gekk.info « articles

Toasting Video: Windows Style

Warning, and I'll keep it simple: Take everything in this article as a mix of speculation, guesswork and wishful thinking. These are the facts to the best of my knowledge, and my best ain't great.

If you know about this stuff and find any factual errors, please email me. I have done my best and info on these topics is hard to find; rude emails will be deleted.

The NewTek Video Toaster is generally known as an addon card for Amiga computers that was extremely popular in video production in the early 90s. I have learned that there are other Toasters which ran on Windows PCs. This is a repository of info I've gathered on them. The lineup included:

Newtek dropped the actual name "Video Toaster" after [2], but the software is all very similar.

This page is meant to accumulate all the info I have about the software and hardware of these systems, as well as their descendents, the Tricaster series.


Please help: Do you have VT[2] software, or any of the PCI cards? Please get in touch! articles@gekk.info

Changes

Video

I have now created a Youtube video covering some parts of this, and demoing the VT[5].

History

Video Toaster: Origins

The NewTek Video Toaster is probably the most memefied piece of equipment in the history of professional video, despite being a piece of TV studio gear for which very few people have an actual use, and which has been surpassed in every conceivable way, decades ago. We can probably assign blame for this to NewTek's unparalleled cross-market advertising campaign, completely unprecedented among video gear, for reasons that are out of scope here - and, well, the Kiki transitions, which are the nerd equivalent of the star wipe.

At any rate, all the cultural and technological factors that led to the Toaster becoming so well known were specific to a technocultural atmosphere that existed only briefly, in the late 80s and early 90s - and it only entered the market in 1990, so it didn't have much of a run. I would say that by 1995, the Toaster was no longer culturally relevant, insofar as it ever really was.

The computing platform it depended on - the Commodore Amiga - was also effectively dead in the market by that time, and it, too, commands powerful memetic status among the computer-familiar. So, it's no surprise that the name Video Toaster is synonymous with the Amiga-based video effects systems sold in the early 90s. You can spend several hours researching on Google if you like, and this is virtually the only thing you'll find mentioned.

However, NewTek did not turn into a pumpkin after the release of the Video Toaster 4000, nor did they remain anchored to Commodore's sinking ship. I have heard it said - though I can't corroborate - that NewTek hadn't even wanted to use the Amiga in the first place, but saw it as the only economical option, due to very unusual (and not particularly market-conscious) features of the platform's hardware design.

The multimedia capabilities of the PC platform skyrocketed in the early to mid 90s however, and by the end of the decade, what had once required specialized silicon became possible on ordinary Wintel hardware. Windows gained full motion, full resolution video support; general purpose graphics acceleration hardware entered the market; and NT married an extremely stable (for the time) kernel and multitasking environment to the feature-rich Windows GUI.

So it comes as no surprise that when NewTek was ready to move on from their 1990-era masterpiece to a new product, they based it on the Windows platform. It was released unceremoniously, without the massive marketing push that accompanied the original product, and it carried the exact same name: Video Toaster, although it is sometimes referred to as the Video Toaster NT.

Even if they had pushed hard with advertising though, by the time the VTNT came out in 1999 the public probably couldn't get too excited. Less than a decade earlier, the ability to create arbitrary, complex video effects on a normal consumer PC simply didn't exist, and nonlinear editing was a distant dream for most editors working outside of Hollywood. Video mixers and effects units were essential for producing all sorts of video projects, like commercials, internal corporate training tapes, even wedding videos. The mixers and titlers you could buy for under $10,000 were not impressive, and the Toaster wiped the floor with them.

With the advent of consumer-priced nonlinear editing in the late 90s however, most of what people were doing with those systems no longer needed to happen in realtime. Adobe Premiere and iMovie made video editing something you could do on an ordinary PC for about $600, with absolutely no special hardware. That rendered the Toaster largely useless for anyone not doing live production.

This is probably why the original Toaster got so much fanfare, but Newtek's following products are almost unknown among the general public. Despite this loss of popular interest however, they essentially never stopped making Video Toasters - even to the present day, depending on how you look at it.

NewTek produced the original Amiga Toasters until the end of the 90s (as far as I know,) then from '99 onwards they made various Windows-based Toasters. Sometime in the 2000s, the brand was retired and replaced with the Tricaster. All of these products are closely related.

The Tricaster was a departure, because it changed NewTek from a software/peripheral vendor to a system vendor. Tricasters were (and are) complete appliances, integrating a PC and NewTek's custom input/output hardware into a single chassis.

Everything predating the Tricaster, however, was simply a component you added to a PC, and NewTek was actually very proud of this fact: all the Toasters (excluding the Amiga original) worked entirely in the software domain, requiring no specialized hardware except for a capture and output card.

I found a newspaper article from the late 90s that said they sold 30,000 of the Amiga Toasters, which is not an enormous number, yet you can find them readily on eBay. I ascribe this entirely to their memetic nature: while most video processing gear gets thrown out once it's obsolete, when people see the Toaster they go, "oh my god, it's that thing," and thus Toasters get preserved.

The Windows-based Toasters, however, do not seem to get this treatment. I'm sure NewTek sold several times as many, simply due to how long they were on the market, yet virtually none show up used, I can find no websites with info about them other than the mid-2000s equivalent of trade mags, and most nerds (who will typically go nuts at the mention of the Amiga product) are unaware they exist.

It doesn't help that, unlike the Amiga models, the Toaster cards made for the PC are almost completely unlabeled. You would be hard pressed to even know that you have a PC Toaster card in your hands without a picture to compare to, and there is no googlable model number. So, in the hopes that these will eventually get better preserved if more info is available, I am writing down what I know.

What Is The Video Toaster?

First, let me summarize the product everyone is familiar with:

Video Toaster / Video Toaster 4000 - 1991

An addin card for an Amiga which replaces several pieces of TV studio equipment: a switcher and/or vision mixer, a character generator, an on-screen graphics workstation, a 3D graphics workstation, and (with addon hardware) a videotape recorder.

Features:

The Video Flyer was a common addon: a separate board which provides hard-drive based video playback, using an array of external SCSI disks, and a non-linear editor, for offline video editing.

The Toaster came out in late 1990/early 1991, and is generally regarded as a revolutionary product that replaced lots of expensive studio gear. The Toaster 4000 came out a couple years later. There was was a gap of at least five years, before the next thing called a Video Toaster was released in 1999, and it looked like this:

Video Toaster NT - 1999

A powerful postproduction suite for the time. Capable of nonlinear editing and 2D and 3D graphic creation, but with no live production features.

