Wow, this turned out to be a book. Hopefully it even posts correctly.
A few years ago I asked Tim why there wasn't a PAL VT and he gave me a long story that I'll do my best to paraphrase.
The card shouldn't have worked in the first place. It was designed based on observation of what the components actually did in the real world, not the published specs. I'm not sure I have permission to go into details, but if you look at the spec sheets for the components, the Toaster isn't possible and a simple overclock won't fix it.
Also, back then the tools to make gate arrays were absolutely terrible, nothing like what we have now and the gate arrays were one of the keys to the whole thing working at all.
There weren't affordable tools to simulate the timing with the precision needed to handle the video signals and there weren't high-level languages to program them with anything non-trivial. Tim spent years hand tweaking every gate by hand to get the timing right and it was only right for NTSC. PAL would effectively be a complete rewrite and he never wanted to go through that ordeal again.
Personally I also think that he had proven it was possible, so the drive needed to expend that much effort just wasn't there anymore. From what I've seen, he's not a money guy, he's genuinely motivated by challenges and curiosity. Doing it again for PAL wouldn't be interesting at all.
As for "why Amiga?", other than everyone at NewTek being Amiga fanatics, it really was the only tool for the job. Years later, others were doing a subset of what the Toaster could do, but not the whole package in realtime and not with anything affordable. The Amiga chipset was incredibly flexible and more importantly, genlockable. The Toaster was designed around the Amiga to take advantage of it's strengths and even quirks. I was literally inseparable.
Truevision, etc had to essentially make their own "custom chipset" and add that to their cards too, driving up the R&D, testing and final shipping costs. The Amiga chipset was already stable, well documented and essentially free with the host computer. We could focus on what we did best. Without the chipset pushing animation data to the Toaster, who knows when a comparable card would ever be available.
We stuck with Amigas for years for much the same reason. The alternatives weren't up to the task. The other was that we not only loved the platform, but we already had a huge investment in software and hardware for it.
Somewhere just before the Flyer was about to ship, it was clear Commodore as we knew it was screwed. We even tried buying Commodore ourselves, but they were still priced crazily out of our reach.
At that point we started testing alternative platforms and even designed a sort of Draco-esque Amiga subset computer with our own custom Amiga-like OS. They had real CIAs, but other than that, all in house and off the shelf parts. If you've seen the tethered "PC Toaster" boxes with the video screen on the front panel that we showed at a few trade shows around 1995 or so, well, that's what's inside the box, a tiny passive backplane, a Flyer card in a Zorro slot and our own 68k(can't remember the model) computer on a card. The problem was that it wasn't better than the 4000 versions, so we scrapped it. I think I have the only surviving systems in my office, one of which still works.
Parallel to that we were writing theoretical code, counting clocks and running the numbers based on Moore's Law to see when it would be possible to do it on software. The hardware side of that project went on to become FrameFactory in about 1998, a basic uncompressed IO board that could bring in some much needed income since Amigas were few and far between.
It took a few more years for PCs to become fast enough, but a couple of years after that you could see us doing demos of what would eventually become VTNT. In the demos we were using dual 500MHz systems (the fastest we could get) and doing some pretty high latency examples of Toaster-like video.
It was almost 2002 before computer speed was adequate and the software was finally ready. When we released VT2, we finally a viable replacement.
Could we have designed a custom chip to do it earlier? Maybe, but at some point you have to pick a path and follow it to the end. Those years hurt, but the basics of the Frame Factory card and VT2 software carried on through the entire standard definition TriCaster line, which is an incredible lifespan by any measure and what we learned there is very much a part of what makes the current high-def TriCasterXD and 3Play lines possible.
