The Gadget We Miss: The Video Toaster The gadget that revolutionized TV in the 1990s.

[Sean Cunningham]

Commodore really screwed the pooch here.  This was, finally, their "killer app", one that compels people, often irrationally, to buy it and buy whatever computer is needed to run it.  Like so many other "killer apps" you have a lot of purchases being made by people who have no real use for it, who won't really use it much at all much less to its fullest, but this is ultimately the goal and a practically necessary component for the economic health of media-based developers.  

You need a large number of people with no business buying your product to, in fact, buy your product so that the people who really use it can enjoy it at a lower cost and with the security that their needs will be supported long term.  This has a "trickle down" effect with related and even non-related titles.  Folks buying Amigas with the idea they might buy a Toaster, folks buying other software to do other things on the Amiga+Toaster they just bought, etc.

[Sean Cunningham]

...
Maybe NewTek was making so much money in North America at the time, they did not worry about making a PAL version, or maybe they were already thinking of moving away from the Amiga due to Commodore's troubles, and they did not want to spend any development resources on anything except moving to the PC.  I have never seen a definite answer why NewTek did not make a PAL compatible Toaster board...

Well, they were making money hand over fist.  I recall a picture of the two founder's matching Ferrari and private jets.  They were a pretty big deal in Kansas before they moved.  The margins must have been insane on those things.  They were living like rock stars.  It was pretty easy to see how the money changed Allen Hastings too.

Given that the Amiga host processor didn't do any heavy lifting at all, no video data passed into the Amiga, the DVE effects were all done by custom chips on the actual VT board, I still don't understand why the folklore at least was that it was intrinsically tied to anything that the Amiga was actually doing besides supplying that NTSC compatible clock signal.  

Seems like making a PAL version or a version that worked in a PC/Mac with an onboard sync generator as an additional component would have been a no brainer.  TruVision had been making pro-level NTSC gear for PC and Mac for years, just not with realtime DVE and switching, which was Grass Valley territory.  NewTek was just loyal, I guess.  It wouldn't be until 1999 that they released their first non-Amiga Toaster hardware if I'm not mistaken.

[Sean Cunningham]

Yeah, I'd like to give Commodore the benefit of the doubt as an engineering-centric company, which means they're predisposed to screwing themselves eventually if they don't get a reality check every once in a while.  They get myopic and quite often don't really understand (or even value) the true end user experience with their creation or its weaknesses.  That's like a cliche situation by now in the tech business.

I said in another thread, the situation with Toasters in 1994 and 1995 after all the Amiga 2000s and 4000s that were ever going to be produced had been produced was Toaster resellers were paying top dollar for dead A4000s (and dying in the first year was quite common, sadly, it happened to mine).  One in LA was offering up to $1500 for an A4000 corpse, so that they could Frankenstein a few of these together into a working VT4000 workstation.  

It was crazy.  The demand was that high and there was just nothing else out there that was comparable for switching and DVE and transitions.  I just never understood why this capability was suddenly so desirable now that it was more affordable than it had ever been.  It was like the device created this need in people to figure out how they were going to use it rather than it being a solution to an outstanding need that wasn't being filled.  It wasn't like all the other hardware you needed to actually make good video was suddenly cheaper.  

The Toaster itself wasn't a panacea for prosumer video which needed to be TBC'd and was pushing a different spec for I/O.  Hi-8mm, ED and Super-Beta and S-Video were ideally S-Video devices but the Toaster was composite only.  Meanwhile, above the industrial threshold professional production was shifting to component for folks still working in analog and serial digital at the high end.  The Toaster was therefore best matched to U-Matic (3/4") which was still a very ubiquitous format in actual TV production, college RTV departments and industrial production but well outside the budget of your average wedding shooter, with a single deck costing as much as an A4000 for a cheap, play-only model.  NewTek reps were very coy and deflective when you started talking to them about whether or not they were going to offer additional models offering S-Video or component I/O.  They'd supposedly designed it to be "D2 internal" which was an odd choice but no way of going in or out but composite analog.

By mid to late 1990s it got so you could recognize when a Toaster was being used on a program or video, whether or not they used the "Kiki Wipes" or those kitschy transitions with falling sheep and wiper blades, etc.  What's sad is all this time, I'm looking at contemporary news programs in HD where, today, a lot of the character generator and lower-thirds and broadcast graphics being done aren't as good as what all the little cottage publishers were making to sell to Toaster users back then.

[Sean Cunningham]

A lot of this I don't buy.  Someone who wasn't "a money guy" wouldn't have been so nouveau riche rap star "baller" about showing off Italian sports cars and jet planes.  They had enough money to hire engineers to do the conversion and, if necessary, re-brute force the development of a PAL compatible version.  They didn't.  The rest seems like excuses.  By his own description, his development process was more about being persistent than clever.  That's actually the sort of thing that's easier for someone else to replicate, quicker.