Features:

If it's not abundantly clear, the name "Video Toaster" does disservice to the original product of that name. They simply aren't the same thing at all: Aura is not ToasterPaint, there is no equivalent of ToasterCG, and Speed Razor is a third party product unrelated to the Flyer editing software. Literally the only thing brought over is Lightwave, which had been available as standalone PC product for years anyway.

I believe it would have been more accurate to call this product the Flyer NT, and in that context - a video capture/output and NLE - it really is very good, and would have done service to the Flyer name, but a Video Toaster, it's not.

Some people were mad about this, but others appreciated it and the package did sell. A couple years later however, NewTek released a proper update: The Video Toaster [2]. Here are the basics:

Video Toaster [2] - 2001

A return to form, providing everything the Amiga Toaster could do and more.

* These apps are all tightly integrated and run as a single program.

Note: Several of the above features require a breakout box, but ostensibly the product was never sold without it, so these can be considered intrinsic capabilities.

So, in other words: The Video Toaster [2] is simply the evolution of the Amiga Video Toaster to the PC platform - which is exactly what you'd expect.

It seems pretty unsurprising that NewTek - whose greatest accomplishment to date had been the Amiga Toaster - would have simply updated it to work on the PC, and that's exactly what they did... but only after releasing some strange, in-between products, which I'll discuss more below.

In 2003, NewTek released the VT[3]. At this point they stopped using the name "Video Toaster," but as far as I can tell it was still the same product with various enhancements. Likewise, after that came the VT[4] and [5], which again seem to be largely the same thing with various enhancements, but using a different PCI card.

VT[3] / VT[4] / VT[5] - 2003-2007?

As you can see, I can't speak to all the differences in the later revisions, but from what I can tell they were pretty feature-complete by VT[2]. Based on the manual it seems like it has virtually everything that VT[5] did, and that's far more capability than the original Amiga Toaster. So, we can be assured that the Toaster product line was not exclusively an early-90s phenomenon, but in fact a through-line from 1990 to the late 2000s.

To get picky: I don't know if Newtek was still selling new Amiga-based toasters after Commodore went out of business in '94. I feel like they would have stopped making new hardware given that they could no longer obtain computers to sell it with. If so, then the Toaster product arguably did not exist between '94 and '99, when the VTNT finally came out - and if you only want to count full-featured equivalents, then really, the Toaster was sold from 1990-1994, and then from 2001 onwards.

The end date on this lineage is a little vague. The original Toaster may have died at the same time as Commodore (assuming that's true at all,) but the end of the PC VT series is less definite.

Newtek's flagship product in the early-mid 2000s was the VT series, but in the late 2000s they pivoted to the Tricaster, a complete video switching appliance. I initially believed that the Tricaster was intended to succeed the VT, but that does not appear to be true.

While the first Tricaster has been described to me as "just a VT[5] in a prebuilt PC," that's not quite accurate. The hardware is identical, and the switching software is clearly derived from VT, but the Tricaster was not designed as a replacement and it offered far fewer features.

While the Tricaster contains substantial Toaster DNA, and there's a lot of overlap between them, "Video Toaster", as a product identity, described a combined mixing, editing, and art creation package; the Tricaster is missing a good third, as explained further in the section at the bottom of this page.

While I think it's fun to say that Newtek "still sells Toasters" (since the Tricaster is still an active product) it would be valid to argue that the last actual Toaster package was sold in the early 2010s - making the lifespan of the series roughly twenty years.

Despite these splittable hairs, though, this all seems straightforward enough. Newtek has made the same product over and over since their inception, adding or removing features to fit the state of the market, like any company... except for one anomaly: the original VTNT. It's a strange speedbump in the lineage, and in fact, I believe we can infer that it was never supposed to exist.

The (possible) VTNT Story (that I made up)

In a nutshell, I strongly suspect that Video Toaster [2] is the product NewTek wanted to sell as the followup to the Amiga Toaster, and had probably hoped to make available as early as 1998, but circumstances conspired to force them to release a subpar product (actually two) first.

The critical thing to understand is that the VT[2] and VT[3] use the exact same card as the VTNT.* That means that, despite the NT having no live switching capabilities, the card it shipped with did, and that makes no sense unless NewTek intended, from the get go, to make VTNT a live switching system.

*Important note: If you are trying to set up a VTNT card, don't proceed until you read the "Core Updates" section below or you could brick it.

It gets even more intriguing when we go back a little further. Before the VTNT, there was another product called the Frame Factory. Very little info about this exists; virtually everything I know comes from old Usenet posts and partially archived distributor websites, but it sure sounds like it was literally just the VTNT, minus the Speed Razor NLE.

Once again, Frame Factory used the same card as VTNT, VT[2] and VT[3]. When the VTNT was released, NewTek sold upgrades to customers for a few hundred bucks, consisting only of software and an FPGA core update; it's not clear to me what either of these did, other than minor tweaks, since the capability list seems to be identical (except for the inclusion of Speed Razor.)

This probably explains why my VTNT box has an "UPGRADE" sticker on it. That had confused me for a while - I thought Newtek had made the card for the VTNT, so what could it have been an upgrade from? I now imagine this card was originally sold as a Frame Factory, and probably every one of those in existence got "transformed" into a Video Toaster and the old box thrown out.

Now, the Frame Factory trademark was only filed in 1998, and the VTNT was being sold in 99, so it seems like there was a very short delta between the two products - and with them being so similar, I strongly suspect the timeline went something like this:

I want to stress that I am really reading between the lines here. I can't prove most of this; it's a combination of vibes from Usenet and forum posts, mostly-broken archived copies of the Newtek and Safe Harbor (a system builder who ran the official fansite) websites, and so on. But it also just seems like the only explanation for the VTNTs existence - why would they have even made something with this name if they didn't intend for it to replace the original product of that name 1:1?

Further supporting this is the condition of the documentation. My VTNT came with the original manuals, and while the books for Speed Razor, Aura and Lightwave are nice, high-quality, commercially printed and bound tomes (Lightwave even has two thick manuals) the book for the Toaster itself is an absolute joke. Perhaps 20 pages, spiral-bound, with the thinnest possible skimming-over of the system features. It is, to be completely frank, an embarrassing effort.

