Paulie’s Perfunctory Game Dev Website
Video Game Development Since 1981
© Copyright 2006 - 2019, Paul Hughes Operation Thunderbolt C64
Operation Thunderbolt was the sequel to Taito’s arcade smash Operation Wolf. Ocean had previously licensed and converted Op Wolf to all the main 8 and 16-bit platforms to great fanfare from the press and public alike, so it was little surprise to anyone that Ocean immediately signed up the sequel for the same treatment. It seemed like a no brainer, indeed for every platform except the Commodore 64 it was. So, what went wrong? Time to go behind the curtain…

Preamble

Although the original author of the Operation Thunderbolt “preview” can be found on the internet with a little Google-Fu, I don’t feel the need to add any more to this person’s embarrassment, something that I am no doubt in part responsible for. To protect the innocent, let’s call them “Roy” after Roy Adams, the game’s protagonist. As the story unfolds, hopefully you’ll see there’s a lot more to the story than just “Ocean released a sub-par follow up to Operation Wolf” – this despite getting hugely positive previews.

A Fateful day

Looking back through old source files It all began on the 19th of February 1990. In just 12 days’ time I would be driving to Ablex the duplicators in Telford, Freeload Master Maker in hand, to kick off the first duplication run of Operation Thunderbolt for the C64. For reasons that will soon become clear, it would be another 29 years before I stumbled across the game’s reviews, now archived on the internet. It started as a quite unremarkable Monday morning, as usual I trundled into the office at around 7:30am, can of Coke and a bag of Walkers Beef and Onion crisps in hand. Gossip of the morning was “Roy” had been let go on Friday evening. Let it be said, up front, Roy was a lovely lad; pleasant, well-mannered and eager to learn his trade. To that end, quite a few of us down in the dungeon had realised early on that he had a bit of a mountain to climb with his lack of experience and so attempted to impart various techniques we’d used in previous titles to make the conversion a little easier. Some people were, well, less helpful in their interactions. Back to Monday morning, Gary Bracey collars Johnny Meegan, Rick Palmer and myself on the code front along with Brian Flanagan and Stephen Thomson on the art side and we convene in his office.

Your mission, should you choose to accept it*

First, a little exposition on the business of “hardware bundle deals” back in the day. As the Amiga “Batpack” bundle had shown, bundling in big name intellectual property alongside hardware can be a jolly lucrative proposition for all parties involved. The Dixons group had concocted a hardware bundle of either a Commodore C64, a ZX Spectrum or an Amstrad CPC bundled together with a joystick, mouse and light gun (no doubt as a means to offload an overstock of peripherals) – in order to do this they needed some big name brands with full support for these “value for money” bundled in peripherals. For Ocean, buoyed by the success of the Batman bundle with Commodore, and the critical reception of its Operation Wolf conversions previously, it seemed like a no brainer to take the licensed sequel of Taito’s Operation Thunderbolt and bundle it with a Light Gun, I mean, the arcade game used a gun, right? Now, understand putting together a bundle isn’t a cheap business; all the component parts need ordering, custom boxes need to be designed and assembled, and huge TV, magazine and radio advertising spends need executing. If any part of that chain were to break down there would be hell to pay – or more specifically a clause that triggered a massive financial penalty for the partner that “dropped the ball”.

The best laid plans of mice and men…

All Ocean needed to do was to deliver three versions of Operation Thunderbolt across three platforms, supporting a bunch of peripherals and the rest would be cash in the bank. Deals were signed, teams were allocated and four months were set aside to get everything done. Taking up Spectrum and Amstrad Z80 duties were the formidable team of Andy Deakin and Ivan Horn who had previously converted Operation Wolf to great critical and commercial success. Colin Porch who had crafted the remarkable C64 version of Operation Wolf wasn’t available to work on the sequel as he had already been assigned to another project that he was currently knee-deep in code. Alas, every other in-house C64 coder was tied up in other titles so a new C64 coder was sought.

A Hiring Detour

The hiring process back in the eighties at Casa Ocean wasn’t quite the science it is nowadays (and saying that, a lot of the processes used these days still have a certain element of gut feeling, rather than a strict formula). Back then, a prospective employee would send in either a demo, or a copy of a previous title they had developed and if Gary was sufficiently impressed, and the price was right, you got hired. This is in no way a criticism of Gary; his “hit rate” for spotting both talent and products to license was second to none – the team he assembled in the Central Street dungeon along with his sixth sense of landing movie tie-ins was nothing short of prodigious. All said and done, “Roy” was hired and given the daunting task of creating Operation Thunderbolt with all its contractual peripheral caveats in the next four months. Don’t be late, and if it could be at least as good as Colin’s Operation Wolf conversion all the better.

