaboration.

Chabot RE-visited

Went to Chabot. Some of the exhibits were very engaging. If our project were to partner with Chabot then it would definitely be less of a floor exhibit and more of an educational outreach or managed classroom-type experience, such as their collaborative "Mission to Mars" experience where students actually man and operate a "mission control" center to fulfill a mission.

Image Stickiness & ProtoTUI

Tested "image stickiness" to the various treatments to the glass. Both the frosted window film and the white oil paint over sandblasting yielded good results and are acceptable. The window film appeared to work better but had flaws in its surface similar to air bubbles trapped under the film but upon examination proved not to be trapped air. This left the surface unaesthetic. The flaws could have been caused by application error--detergent film left on the glass during application. Therefore, another attempt may be made using this technique.

Tracking of markers from below the glass treatments was tested and actually worked--providing the basis for our TUI prototype!! Testing revealed certain optimum camera parameters--low gamma, backlight on, color correction of the marker through the glass, and switching to Black & White mode. Though it did track, it was a little jumpy. The size of the marker used was determined to be smaller than should have been used for the camera-to-tabletop distance. Larger markers would help stabilize the tracking as would use of ARTag Forward Error Correcting markers.

Some experimentation was also done with overlaying dynamic text over the scene (such as for x,y coordinates or rotation of objects).

Research

Researched symbology we will use for the different time periods. For example, zodiac symbols may be use for the Greek or Galileo time periods, while symbols from the Mayan calander could be used for that time period.

More on Window Treatments

Having acquired white oil paint and a brush, Stephen set about to do the final treatments to try out. The "Vellum Paper and Spray Adhesive" option section was cleaned off and the frosted vinyl window film was applied in its place. Then the oil paint was brushed over a masked section of the sandblasted section. The excess was then simply rubbed off with paper towels leaving a finish pretty much as we had hoped and anticipated--similar to the "White Spray Paint" section that did so well gluing the image yet still quite transparent from behind.

Afterwards, we learned how to map virtual "obstructions" to coincide with real objects surrounding the marker panel to get foreground occlusions, thereby enhancing the illusion that the augmented objects are really where they appear to be. Although this worked, the alignment of the virtual obstruction with the real world obstruction would fail. It was decided that this had to do with providing DART with the proper camera distortion and calibration information. Since our previous attempt to learn how to do this ended with the information being in an unusable and untranslatable format and now seeing WHY it is so important under certain circumstances to properly calibrate the camera, we researched a method we could place in our pipeline to do this. After much time, we finally came across "GML Camera Calibration" which seemed to provide the desired information in a usable format. Several trials were made but we will wait til we have more time to do the actual calibration.

Chabot Visit

Decided to finally go to Chabot as planned to see their exhibit layouts and direction. However after making the drive out there we got to the front steps only to find that they are closed on Mondays and Tuesdays--something we seemed to have missed while checking out their hours on their website. We will attempt to reschedule for later this week.

Puzzles

Discussing collaboration, Song and I developed an idea that would add interest, depth, and a goal worthy of collaboration to wrap our project around:

Each of four "Age" phicons would be assigned a color corresponding to its Age. Yellow for the Greek Age, Red for the Mayan Age, Green for the Scientific Revolution Age, and Blue for the Modern Age. Each Age phicon would have a set of 6 symbols on it relating to its Age. If any single Age phicon is presented on the table, the Age's content will begin. Two or more Age phicons presented at once results in an interface with a bar of 4 colors. Presenting the correct symbol for each Age Phicon in the right order then unlocks a final content package tying all the Ages together (the chances of doing this correctly by chance are 1/1,296). To discover the correct symbol for each Age, each Age's puzzle must be solved. Along the way, information about each Age will be learned as well as clues given that aid in the solution of the puzzles for the other 3 Ages.

More on Window Treatments

Purchased frosted vinyl adhesive film.

Where's the fun in collaboration?