For my money, I'd bet NewTek had already produced a beautiful, high quality manual for the complete Toaster system that they hoped to sell, then found serious problems with the switcher product after announcing it. panicked when they realized it wasn't ready, quickly whipped up this pamphlet, and released the whole mess to avoid further embarassment.

I think that when VT[2] came out, NewTek breathed a collective sigh of relief as an increasingly frustrating and stressful period in their history finally drew to a close. I imagine they were sweating bullets for at least two, perhaps four years, as they watched their early victory become increasingly ancient at the same time as their next product slipped further and further into the unknown future.

From what I've read, at least some customers got upgrades to VT[2] for free, since it only required a new piece of software and an FPGA update - but knowing how small/medium businesses operate (especially when they feel embarrassed) that probably ended up being basically everyone. Probably an awful lot of people bought upgrades too, so I'd guess there are very few plain "video toaster NT" systems remaining in the wild.

The above is my closest approximation of a history on these products. The following sections will address more specific details of the hardware and software. Then, if you're curious where NewTek is today, you can check out the section at the end about their modern product line, the Tricaster.

Hardware

There are two Video Toaster cards for the PC platform. Here is some info applicable to both of them.

Copy protection

The VT cards act as a copy protection "dongle" for all the software included with the various Toasters. None of it will run if the card isn't detected, and while that's obviously true for the Switcher (VT[2] and later,) much of the rest of the software is actually quite useful without it. That includes Aura and Lightwave (both of which were actually sold as standalone products with their own dongles) and the Toaster video editor.

The software included with VT[2] and onwards, and all the Tricasters, must be activated with Newtek support to be used longer than a short trial period. There are ways to bypass this out there, but I am told that if you call Newtek support, they will usually transfer ownership to you, even for older products. YMMV

Software Compatibility

FPGA core update warning Make sure you read this section to understand the possible risks of mixing and matching software and hardware versions.

The two VT cards were originally sold with different versions of the software, but I have been told (though I haven't been able to verify) that they actually have great intercompatibility. Supposedly, you can use an original VTNT card with VT[5], and possibly even vice versa.

That's not to say all the cards work the same way though. Each version of the VT (and Tricaster) software includes an FPGA core updater - in other words, a firmware upgrade. If your card firmware doesn't match the software version, it will try to install the update every single time you start it.

I urge you to use great caution. Based on forum posts, there is an extremely specific procedure that needs to be followed, including powering your machine down for several minutes as soon as the update is complete. Users report that failing to follow this can brick the device - and it can still brick randomly even if you do.

It's your call, but personally I am pretty skittish about trying this! In 2004, if you bricked your card, you could still contact NewTek to have the thing resurrected. I doubt that's still the case, so my advice is to avoid updating the core.

I have used a Tricaster card with VT4 and 5 software without updating the core. It crashed when I did one or two things, but I didn't need those things, so it wasn't really a big deal. Everything else seemed to work. So you can just hit "no" every time it asks you to update the core, and if the card works for your purposes, you don't need to upgrade it. I renamed the updater (VTCoreUpdater.exe) to make it impossible for it to run.

Consequently, I don't know if you can ever downgrade a card. It's possible that my TC card *can't* ever be converted back into a "VT5" card - and, if I were to upgrade my original VTNT card for use with [2], maybe VTNT would never work with it again. This is why I'm trying to find a second card so I can keep that one intact.

But if I did want to upgrade it, apparently the sky's the limit. I have found two people who claimed to be selling original VT cards that they had been using with the VT[4] or [5] software, and I've been told this should work, it would just have fewer features (I don't know which.)

VTNT, VT[2], VT[3] card

Connections

If you only need to input and output a single video signal, you can use common "VGA to BNC" cables. You can find them easily, and they usually break out to red, green, blue, white and black cables. The composite video line will be the white wire.

A note on testing the card:

The in and out ports are wired identically, so you can test both with the same cable. To test in/out at once, plug a plain VGA cable between the ports for a loopback, then use ToasterVision In to monitor the input while you output something. If all's well, you should see the same input and output. It's possible there are other signals on those ports that are not meant to be connected together that could eventually cause damage, but I don't think it's likely.

By itself, the VT card can only do one video channel in and out, so if you have the VT[2] or later software, you can take input from one camera (or other device) then perform a transition to a video playing in a DDR, or a title card from the CG tool - but you can't transition from one camera to a second one, because there is nowhere to plug it in. For that, you need an SX-8 or SX-84 breakout box (BoB), described below.

The SDI option board takes up a second PCI slot, and requires a special "bridge" board to connect to the card.

If you want to hook up a BoB, you will need a separate backplane bracket that adds two more "VGA" ports (it hooks up with four VGA cables in total; yes, ordinary-pinout VGA - with a catch though, read the BoB section for details.) You will also need a special bridge board. If you want SDI and BoB, you need a third "combo" bridge board.

Details

There is no specific name for this card, and it's completely unlabeled. There is absolutely nothing silkscreened on the board, not even an internal part number. I am calling it the VTNT card, although it also works with Frame Factory, VT[2] and VT[3].

I will make this point in more detail down in the Software section, but: The card is really, really simple. Many video editing/processing/switching systems use lots of custom hardware to apply video effects, but the VTNT card is nothing more than a glorified video capture device.

I do not know all the details of how the card works, but I have combined my speculation with the info in the intro to the manual (which I will endeavor to scan when possible.)

While inexpensive, high quality capture cards were readily available on the market in 1999, probably none of them could move video data to and from system memory without CPU involvement. The PCI Toaster uses a bus-mastering approach - basically, the card itself can access system RAM independent of the CPU. That allows it to send captured frames directly into RAM, and read frames from RAM for output, with no impact on system resources, and that allows the software to use 100% of the CPU for video processing.

I'm curious why it has an SDRAM stick in a normal slot. I mean, I'm not surprised that it has RAM, since it needs somewhere to buffer frames in case the PCI bus is busy, but why is it a discrete stick instead of a couple chips soldered to the board? My guess: The supply chain was unpredictable and they were having to stuff theRAM in the board at the last second during manufacturing, using whatever they could get. Or maybe chips were failing a lot and they needed it to be replaceable.

VT[4], [5], Tricaster SD card (VT Pro)

This card was used with the last two VTs and the first generation tricasters - you can read more about those at the bottom of the page. Newtek called this card the VT Pro. I've seen it silkscreened both "MDC-G4" and "MDC-E"; they should be interchangeable.

