Anatomy of a Project: HDTV Animation for a Large Screen Venue

The D-ILAs are capable of projecting high definition images on screens upwards of 20 feet, thereby targeting cinemas and high-end home theaters. JVC's product manager, Frank Taylor requested us to come up with a one minute commercial that would help make the name, D-ILA more familiar. No limits were placed on the creativity or concept, but the budget was tight.

Hunkering down with Jack Ehrbar, our chief animator, I decided our firm would submit two storyboards, one by Jack and one by me. Using the simplest story telling tools, our hands, voices and bilabial sound effects, we honed our offerings down over the period of a week into my board that spoofed the old Victor trademark, featuring a dog and a Victrola (we experimented with a lot of ways the dog could show his disdain for analog). Jack's board, featured a pair of archeologists trying to open the secret chamber of an Egyptian pyramid, there to find the Pharaoh, alive, popcorn bucket in hand, watching a D-ILA projector.

On presentation to the clients, I learned that JVC no longer owned the rights to the famous dog logo, which left me desperately pitching one idea. The experience is a lesson in group satisfaction. One member of the committee wanted the history of JVC somehow presented in the piece. I suggested that the hyroglyphics on the walls of the tomb comically feature key symbols for JVC's pioneering moments in broadcast history. As the archeologists press these symbols, the inner chamber is opened. Another committee member wondered if Egypt was the best locale. Someone remembered that a pyramid had figured in a recently successful print ad campaign. Still another suggested lengthening the concept to a five minute film. A reiteration of the budget quashed that idea. Eventually the charm of the storyboard won out (or no one really wanted to waste another week while we churned out another concept).

With approved board, I gathered the creative team together to build a workable schedule and divvy up the work. Jack had recently installed Discreet's 3-D Studio Max and several of Digimation's plug-ins and would be responsible for all the digital animation. I would write the dialog script, cast voice over talent, execute the soundtrack and supervise the marriage of sound and picture to a D-5 HDTV master tape.

Since the deadline was short, only four weeks, we decided to work simultaneously in spite of the fact that usually the sound gets done before the picture. To achieve this compression of effort and time, Jack listed all of the tasks and scenes that did not require lip synchronization, while I schedule the minimum time to get a script written, approved and voice recorded. While I was scripting and casting, Jack could be modeling the characters, designing the pyramid scenery, and building the correography. After I had completed the voice recording, Jack would begin the lip synching while I worked to find the music and effects for the final mix.

Since our shop is not yet equipped with HDTV recording capability, we planned to work in 525 digital format at a 16:9 ratio, doing all sound to a BetaSP "proxy copy" of the piece and laying the mixed audio to a SMPTE coded digital audio tape (DAT). With the exception of a 12 frame test, we would not see the final HDTV result of our labor until the final stage, when we carried our 1,800 HDTV frames to Tape House in midtown Manhattan, for the frame-by-frame record to a Piranha digital disc recorder and thence to D-5. Like ancient Egyptians, we were playing with fire.

Discreet's 3-D Studio Max 4.0

Barely had we begun our modeling work on 3-D Max 3.0, than a review copy of 4.0 arrived. Aware of the significant advantages of the upgrade, we decided to proceed with the newer version. Regrettably, the 3.0 character models did not port well into the new version of Max, no doubt due to the huge amount of reworking that was required of the core program in order to achieve the significant improvements of 4.0.

One can complain about the issue of upgradability all one likes. It's nice to have all work, archived in Version 3 instantly portable to Version 4. The facts of life, however, determine that often a product must undergo such severe core code reworking (the kind of rework that really makes an upgrade worth buying) that work done in early versions is difficult to plug in to newer versions. Such is the case between Max 3.0 and 4.0. Now you know. If you are planning a 4.0 project for a client that did a lot of work in 3.0, you can decide if you want to stay in 3.0 and bill for that level of effort or redesign the elements and build in 4.0. My opinion is that the time saved by 4.0's significant improvements will compare favorably with working in the older version. It's not a big deal if you know the answer before you bid the project.

That said, what are the advantages of 4.0 when considering an upgrade or a first purchase of Max? We see the improvements breaking down into four areas; Interface, Tools, Plug-Ins and Integration.

The 4.0 interface allowed us to right click on any object to obtain all the commands appropriate to that object without referring to the command panel. Our IBM Intellistation, equipped with two, 1 gigahertz Pentium processors, is equipped with two monitors, which allows the 4.0's "Quad" display window to neatly segregate functions into "active" and "accessory" levels of priorty. Jack can therefore work on screen A, 90% of the time, but refer to screen B, for immediately essential control functions. When JVC's clients came to see the work in progress, the Quad layout allowed them to concentrate on the large frame displays without getting distracted by the control screen, which we simply shut off during the check rides.

Usually, we obtain client approval at each critical stage of animation, beginning with written approvals of the storyboard, the characters, the physical setting, correography, lighting-texture-atmosphere and audio mix, in that order. Max 4.0 seems well designed to this flow. For instance, Active-Shader is a new, "pre-render" timesaver that updates the frame preview as each atmospheric, lighting and texture parameter is changed. Since we adjust these parameters after the choreography is approved, we avoid this function's only conceivable shortcoming, which is that it takes a bit of time to redraw the frame after an object's position has been altered.

Since our project involved the creation of a "real world" environment, employing cinematic effects, 4.0's MuliPass feature provides instant previews of Depth of Field, Motion Blur and various camera angles without rendering entire frames. The speed of these feature is determined by the processor(s) and video card, which, in the case of our IBM's Oxygen Wildcat card, saved us an aggregate of nearly one workday during the course of a month's project. It also helps to have a large amount of RAM onboard - in our case, half a gig.