Let the game commence

With a programmer primed and paired up with Brian Flanagan on art duties, a full-sized Operation Thunderbolt machine was delivered downstairs to the dungeon, sitting outside of Gary’s office, as it was way to big to fit into our usual hidden corridor, affectionally known as “arcade alley”. After a couple of weeks of coding and several somewhat tipsy debriefs at the local hostelry, The Square Albert, (fittingly located on Manchester’s Albert Square) it became very apparent that this title was a lot bigger than any other title Roy had ever attempted. He had no experience of creating horizontal scrollers, multiplexing sprites, or sequencing attack patterns together. A bunch of the programmers started offering advice and chunks of code to get things up and running; Andy explained how his Operation Wolf had worked and what he was doing on the Z80 versions of Thunderbolt – things appeared to start coming together. Brian’s graphics were going in, and some big sprites were trundling around the screen, but alas all was not as it seemed.

The hype trail

As the bundle’s release date started to get closer, the publicity started to ramp up with exclusive previews being assigned to a few of the mags (yes, I know, not strictly a solid definition of “exclusive”, but we’ll let that slide for now). ZZap 64 saw a version of the game that wasn’t playable – it was seen as showing off the game’s attract mode. With Ocean’s pedigree (and no doubt massive advertising spend) – the preview was glowing and didn’t particularly mention the bit about not being able to play the game. As the days went on the game “wasn’t quite ready” to go into QA each day, and then it was pointed out that the new pseudo 3D sections that Thunderbolt had added to the franchise hadn’t yet been started. Oh, and all the peripheral support, the reason for the bundle pack, was ominous by its total absence.

The Penny Drops

Like any pimple popper will tell you – eventually all things must come to a head (groan!). After several days of QA raising concerns about not getting playable versions and the other developers urging Gary to intervene, a meeting was called which, alas, I wasn’t privy to, but it was recounted to us later that what there was of the game was demoed and the company parted ways with Roy.

An indecent proposal

As mentioned in the opening salvo, the penalty clause was restated to us and a plea to pull together something by any means possible, as fast as possible, but before the 3rd of March as cassettes needed to be delivered no later than the 5th of the month! I suspect it won’t come as much of a surprise to you, dear reader, that the dungeon crew were used to burning the midnight oil, suffice to say over the next thirty-odd years I’ve been involved in some almighty crunches to get games out of the door, but I don’t think I’ve ever worked so long without sleep since we expelled Operation Thunderbolt unbidden into the world! The last three days were a total blur fuelled by Jolt Cola, Pro Plus and Pizza Express Pizza, culminating in a 150-mile round trip to the duplicators entering in the morning with a master maker (did I mention I had to create the Freeload tape master as well) on floppy disk and leaving that evening with a boot full of boxed product.

12 freakin’ days! No wonder it sucked, right?

Here’s the rub; there’s a lot more to it than the “condensed” time frame. Whilst I can take a little solace in “it was the best you all could do in the time available”, it wasn’t just that. Let’s leap forward a tad, to, well, 2019. I follow a Twitter freed called “Yore Computer”; a play on the classic British computer magazine “Your Computer”. This feed digs out pages from archived computer magazines of yesteryear and tweets them out. Planets aligned, and random number generators coughed forth the ZZap 64 review of Operation Thunderbolt. It may come as a surprise that I never bothered to look at any reviews of Operation Thunderbolt back in the day, because, if truth be told, I think we were all embarrassed by it. It dug Ocean out of a hole, but it was by no means any of our finest hours. Much to my amazement and incredulity it was awarded a ZZap Sizzler. Now, I’m under no illusions that the game remotely deserved that, in fact, I have no doubt things were said or done that led to a swathe of positive reviews, but upon my chuckling about it on Twitter, it appears that the biggest common gripes about the conversion were things that we all totally agreed with but in some cases were powerless to change…

What went wrong, a 30-year old retrospective

I present the following purely as a look back on what happened during those fateful twelve days no more, no less. When push comes to shove, the end user doesn’t give two hoots what kind of troubled development a game had; they just want their nine pounds ninety-nine to be well spent. We’ll take it as read that hastily putting together a game in 6502 assembler in 12 days flat had more than a passing impact on the quality of the final game, but with three decades of hindsight, I think there’s more to it than meets the eye.

How was Version 1 allowed to go on for so long?

To even the most the casual reader, the first observation must be “why on earth did it get so far down the line without alarm bells ringing?” Looking back, I guess that was just a corollary of how much work Ocean had on simultaneously along with a somewhat misguided do or die mentality when it came to shipping a product. Somehow, by apparent force of will, games came out of the dungeon despite any development hell it may have suffered, and with so many titles in flight, we had more than our fair share of “issues” yet still we put out (mostly) good games. A lot of the early warnings were perceived as a bit of banter; “company X is a bit shit”, “programmer Y can’t code for toffee”. Think of it as the games industry equivalent of crying wolf. Alas, on this occasion, the wolf came along, blazoned in a luminous hoody bearing slogan “I’m going to screw your game” and promptly devoured all of the time.