This card is still PCI, but is capable of running at 66MHz instead of 33MHz. It can do three concurrent inputs and two concurrent outputs, which is necessary for several purposes:

There are several Newtek forum threads about this card not working in an older P3 board, which turned out to be due to the board not providing one voltage rail or the other; I did not have this problem with my P3 board, and it's unclear from the thread which voltage was problematic.

You will want to use a machine with 66MHz PCI, or you will have problems when attempting to enable the Preview (or Alpha) output. In my case, it produced flickering for a few seconds, then crashed the drivers. 66MHz PCI is nearly synonymous with 64-bit PCI, AKA PCI-X, which does not appear on any consumer motherboard. In short, you need a server or workstation board, usually from the P4 era, with the "long PCI slots". All other features seem to work fine on normal 33MHz slots.

You should also to set your PCI latency to 128. I have no idea what this means, but people insist it's critical, and can sometimes cause the above problem.

There is a power connector on the card which accepts a normal floppy power plug. Without this the card will work halfway and crash randomly, so make sure it's connected at all times.

The ports on the back are the same as the original VT card, and again you can use this with nothing more than a VGA-to-BNC adapter, if you don't need to switch multiple video inputs. You will be able to do all the same things that card could do under those constraints (explained above.) If you want full switcher functionality you will need an SX-8 or SX-84 breakout box (described below) and the backplane bracket that adds the two extra "VGA" plugs in order to connect it.

As before, to connect a BoB you need a breakout board which adds two more "VGA" plugs on your backplane. The BoB uses four VGA cables in total (yes, literal normal-pinout VGA cables - with a catch though, read the BoB section for details.) However, a change from the original card is that instead of connecting to the VT card with a "bridge" board, the new breakout plate connects with a pair of flat blue ribbon cables.

These are actually micro coax. Yes, each of those little blue "wires" is apparently a coaxial cable - it makes sense for analog video, but still, wow. If you got the VT card and breakout without the ribbons... good luck replacing them.

I have been told that the blue cables are VERY fragile and can be damaged by vibration if you knock them around, so be very careful moving these and, if possible, don't have them shipped. If the cable gets wadded up, it might get destroyed.

These cables sure look like Quadrangle SGC jumpers, so it might be possible to find them in non-Newtek context, but I don't have part numbers.

If you only have one ribbon, you can still make a BoB work, you just won't have any audio. On the Tricaster, I believe you can't get away with using less than two ribbons (the one near the backplane, and the one on plug J5.)

Breakout box / BoB / SX-8, SX-84

The VT cards, by themselves, can't handle more than a single input and output signal. To support two inputs (for transition effects) and to allow switching between a large number of inputs, NewTek sold the SX-8 and SX-84 breakout boxes (BoBs.) They were originally sold with the VT[2]/[3] and VT[4]/[5] respectively, but are virtually identical (see compatibility notes further down.)

Adding one of these to any of the VTs (as long as you're using VT[2] or later) gives you a full fledged live vision mixer system, minus one now-common feature: Most modern vision mixers offer a "multiview" in which you can see all possible inputs as thumbnails at all times. Video Toasters do not.

The original VT card, even with the SX-8, can only actually handle two inputs at a time. When you select a Program and Preview input in the software, the BoB switches those into the card's inputs. To monitor inputs that aren't selected, you need to split them externally and send them to discrete monitors.

The VT Pro card has three inputs, but it still can't monitor more than two at once. The sole value of the Pro card is to allow three simultaneous sources to be combined in a LiveSet, which only exists in VT[5]. In VT[4] it serves no function I'm aware of.

The same card is used in the Tricasters, where they can in fact monitor three camera inputs at once, though only in three-camera mode. The six-input models have a 6-camera switching mode which does what it sounds like, but disables the ability to monitor any of the inputs.

Notes on SX-8 / SX-84 compatibility

Cable warning! The SX-8/84 uses "normal VGA cables" - except it turns out most VGA cables aren't "normal" and will damage it. Read the below section on "Cable / troubleshooting" before connecting a BoB!


Brick warning! When you first run diagnostics or start up the VT software after connecting a BoB, it will check the FPGA revision and then upgrade it if need be. You can probably brick the SX by unplugging it during this, so make sure that, once you have the box powered up, you don't unplug it for a good minute to be sure.

According to Newtek (from an employee on the forums):

The SX-84 is a replacement BoB paired with the new hardware, but for customers who already own an SX-8, there are very few differences:

Supposedly, to make an SX-8 do four channel sound when connected to a VT Pro card (I don't think the original VT can do it at all):

Here's what you'll need to do to make it work. First of all, break-off pins 12 and 15 from the VT5 (control) cable. Hook-up the old SX-8 audio cable to the VT Pro card's audio I/O jacks. Hook-up the VT3 (in) and VT4 (out) cables to the VT Pro card as well.

Next, the VT Pro daughtercard. Connect the VT5 (control) cable to the port marked control on the daughtercard. The audio connectors on the daughtercard will be for the "rear" channels. You can monitor the rear channel output from the audio out connector. To get the rear channel input you will need to run a cable from the Aux send of the SX-8 to the audio input on the daughtercard. (a RCA pair male to 1/8" stereo plug)

At this point you should have a functioning SX-8 with 4 channel capability. The odd inputs will be front and the even inputs will be rear just like on the SX-84. Both sets of XLR inputs can even be selected from line/mic and the preview/alpha out works too.

However, the above is not the full story, as will follow.

Cable Problems & Troubleshooting

While the SX-8/84 hook up with "normal VGA cables," it turns out there is disagreement on what constitutes "normal." A proper cable should have separate wires for every ground pin. Many, many cables do not, and simply gang all the grounds together.

As it turns out, one of the pins - #11 on the VT5/Control cable - is a ground for VGA, but for the VT, it carries +10VDC that powers the BoB.
This means that MOST VGA cables will short out the VT card! That is probably as bad as it sounds.

To check a cable: Unplug your VGA cable at both ends. Use a multimeter to look for conductivity between the shield (outer shell) of a connector and pin 11 - you can google a pin diagram, or look real close at the plastic below the pins, it's marked. If you see conductivity, you have an unusable cable. Repeat this test at both ends. Ideally, check also between pin 11 and all the other pins. Nothing should be connected to anything else.

If you don't have a meter, get one, they're cheap. Don't be impatient or you may fry your irreplaceable VT card.