Did research into possibilities for developing more a more puzzle-like game feel to the project. Myst-type games are fun to play with others since each person has a different approach to problem solving and each person may also notice different details and clues. A puzzle presented to two or more people is a goal-oriented task that lends itself to, and may encourage, collaboration. Researching other possibilities for content, the idea of the Mayan calendar being the function of two wheels with different ratios lends itself to the idea of being part of some sort of puzzle (such as finding a particular celestial event). Each set of Mayan calendar symbols could be AR markers that when in the right combination could be ID’d as a marker panel to unlock AR content. Supporting clues could possibly come from the content areas of the other time periods (such as learning in the Greek time period that the circle/dot symbol is for the “sun” god Apollo and using this to decipher and deduce the Sun Glyph position in the “orbits” puzzle in the Scientific Revolution time period content).

More on Window Treatments

Discovered a frosted vinyl adhesive film to also be tested as a potential glass tabletop treatment.

Camera Calibration

This afternoon we decided to make a camera calibration file in order to see if this would improve the tracking ability of the camera while running DART. We made the file but realized it is probably compatible with ARToolKit only and needs to be rewritten specifically for DART.

More on Window Treatments

This afternoon we decided to calibrate our camera and make a camera calibration file in order to see if this would improve the tracking ability of the camera while running DART. We made the file but realized it is probably compatible with ARToolKit only and needs to be rewritten specifically for DART. I compared the results of the 6 different glass treatments and their practicality as the basis for our TUI tabletop. The treatment that had the highest projection “stickiness” was simply the white spray paint on the top of the glass. However, its durability is not practical, not even if it were sprayed on the bottom of the tabletop glass where marker visibility would be impossible due to additional light diffusion through the 1/4” glass. Durability was an issue with some of the other treatments as well even though they showed signs of high clarity for viewing marker patterns through the glass. The sandblasted section seemed the best treatment overall but the optical characteristics of its surface are easily altered by oils and perspiration filling blasted gouges in the glass. A treatment will be attempted using oil-based paint over sandblasted glass to try to fill the gouges with white pigment (increasing projection “stickiness”) and keep the paint from being easily scraped off the glass (increasing durability). Met with Scott.

Sketches

Made some sketches for Apollo carrying the sun across the sky in his chariot.

Window Treatments

Learned how to Sandblast and acid-etch glass as possible treatments for the table top for a TUI. Treated sample glass with following six treatments: (1) Sandblast, (2) acid etch, (3) Acrylic Matte Spray, (4) Vellum paper with spray adhesive, (5) white spray paint with matte spray (for protection), and (6) wallpaper paste with titanium dioxide white paint pigment powder and matte spray (for protection). Met with Bill Nye regarding tracking technologies (including ARTag markers). Gavin reported that Director WILL be purchased for both platforms afterall given our needs.

C++, C--

Took another stab at re-writing the “mk_patt.c” code to run on the Mac. Was able to get the code to work again and established that it was identifying markers but still couldn’t get it to draw the “marker ID square”. The functions associated with these are complex and part of the obsolete “gsub.h” set; re-writing it looked to be the last ugly option before us. However, Song suggested a checking to see if a new version of ARToolkit were available. It was, and the notes indicated it was more stable across platforms and had more updated routines that didn’t rely on “gsub.h” function sets. We compiled this, trained the system to use our own custom markers, and defined marker panels allowing us to utilize them in DART. We tested this by successfully taking our P-38 3D aircraft and augmenting our mousepad markers with it. Turns out that the re-written code we made was working fine minus the “marker ID square”. The square was just an indication a marker had been found and was functionally replaced by a printed feedback statement I entered saying “marker found”. The square was not required as an alignment template as originally interpreted from the code and instructions. Gavin reported that we would need to discuss with Scott if we could work with just one platform installation of Director for the time being. We wrote back noting that to keep on track with our project’s Milestones we’d have to have the software running on both PC AND Mac platforms at the same time since we would be testing communication features we would utilize heavily later on in the project.

Softimage

Viewed 3Dbuzz videos as an introduction to Softimage.

More on the Software Request

Met with Scott regarding our progress. Gavin reported that the Multimedia Graduate Department would not be able to supply us with our own workstation copies of SoftImage or Cinema 4D. He also misunderstood our request for Macromedia’s Director MX 2004 to be Macromedia’s Dreamweaver MX 2004. I made a chart of possible glass tabletop treatments to be used for our TUI and evaluated the potential costs/benefits associated with each method.

Got da Glass

Found glass tops to experiment with for Tabletop TUI specs.

Custom Markers (continued)

Continued designing custom markers.