The only thing the "chipset" actually provided was the genlockable character generation and scrolling text.  It didn't do the DVE effects to the video itself.  Amiga titling was generally better than the options available from PC, Mac and IIGS solutions.  Of course that's not insignificant but even at double the price for a non-Amiga solution you'd still be looking at a significantly lower cost than just a professional, standalone character generator and those folks were used to paying more for their toys in any case.  

A Truevision AT/Vista board was $3795 and that was mostly a paintbox and simple display buffer solution without the ability to process live video or playback the kind of animation that you got with the Toaster4000.  It was a much better paintbox for sure than anything on the Amiga but that's just to illustrate the kind of pricing NewTek was competing with for much narrower focused devices.  Double their Amiga prices would have still been a bargain.  There wasn't anything "impossible" going on, as this was always religious hyperbole and folklore.  The real question in the context of non-Amiga solutions was whether or not it could be done at the same price as an Amiga solution.

The info later in the post, talking about waiting for processors to get faster, etc. is obviously related more to being able to process the DVE effects on the host computer, take the video in, transform it, animate it, and then push it out.  The Amiga VT didn't do that, there was an onboard DVE processor that did this, crudely in the first version and with a bit more polish in the VT4000 (ie. antialiasing).  Having a software based DVE where the board itself is just handling the high speed video i/o from multiple sources is an entirely different beast than what the VT was and there's never been an Amiga that could have handled that sort of business with even standard definition video, I'm sorry.

The real story is they were loyal to the Amiga, beyond it's death until they had no alternative for continued survival.

[Sean Cunningham]

Problem is, this 'killer app' was 1) too expnsive for casual users and 2) too niche to effect sales enough.

Photoshop was not an inexpensive app.  Neither were any of the page layout tools that kept Macs relevant, none of which directly brought any dollars into Apple either.  All of these very niche as well, yet they led to not only more people buying software they didn't actually need (this is the lifeblood of the industry overall) and machine purchases that allowed for the "one day" upward compatibility.  The killer app is the tide that raises all boats.  Because it's there, the hype surrounding it, all sorts of other smaller developers benefit.  The killer app creates another tier of supporting apps.  Commodore benefits from all that activity if they seize on it and support it and they didn't.

There was no existence in desktop publishing without spending thousands, in addition to whatever host computer you were using, and generally none of this additional expenditure went into the pocket of whomever made this computer.

Killer apps aren't about practical accessibility, or at least not exclusively so.  They're about selling an idea and creating the feeling that you "needed" it.  Since the dawn of the personal computer age most of this is irrational consumerism.  The Toater is hardly the most expensive item people have lined up around the block to buy for their computers that they don't, ultimately, need.

[Sean Cunningham]

A
That's not actually correct.  Watch the control monitor of Toaster system when doing a transition.  The animated effects are done by the Amiga, you just can't see them well because they are encoded with extra data that doesn't look correct on that monitor, somewhat similar to a DCTV image.  It's a combination of overlay, matting and UV coordinate information that the Toaster card knows how to interpret. The gate arrays on the Toaster card then use that data to mix the video signals and potentially the decoded Amiga graphics.

Okay, but that's very sparse data.  It's not heavy lifting at all.  It's very clever, allowing new effects to be added without the need of firmware updates to the board.  I would put money on this actually being less work than a full page of Kara Fonts characters scrolled from the bottom of the screen to the top.  Some polygons and a few "magic bits" is hardly "Only Amiga Makes It Possible" material.

The Amiga Toaster had chips that processed video, but the mix was driven by the Amiga.  The video didn't need to get to the host computer unless you grabbed a still, which is a very slow process on a Toaster.

You're using "mix" here as if it was actual work or some kind of special engineering.  The video was processed on the board and sparse data sent from the Amiga instructed the board.  It's not magic.  If I'm watching Netflix streaming over a Chromecast device yet I'm controlling the Chromecast with an app on my phone I don't attribute any of the streaming or display or decompression, etc. to that mobile device in my hand.  

That's a very simple analogy, simpler in many ways compared to what is happening in the Amiga to Toaster connection but there is nothing stopping an extension of this very sort of one-way relationship of control such that full scrubbing and editing and higher level application control is established in such a paradigm and no matter what I would never attribute any of the real work to the handheld device that's ultimately controlling and "driving the mix".

Going with smarter, more specialized hardware would also limit the lifespan and flexibility of the design.  Those cards only became outdated when HD video became the norm.

Mmmmm, in local TV circles and the like, perhaps, which stay with hardware long passed its sell-by date.  The Toaster became anachronistic with just a notch above the the more common industrial level (limping along on U-Matic) not long after the release of the VT4000.  Because composite video.  It would be the late 1990s but well before HD when the VT and its composite video became a real problem at the prosumer level.  Because IEEE 1394, otherwise known as Firewire.  

As soon as video acquisition and recording went digital the Toaster was pretty much public access television material.  But by 1993 even, component video devices were making their way into prosumer hands even, via devices like the Personal Animation Recorder and later Perception Video Recorder.  These rendered the framebuffer and video output on the Toaster a null and void issue for pretty much any application not involving the switching or processing of live video.