Symptoms:

I can't stress enough that you shouldn't wait for symptoms. A power/ground short can destroy a device rapidly, so check your cable as above, don't just try it. But let's assume your cable is bad in a different way. Do the following:

Please note that if newtekrtme DOES detect the BoB, it may start an FPGA upgrade. You'll be able to tell, it says exactly what it's doing. Don't interrupt it!

Does this problem affect the other cables? I haven't mapped out all the pins (although we do have a pinout provided by Newtek staff) but at a glance, there are no voltage rails on the other three cables, so there is no risk of damage. Any inappropriately grounded pins with signals will be silent/blank, and it looks like that includes audio Ch3, which is also on pin 11.

I have been advised that if you want to buy cables that will definitely pass this test, the keyword is "DDC and EDID compliant VGA cable"

Control surface / RS-8 / LC-11

In the VT era, NewTek offered the RS-8, a vision mixer style console with a fader handle, program/preview buses, and various other buttons.

When the Tricaster came out, they made a larger controller called the LC-11, which does the same thing, but more of it.

Both of these controllers connect via USB, and then simply adjust (and reflect) settings in the switcher interface. The LC-11 works perfectly on VT5 (version 5.2 and later) but I do not believe it's supported on earlier software versions; I also do not know if the RS-8 works on Tricasters.

Software

As you can see, the hardware component is little more than a capture card, and a box for automatically choosing signals to send to it. Indeed, the Video Toasters are intended to be all-software products, from the ground up. In all reality, the hardware probably could have been replaced with a cheaper commodity product - VT is in fact meant to be used on otherwise ordinary commodity PCs, and if you have both the hardware and software, you can probably make it work in whatever machine you have.

The rest of what defines the Toasters is in the software. It's quite a project to describe everything - notionally, reading the manual would be the best way to get it all, and you can get some of those on Newtek's FTP, or at the bottom of this page. Here are some features common to all the VT's (except NT):

I'm pretty sure all of this is available in every version of VT (except NT, which only has lightwave, aura, capture, and a third-party editor.)

Versions

The capabilities of every version from VT[2] to [5] are mostly similar, with some substantial differences, although I'm not sure what all of them were. Newtek did not do a great job of explaining, especially since their promises appear to have shifted and changed over time in the leadup to a release, and in some cases failed to meet up with what was delivered entirely.

This forum thread is a great example: VT[5] was supposed to include support for multiple DSKs, a long-awaited feature, but ultimately only delivered one, and was also delayed heavily so that Newtek could add a feature that many users had no interest in.

Streaming

Beginning with [2], all the Toasters (and Tricasters) support online streaming - which is frankly pretty wild when you consider that Video Toaster [2] came out in 2001. It seems to be primarily based on Windows Media Encoder, although later versions support Flash Encoder.

I've tested this in VT[5], but I suspect there are similar options in all of them. The three options in VT[5] (out of the box) are:

You can only stream the program output, unsurprisingly, bitrate is fixed, framerate is either 15, 29.97 or 30, and resolution is no higher than 480x360 for unknown reasons. There is a significant delay (upwards of five seconds) between stream output and local output.

Screenshots

VT[3], VT[4], Tricaster, Tricaster Pro and Tricaster XD screenshots forthcoming, since I own all of these. VT [2] when possible, if I ever get a hold of it.

Video Toaster NT screenshots

Video Toaster [5] screenshots

PC requirements

Toasters are meant to be used in normal PCs. NewTek was very insistent about this:

NewTek's position was that the Toaster was capable of doing as much as you wanted it to do, and its capabilities all depended on how much hardware you threw at it. They basically insisted that you could use whatever you had laying around, you'd simply need to get better hardware if you tried to do something and found out you didn't have enough oomph.

So, there are no specifications; nothing that says you could do "four concurrent video streams" or "five simultaneous effects" or whatever on a given CPU, you just have to try stuff and see what works. All versions support multiprocessing, so you can give them gobs of gigahertz and cores if you want to pile up more layers of video and more effects.

Conversely, however, NewTek made it very explicit that the Toaster didn't require you to build a Monster Rig. They explicitly stated in their manuals that if you were sure your requirements were minimal, you could safely go under the minimum requirements, and the software would try to accommodate you with lower quality previews, etc. They also stated, in the runup to the VT[2]'s release circa mid 2001, that they were still targeting the Pentium 3 despite the significant benefits of increased memory bandwidth on the P4. So they were pretty dedicated.

Very rough system recs:

Video Toaster NT

The lowest set of requirements I've seen listed for the original Video Toaster NT was a Pentium II 266, 128MB of RAM and Windows NT 4. I am pressing X to doubt this. Maybe it technically worked, but a 266 is pretty sluggish even running Windows 98. Sure, it might have been a damn sight quicker than the Amiga version, but I doubt it was really very practical, and the software version that came with my VTNT includes a readme that calls for a P3 450 - meaning that by the time the disc made it to presses, Newtek had changed their mind on that P2. And remember, this version can't even handle more than one stream at once.

VTNT requires Windows NT 4 (SP3 or 4 minimum, I think) but I don't think you should waste your time with it unless you're a real NT4 wizard. I spent two days trying to make it go and hated every second of it. The software runs like a champ on Windows 2000 SP6, and that's far more worth your time. It will not run on XP, it just crashes. There were several patches that might have fixed this but all are long lost to time; trust me, I tried.

Video Toaster [2] / VT[3]

VT[2] (and as far as I know VT[3]) requires a P3 at minimum, because it needs SSE. Later versions of the docs recommended a P4. It also requires Windows 2000 at minimum; I believe I've seen people say they used XP.

VT[4] / VT[5]

VT[4] and [5] require a P4 at minimum, because they depends on SSE2. They also want Windows XP or later.

I have been told that VT[5] (at least) will run on everything up to Windows 10; it is very much just a program, so if you're more interested in the device itself than having a Vintage Experience, I highly recommend just doing that.

Graphics/Sound

Besides the SSE or SSE2 dependencies of the various versions, there is only one special hardware requirement for all PC Toasters: If you want a live picture on your desktop of what the Toaster is inputting or outputting, you need a Direct3D capable GPU..

This is completely optional - everything works without it, you just won't be able to see what the card is doing. If you're capturing video you'll certainly want to be sure if the card is receiving anything though, so you'll have to get this working. Any card will do. I'm using a Geforce 4 MX440 for VTNT, and I imagine that would work just fine for later versions.