The new Tools included in Max 4.0 that assisted in this project were, HSDS, Skeletons, Skinning, Polygon Modeling. In the past, we tended to create a basic humanoid character and then modify it into the cast that would fill an animation. This produced a similarity between characters that was determined by what was easy to individualize from character to character.

Hierarchical Subdivision Surface modeling, or SHDS, grew out of seminal work at Softimage which provided isolated areas of functionality, based on whatever object divisions that artist cared to create, such as hands. Max 4's advancement on this basic concept allows high degrees of detail, color, texture to be manipulated on one part of a character, the face for a close-up scene for instance, without affecting the complexity or render times for other areas (or other appearances) of the character, saving enormous amounts of design and rendering time. The use of "proxy" objects, commonly used in earlier 3-D software products, is eliminated.

Our Arabic character, Farag, for instance, required such detailed skin texture in one close-up shot, that Jack had to actually scan my swarthy Mediterranean cheek to get Farag to look authentic. The cheek textures were controlled at the face level hierarchy, which was eliminated for long distance shots where the texture was not apparent.

Skeletons, bones, whatever you call them, have always been a blessing and a curse since we discovered them in Lightwave 4.0. In Max 4.0, they now represent both the "physics" of the inverse kinematics (IK) as well as the shape and size of the character they occupy, making the task of switching between characters and tweaking the choreography a lot easier.

Another useful tool in 4.0 is the Render Elements feature that allows the user to separately render the elements of a composited picture. Z-Buffer, Alpha Channel, Shaddows, Reflections or Refractions and others may be rendered independently into their own windows and exported to compositing tools like Discreet's Comubustion, Eyeon's Digital Fusion or Adobe's After Effects.

Plug-Ins from third party vendors need not be a concern to animators migrating to higher versions of Max. Traditionally, Max has either incorporated key plug-ins or the third party developers have courteously upgraded their products at little or no cost to be compatible with the new version. The following plug-ins have now been incorporated into Max 4.0:

To date, we have had no problem with plug-ins offered by Digimation, the leading third party vendor for Max plug-ins. In fact, Digimation's personable technical staff go way beyond the call of duty to make sure every product is supported and converted with successful results.

One particularly useful Digimation plug-in, Stitch, worked incredibly well with 4.0, allowing us to build flowing, Arabic robes for Farag. Stitch is a fabric modeling software that works much easier than previous attempts at fabric rendering by other manufacturers. On previous assignments, a robe or dress would always cause a problem if the material was caused to stretch or unfold to accommodate a spread of the underlying joints, arms or legs. Unsightly gaps, or bizarre creases would often form, requiring intricate patching or hand retouching of affected frames. Stitch figures allows garments to be designed from the same kind of patterns that a seamstress uses, with "real world" logic, thereby eliminating repair routines and making the learning process highly intuitive. (Jack intends to review Stitch in a future edition, so enough from me here.)

As a result of Discreet's acquisition of 3-D Max, users can expect a high degree of integration between this version and Discreet's other products. While we did not test Fire, Flame, Flint, etc., we did output one sequence of our animation from Max 4 to Combustion in order to composite a background scene. This was facilitated by using Discreet's new RPF file format which incorporates 13 channels of 3D metadata in frames that are carried from the animation program into other, compatible programs (Combustion is one). Other programs using the RPL format offer only 6 channels.

Case in point: One of the few Digimation tools which has not yet been upgraded to Max 4.0 is Phoenix, which we needed in our production to create torch flames. For these elements we fell back to Phoenix running on 3-D Max 3.0 and rendered these elements in RPL format, storing the alpha, G-buffer and UV map parameters in the metadata channels. We had to sacrifice the reflection mapping data, but our schedule dictated that there wouldn't be time for rendering torch reflections anyway. Had the schedule been longer, RPF metadata, dumped into a compatible compositor, would have been essential.

At the time of this project, we had just received our first copy of Combustion and did not feel confident enough to use it under "client pressure." We used Eyeon's superb Digital Fusion software to composite the torch flames into each shot and the results were predictable and fast.

While Jack was working the animation, I proceeded from approval of the script to casting the voices for the archaeologist, Farag, the Arab guide and the Pharaoh. The client immediately liked Tony Wellman and I resurrected my best horror voice for the Pharaoh's climactic appraisal of D-ILA. For Farag, we had to go to a genuine Egyptian, Fawzi Abdeltawab, who translates Arabic for our firm's foreign language translations service.

Digital Fusion also assisted us in the next phase of production, which was to check our animation results in 525 digital standard. The rendering of HD (1,080 pixels by 1,924 pixels) frames and their frame-by-frame assembly on D-5 tape is a very expensive process. We wanted to nail down every detail "offline" so to speak.

To do this, we first generated standard digital television (SDTV) frames in "letterbox" style, and then assembled them in our DPS Velocity nonlinear (NLE) editing suite. Here, we caught major errors, adjusted cut points between scenes, checked the accuracy of the lip synch, color balance, etc. After making the "gross" adjustments to the animation, we re-rendered the program as HD frames and then batch composited them in Digital Fusion to be letterboxed 525 frames. After reassembling the frames in Velocity, we examined the finer points of the animation to see if we could find anything wrong. We did, and corrected only the scenes that were problematic.

There's only so much one can do to check an HDTV animation in the SDTV universe. One mistake was a hole in the back of the Pharaoh's head that allowed light to shine through his open mouth. Instead of going back to the animation, Jack merely took the twenty offending frames into Adobe Photoshop and painted in the mouth.