The Talisman graphics initiative has turned out to be one unlucky charm for Bill Gates and company. Microsoft launched the effort one year ago at Siggraph, the computer industry's pre mier conference for all things visual. It was presented as a gold-plated plan to build advanced accelerator hardware that would enable PCs to deliver high-quality 3-D images. Boy, was it ambitious. Higher resolutions, faster frame rates, fancy "anisotropic" filtering--every whiz-bang feature you can think of was folded into Talisman.

Microsoft's marketing strategy was to license Talisman to chip makers. The latter would create the integrated circuits and add-in boards that would drive the technology into the hands of millions of consumers.

Cirrus Logic, Philips, Fujitsu and Silicon Engineering signed on for the first step toward that goal--the creation of a Talisman reference board, called Escalante.

But a few months ago, Escalante fell apart. A big stumbling block was cost. Reportedly, some $300-worth of parts were required just to build the board. To turn a profit, vendors would have to price it so high that only a Bill Gates could afford it. Mere mor tals preferred garden-variety, low-cost accelerator cards from the likes of ATI Technologies, Orchid, Number Nine and Diamond Multimedia.

Now, Talisman appears to be tilting further away from a hardware-based graphics architecture and reemerging as a technology that will be implemented largely in software--namely, as features supported by Microsoft's DirectX application-programming interfaces (API).

True, a couple of companies--Fujitsu and Trident Microsystems --are hard at work on Talisman chips. But the more common tack is that being taken by graphics-chip vendors like S3 Inc. and ATI, which will support Talisman not by building Talisman chips, but by constructing more generic accelerators that also support DirectX.

In real-world terms, programmers who want to take advantage of Talisman will do so by accessing Dir ectX. More specifically, they'll be able to use features like anti-aliasing and texture-filtering that have been folded into the Direct3D component of DirectX.

Since Microsoft has been rather slow to clue the public in on Talisman's software shift, I put in a call to Jay Torborg, the company's director of graphics and multimedia, who graciously fielded my questions. For one, he hotly disputed my characterization of Talisman as a non-starter. But rather than respin Torborg's answers, I'm going to go to the videotape so you can judge for yourself.

Wintel Watch: "The buzz is that Talisman isn't hardware-driven anymore. Instead, word on the street is that you're going to see its graphics functions implemented via Microsoft's DirectX software application-programming interfaces."

Torborg: "The intent all along was to integrate the technology into our mainstream APIs. Talisman was never a product development--it was a technology initiative to help the industry move forward with some advanced technologies more rapidly than they might have otherwise. It was an opportunity for Microsoft to take a leadership role in driving the direction [of] hardware accelerators. Our real goal is to create these technologies and then integrate them into our mainstream APIs."

WW: "Is that a sophisticated way of saying that the ATI and S3 parts aren't dedicated Talisman accelerators--they just support its features through the DirectX APIs?"

Torborg: "That's the way that we expect all of them to be, yes."

WW: "Is Talisman dead?"

Torborg: Talisman is far from dead. From an independent software-vendor perspective, it is and will be embodied in our DirectX technologies. From a hardware perspective, we continue to work with the industry, and we've licensed it to the top eight or 10 graphics-chip companies. And they're using many of these ideas in their forthcoming products. DirectX 5.0 provides support for quite a few of the features that we developed for Talisma n, including anisotropic texture filtering, chunking, anti-aliasing, and range-based fog. The rest of the features will be exposed through DirectX 6.0.

WW: "If you had to do it over again, would you announce Talisman in the same way you did last year at Siggraph '96, as something that was presented largely as a hardware-centric initiative?"

Torborg: "I think we'd probably do it a little differently. When we announced Talisman at Siggraph, we focused on the feature that was the most unique, which was its image-based rendering mechanism. A lot of the industry got the perception that that's what Talisman was all about. It's very interesting, but it's also the thing that's the hardest to take advantage of in software. So I think we made a mistake by focusing on that feature, because it tended to make the industry believe that this was some kind of fringe effort--not something that was part of our mainstream plans."

We ll, you gotta give Microsoft credit for trying. After all, this is a company that's made big bucks by throwing form-wizards into word processors and calling it an upgrade. With Talisman, they actually floated some worthy and highly ambitious technology. Unfortunately, Microsoft bit off more than the PC industry as a whole could chew. This might come as a shock to Redmond bashers, but Microsoft wasn't wrong--they were just a bit ahead of the curve. Let's hope they learn from the experience and make us proud next time 'round.

Free Subscription to EE Times First Name Last Name Company Name Title Email address   Click here for your Free Subscription to EETimes Europe