Also, the software will barf and explode if you don't have a working sound device.

Storage

The biggest stumbling block for the original Toaster NT is that you need a fast hard drive.

The VTNT dealt exclusively in uncompressed video in NewTek's RTV format. To make that work, you need a sustained 21 MB/s read and write for each concurrent stream. This is not a suggestion. If you don't provide sufficiently fast storage, the software will simply throw up its hands and give up.

A basic IDE drive will usually not work. I tried a consumer one from the early 2000s and I couldn't even play back a single video without erroring out. NewTek urged that you use either a striped set of IDE drives, or an Ultra160 SCSI disk. I am using an Ultra160 card and a 15.6k Seagate on a VTNT with no trouble.

The original VTNT software can't do anything in realtime with more than one stream, so you never get beyond this requirement. If you have a VT[2] or later and you want to play two RTVs at once however, you'll need 42 MB/s; if you want three, you'll need 63, and so on.

Beginning with Video Toaster [2] Newtek added support for compressed AVI files which are much smaller, so this hard speed requirement is relaxed. However, they still advise using uncompressed video whenever possible to prevent generational loss, and their recommendation for realtime two-stream editing with uncompressed video is a disk array capable of 70 MB/s reads.

I suspect that if you can get the VT software running on a machine that supports SATA, a single spinning disk will deliver more than enough performance; an SSD will probably wipe the floor. This is true for VT5, certainly, which I have running on a 3rd gen Core i7 with a single 7200RPM SATA drive, though I have not thoroughly exercised the uncompressed video capabilities.

Design principles

All-software processing was a core philosophy of NewTek in the design of the PC Toasters. Literally, per the manual for the [2]:

When we started Video Toaster for the PC, we made a fundamental decision that shaped the entire product and resulted in Video Toaster [2]'s drastic departure from other video products on the market. We observed that computer speeds were increasing at a staggering rate (as hypothesized by Moore's law) and recognized that eventually the general purpose CPU at the heart of any modern computer would have sufficient power to perform all the real-time processing for video production without dedicated hardware and DSP circuitry. We could now concentrate on developing unrivaled video quality instead of complex image-processing hardware that would be made obsolete.

In fact, that foreword goes on to make some other very good points, saving me the effort:

Arguably the most important consequence of our 'host processor' design philosophy is that we are not locked into any particular architecture or algorithm that would need a new hardware design each time it changed. The benefits of this approach are illustrated by an example from the development of ToasterEdit.

After we implemented the picture-in-picture code (PiP) that places, moves and fades images and overlays in the editor, we realized that we could make high-level assumptions about the typical alpha channel topologies of video images.We were then able to increase the quality of our PiP and increase performance.

If we had been tied to a hardware architecture, this improvement would not have been possible.To highlight this benefit in a different way, consider that the next release of Video Toaster will be available to you as nothing more than a software upgrade; we can add as many new features as we want without needing new hardware. As computers get faster, we will be able to perform more and more complex manipulations without requiring you to purchase a new solution!

If you're not aware what they're contrasting to here: Prior to the late 90s, CPUs simply had not gotten fast enough to handle video, beyond playing back low-resolution "thumbnail" clips. At one point in the 90s, even playing a 640x480 MPEG on a PC required a hardware accelerator.

So in say, 1996, it was impossible, full stop, to do realtime video processing on a PC. It simply wasn't happening. That's not to say PC-based video editing didn't exist - it did, but it was either very slow, or it required gobs of custom chips; either ASICs (actual custom silicon) or FPGAs (fantastically expensive programmable silicon.)

The above (courtesy @compgeke) is a Pinnacle real time effects card, utilizing more FPGAs than I've ever seen in my life.

When you edited video on a setup like that, the system's CPU was basically doing nothing. At most, it grabbed video data from storage and stuck it in RAM for the card to retrieve.

All the chips on that card are "hardwired" to do things like mix two videos or transform a picture. Everything was fixed-function. In a sense, these were like modern GPUs, but without the flexibility; highly parallelized, but optimized only for specific tasks. They had hard limits on how many effects or video clips you could work with at a time, because each one literally took up a chunk of physical space on the chip.

NewTek experimented with this in the mid 90s, but ultimately decided to bank on CPUs getting fast enough to do it all in software - which is exactly what happened, to great benefit.

The point being made above is that, when NewTek discovered suboptimal code in their PiP routine, fixing it in a hardware-based system would have required thousands, or hundreds of thousands of dollars of rework. Instead, they just... compiled a new EXE that could then be downloaded from an FTP by customers, which is exactly what they did.

This dig at unnamed companies making hardware-based systems was present in the Video Toaster NT manual as well, and I am inclined to think it was explicitly directed at Play, Inc.

Play was formed by some ex-NewTek employees (including one of the founders) who went off to do things Their Own Way in the mid 90s. I suspect there was a schism over whether the future of video processing was in custom hardware or CPU-bound software. Play, Inc. produced one product, the Trinity / GlobeCaster, then went out of business; NewTek still exists, so I think we know how that argument turned out.

The Tricasters

As far as I can tell, the Video Toaster NT series was very successful, since NewTek made five versions of it, the last one coming out sometime in the mid-late 2000s. At that point, NewTek pivoted. The Toaster brand was retired, and the technology was rolled into a new product line called Tricaster. It was not quite that straightforward, however.

First, let's talk about how the two products differ. Then we'll talk about what the Tricaster itself was meant to be and how it evolved.

Tricaster vs. Toaster

Up until that point, the Video Toaster series had been a PC accessory. It was mostly a sophisticated software product, with a somewhat interesting hardware component. You were meant to put it in a PC you built yourself, running software of your choosing; there was a big business for system builders to make prebuilt Toaster machines.

The Tricaster was sort of like Newtek bringing that business in-house, but not quite, because the Tricaster was intended for very different purposes, and it's software design and capabilities are quite different from VT.

First, the Tricaster lacks an entire third of the Video Toaster's creative software suite. All versions of VT included a raster graphics package (ToasterPaint, and later Aura) and Lightwave 3D. These are not included with any Tricaster.

More significantly, the Tricaster is not modular. Let me explain that in detail.

The original Amiga Video Toaster was a fixed-function product. It supported four analog inputs and two digital framebuffers, and nothing more. However, VT[2] through [5] were heavily modular and flexible. They supported many analog and digital input formats and an integrated switcher with up to 24 analog inputs, which could be assigned to the input bus in any desired combination, in addition to a ton of internal inputs. You could make a VT[5] system out of nothing but DDRs and simply switch between local video files, or CG pages, or static images, spread across 24 simultaneous inputs.

These systems were also designed to express Newtek's then-philosophy of fully CPU-based design. You could build whatever system you needed, and if your machine wasn't fast enough, simply add more hardware. Tricasters, however, are "appliances"; opening the case voids the warranty, so you can't add more CPU cores, or more RAM, or a faster hard drive (at least internally.)

Perhaps this is why, while the Tricaster software is clearly based on the VT software (you can even still hit Ctrl+Alt+V+T to instantly exit) Newtek removed all the modularity. You can't rearrange the interface, add or remove modules, or remap inputs. You are hard-limited to a certain number of DDRs, title pages, static images, and so on - usually two at most.

These limitations may have been the only reliable option to ensure that users couldn't overburden their Tricaster. If they allowed modularity, then a user might manage to create a combination of features that bogged down the hardware, but if Newtek simply tested every combination and then placed hard limits on a reasonable set of features that could work on that set of hardware, then nobody's surprised when their three-input Tricaster can't handle 9 concurrent DDRs.

But it's also possible that they simply wanted to stratify their product lineup. The VT was based on being the cheapest option in the market - but by the late 2000s, software-based video switching was becoming very cheap, so Newtek may have needed a different angle to stay profitable. A hardware "appliance" is a lot easier to break into separate SKUs: you buy a cheap one when you have two cameras and only need one CG page. If you want more of those things, your only option is to buy a bigger, far more expensive Tricaster.

I'm not sure that's it, though. Arbitrarily limiting software features in order to create more expensive license options is an ancient and time-honored tradition; it feels like they could have just licensed these capabilities if that was the sole concern.

And likewise, they may have cut Aura and Lightwave so they could sell them separately, but it's also worth pointing out that they're VERY heavy applications. The VT software would let you alt-tab and run those apps, and they could severely bog down the machine. Tricasters absolutely do not allow you to switch tasks - as long as the Tricaster control center is running it dominates the entire system, I don't even think Explorer is running.

This makes sense if Newtek was focused on reliability. If you could run Aura, you could run Photoshop - on a VT[5] system, if you bogged it down with a 50 layer project, Newtek could say "Buy a bigger CPU." Here, they can't, so the safest option is to simply state "The machine can do these things, and is not meant for anything else." Indeed, the TCXD-era software outright pleads with you not to install anything other than the TC software or you might experience "instability." It's a PC, so they can't prevent you from running other stuff, but it's clearly a bad idea in their eyes. So I'm inclined to think their focus was less profit and more reliability.

Tricaster History

As I've said several times in this document, the Tricaster was not (as far as I can tell) intended to be the evolution of the Video Toaster, nor meant to replace it in the market.

Modern Tricasters are replacements for the VT, and more. They're positioned as complete end-to-end video solutions for broadcast and many other things. I don't know much about those, but the original was not positioned even remotely like that.

The first Tricaster (just called Tricaster, model TC-100) was sold in 2005, and targeted exclusively at small-scale, no-budget productions. Newtek's marketing made almost no mention of broadcasting except regarding cable access, and that came after teachers and HR reps. Newtek primarily imagined it being used to make internal training videos and internet streams - which were nothing like they are now in 2005. Nowadays streaming content can include major productions, but back then it was often things like "church services" or "seminars."

The name "TriCaster" was in fact coined to describe three output methods - recording, streaming and projecting - that Newtek was focusing on. Those just happen to be exactly how every "cutting edge" town hall meeting was being made available to the public in the mid 2000s; indeed, my Tricaster Pro came from Dallas-Fort Worth city hall.

This version of the Tricaster was heavily angled towards non-professionals, to the extent that Newtek actually used more generic terminology to make it more accessible, and removed some features and UI elements.

They also removed the title generator, although the video editor remains, and there is no Preview window, only Program (or as they called it, Live.) Although Preview comes back in later Tricasters, I assume their reasoning here was due to the TC being able to display all three camera inputs in "thumbnails", so the user didn't need to select one on the Preview bus in order to see it.

The TC-100 was on the market for a year or so before, in about August 2006, Newtek produced the first Tricaster Pro, with more input and output formats - and BNC connectors instead of RCA, the universal indicator of amateur hardware.

At this point, VT[5] hadn't even come out - VT[4] was still being sold. VT[5] wouldn't come out for a year or two, and although I'm not sure when they stopped selling it, they were still updating it as late as 2010 - meaning VT and Tricaster were being maintained concurrently for five years.

Based on those facts, I don't think the Tricaster was supposed to replace VT, but at some point Newtek must have realized that it made sense. My guess is that, by 2010, it was becoming apparent that software solutions like vMix combined with inexpensive hardware capture cards were poised to obsolete a "PC addon"-type product like the VT, and since (as I'll detail later) Newtek had already discovered that the existing software needed to be completely rebuilt from the ground up to handle the now-dominant HD format, it was a good time to call it quits and move on.

I am being sent a much newer Tricaster. If it works, I will be able to continue this history into the modern era.

Hardware

SD-era Tricasters

TC-100

TC-100 UI screenshots.

The very first Tricaster (TC-100) from 2005 was built around a Pentium 4 Shuttle system - it literally says Shuttle on the side. Part of me wonders if they even got these from Shuttle directly or just bought 'em at Fry's Electronics.

It uses the same VT Pro card as the VT[5], but the front panel of the case has been replaced with a miniature three-input switcher (internally known as the SX-3 - the later 6-input models had an SX-6.) The TC-100 offers only RCA composite and S-video inputs. Anyone who's worked with SD broadcast gear knows what that means: consumer market. Indeed, the intended market was ostensibly very low-end, and the interface did not resemble a serious piece of professional equipment.

Tricaster Pro

Tricaster Pro UI screenshots.

The next newest machine I have is a Tricaster Pro, from probably 2007. It is a newer, Core 2 based Shuttle (an FG31 motherboard, I believe) with more inputs and outputs, and using BNC instead of RCA connectors (the universal sign of Professional Video.)

It's difficult to photograph the insides of these since they're very dense, but here are a couple overview shots (plus the one above.)

Newtek made a number of Tricaster models with various improvements, but for the whole SD era they were nothing more than a PC running special software. You could simply drop out of the app to Windows XP and run any software you liked. Those models include the Duo, Pro, Studio, Broadcast and just plain Tricaster.

The TC100 and TC Pro are in nearly identical chassis, although the motherboards are completely different. Here are some comparison photos: 1 2 3 4 5

Important info about capacitors

I call the 2000s the "Blowncaps Era" (same tone as "gibson lawsuit era.") The capacitors on every motherboard of this era are bad, and Shuttles are particularly bad. This is not a likelihood, it's a certainty: You need to recap this motherboard or the machine will slowly stop working. My TC Pro ran for about an hour before crashing when I first got it, and by the next day it couldn't even boot without freezing. After a recap it works perfectly.

Recapping is not expensive, nor is it difficult if you have soldering skills and patience. If you don't, you should find someone who does, because if you damage this motherboard you will not easily be able to replace it. Shuttle motherboards are proprietary.

If you have the Tricaster Pro with the FG31 motherboard, here is a list of the caps you'll need, and a picture of their locations. I only replaced the "tall" caps, which were more likely to be bad (based on badcaps.net) You can do more if you want, but it's more opportunities to overheat and damage the board. I recommend solid polymer caps, because they're shorter.

If you have other models, just look for caps that are visibly blown or bulged, and if all else fails, just replace everything larger than 1/8" in diameter with a vent on the top.

The GPU included with my Pro was a Geforce 7300GS. All the caps were blown. You could recap this - they're clearly labeled and easy - but the card is actually unnecessary. I switched to the onboard graphics chip and it works perfectly.

Important info about cables

The SX-3 connects with three of the awful little blue micro coax cables used in the VT[4] and [5]. I'm told that these cables had a tendency to wiggle loose in transit, or even break due to vibration. Eventually Newtek began hot gluing the cables in place, which is what I found in mine.

The cables feel fragile as hell and I don't recommend putting too many bend or plug cycles on them. If you do need to unplug them however (and if you're doing a board repair, you will need to) please follow this advice:

Activation

As with all of Newtek's PC products, the software works for a couple weeks and then demands activation, which is done over the phone. I am told that if you call Newtek, even now, they will transfer ownership of a machine of this era on request. YMMV

The HD Era

As I've heard it told, NewTek thought they would be able to move into the HD era by simply adding more CPU power, but their philosophy ran out of steam. I've seen Usenet posts where they asserted that HD was going to happen with the original software-based approach, but that never came to pass, and I've also heard that that they made a prototype - and it couldn't exceed 720p.

Newtek had tapped out CPU-bound video processing and had to add some kind of hardware solution, so the HD Tricasters were created, starting with the TCXD300, which appeared on their website around Dec 2009 (but may not have been available until early 2010.) They made many more submodels that varied mostly depending on number of ports.

Architecturally, these systems are actually still fairly close to NewTek's software-only philosophy. They use commodity motherboards; they run Windows; they still use a special card, but it's still just a capture and output device, simply upgraded for HD.

The big change that made HD possible was the switch to hardware compositing - but even then, only using commodity hardware. While all previous VT/TC products required a GPU, it was solely for live onscreen previews (since it's very difficult to put 60fps video on the screen with a plain Windows GDI surface.) The Tricaster XD series, however, can't function without a GPU, because the entire compositing engine runs inside a Direct3D context.

The TCXD (and, as far as I know, all subsequent Tricasters) basically just shove the frames captured by the input card into VRAM, then create a D3D scene that applies them to a surface, then transforms the surface and uses pixel shaders to accomplish effects.

This obviously decreases the CPU load tremendously, but without involving the hated "pile of FPGAs" that Newtek had worked so long to avoid. The GPU is an off-the-shelf model - in the specimens I've used, it was simply a GeForce, not even a Quadro. I think the TCXD850 I borrowed had an 8800GTS. Very little oomph is required, because after all, we're talking about just a couple of 2D images. I suspect that if I transplanted the card to a machine with an Ivy Lake Intel chip or something, it would probably work just fine on the onboard graphics.

But, going back to that I/O card: Newtek's "taco bell menu" approach (e.g. a hundred products, all made from the same ten ingredients) had been going on for many years at this point, and it hadn't ended yet. The Frame Factory became the VTNT/[2]/[3]; the VT[4] card became the VT[5] and a dozen models of Tricaster. And so it's no surprise that when you open a Tricaster XD, you don't find a card labeled "Tricaster."

The card is branded '3Play.' What was that? An (HD-capable!) live video replay system, which... very curiously resembled a Tricaster. Announced in Feb 2009 - so, once again, we can guess that Newtek got the hardware for the new "Toaster" done before the software, and decided to sell the hardware as a separate product. And then once the new TC was done - hey, why bother changing the silkscreen on the PCB, when opening the case voids the warranty?

I have no particular observations on the card, but I do want to mention that the "SCSI" cable you see there probably is just that, but not being used for SCSI. It's just a big bulkhead connection to the breakout box on the front of the machine (pretty much just like the SX-3 on the first Tricaster.)

I believe the SD Tricasters were running analog video over their little blue ribbons, so they had to be made with "micro coax" to preserve signal integrity. The TCXD uses digital signals between the BoB and card, if I'm not mistaken, and that probably allowed Newtek to switch from fragile, discrete coax to these nice, clean, high-density ribbon cables, of which 68-pin SCSI was readily available.

Regarding how the machines work: I have used a TCXD850 briefly - a huge, 8-input rack model - and I own a TCXD300, another little Shuttle PC with 3 inputs. They are, again, just logical extensions of the original Toaster; more capacity, more framestores, more effects, etc.

The early 2010s machines like mine can only do SD, 720p or 1080i, which was extremely common for the era and appropriate to the intended application: ATSC broadcast television, which was and is still 1080i. As far as I know (not being a working technical director) I don't think there's a technical reason you couldn't produce an ESPN hockey game in 2023 on one of these, if your whole $600,000 Sony board went up in flames and you needed something Right Now.

I don't have any information about Tricasters made after the early 2010s. I suspect they are not very different in principle, just using newer motherboards and software. They have, of course, confused the hell out of their model numbering just like all companies that survive longer than a decade, so they had TCXD850, and then TC860, and then TC1, and so on, so you really can't guess when any of these are from without trying to find a press release for that specific model.

Resources

Downloads

I have a few files I've collected.

List of Articles