Laser Sights and Scaling Sprites

Operation Thunderbolt added a couple of new bits of business; first, a new pick-up; the unassuming laser sight. This was the über pick up, in that you had a dot projected into the screen showing exactly where your bullet would hit – a useful addition when you had a big haptic gun strapped to the arcade machine. When you were playing on a home computer screen, deriving input from a mouse or a joystick, we couldn’t just have a cursor on screen as we did with Operation Wolf as that would negate the advantage of the laser sight. Instead we had to “kick up dirt” where a missed bullet hit so the player could intuit where the gun was roughly aimed. Only when you got the luxury of a laser sight would you get a clear reference for your aim. To this day I still say this was a HUGE mistake. In particular we didn’t like it on the horizontal scrolling sections, as you were almost spraying rounds into thin air hoping to hit something by chance rather than skill and then try to keep things pointing in that rough area. It goes without saying that most players thought the same; it killed the fun of Operation Wolf when omitted from Thunderbolt, unfortunately Taito wouldn’t budge on it – having an on-screen cursor broke their mechanic, even if it did hinder the home conversions. Now, with that said, all the other platform conversions had the same restriction without too much critique, so it clearly wasn’t the only deal breaker. What other things were afoot?

Pseudo 3D

Let’s talk about the newly added pseudo 3D sections with the antagonists hurtling towards you. The arcade game had scaling hardware, so with just a few frames of animation and a rotozoomer you got a reasonable interpretation of 3D characters, boats and props moving towards you. This was an “interesting” challenge at the best of times – trying to give the appearance of something moving towards you with a limited number of memory hungry pre-scaled objects on a character mapped screen – even the best C64 pseudo 3D games like Turbo Out Run struggled to convincingly move lots of scaling objects towards you without it looking like a big mess of blocky characters overwriting each other. Johnny Meegan and Ste Thomson, for my money, made a valiant attempt at pulling this effect off with carefully positioned character and sprite objects that didn’t do too much intersecting to give away the character block nature of what was going on. Space Harrier and Power Drift were other examples of a soul focus on presenting pseudo 3D on the C64, and even then those titles that exemplified the technique couldn’t sufficiently nail the effect of emulating hardware scaling. Once again, however, the same criticism could be laid at the door of the other platform conversions, so this on its own wasn’t a calamity in and of itself.

VHS – A Study in Gameplay

During the game’s original gestation, I think we all had a good old blast on the Op Thunderbolt machine; it greeted you as you walked down the steps into the dungeon. If you weren’t specifically tied to a title you didn’t really over analyse what a game was doing; how the enemies moved, how much animation it had how many sprites did it push around. To that end, when we got thrown in at the deep end we literally didn’t have any time to deconstruct the game, nor did we have any extra info from Taito (we very rarely got anything other than the arcade boards, no raw art, no documents, no code), so we relied on videos that QA had made playing the game through to the end. Shuttling backwards and forwards through those tapes we got enough info on level structure and attack patterns to start putting code and art together. Time for a slight diversion again; porting to different platforms. It goes without saying that the various computers had their own strengths and weaknesses; the C64 had its hardware assisted scroll, SID and hardware sprites (latterly augmented with Y multiplexing). The Spectrum on the other hand had the much faster clocked CPU with the ability to shunt data 16 bits at a time via the stack. Whilst the C64 had its hardware sprites there was a hard limit of 8 on any horizontal line, the Spectrum had to draw everything via the CPU, but the limit was how many sprites you could draw in your desired frame rate; if you didn’t mind the odd frame rate spike you could go nuts with no glitches. So what I hear you ask! One of the options we considered was Andy had virtually completed Op Thunderbolt on Spectrum and Amstrad and despite them being written for a different processor we could potentially lift all his attack pattern data that he and Ivan had discerned during their port and convert his movement code to 6502. Except we couldn’t. Andy threw an inordinate number of sprites around the screen with his super optimised sprite draw routines and his pre-shifted background scroll; way, way more than we could fit on a horizontal line. Whilst we could multiplex on the C64 via raster interrupts (repositioning a sprite’s Y coordinate value once the TV’s beam had traced out the sprite) we couldn’t work around the “number of sprites on a line” hard limit so the data was pretty useless to us. In the end Rick had to bite the bullet and tried to map “the beats” of the levels when specific key moments of the action happened and then filled up the gaps as closely as possible to the playthrough video as we could with the limitations of the sprite hardware. This was essentially how Colin tackled the port of Operation Wolf to the C64; he just had more time to refine it. Which leads me to the next perfectly valid question;

Why the hell didn’t you use the Operation Wolf codebase as a starting point?

I asked myself this question the other day, and with hindsight it was an obvious move, but (and there’s always a “but” in stories like this) firstly and importantly the key to how Colin got the gameplay right, was, well, getting the gameplay right! It takes time to study, setup, refine and balance. Time that we didn’t have. Even if we had taken Colin’s code there was the time required to get our heads around it (disturbing him from his current workload) and then create all the new gameplay data, so the time would most likely have evened out giving us the same quality of gameplay just in Colin’s engine. Then of course there were the added non-negotiable spanners thrown into the works that Thunderbolt had; two players playing simultaneously, and those two players could be using combinations of a mouse, a joystick, a light gun and the keyboard.

Peripheral Hell

There’s one thing agreeing to support a raft of peripherals; its an entirely different thing trying to make it all work in the context of the conversion; in theory either player could use any peripheral they wanted. In practice hardware input limitations cancelled out certain combinations; for example, you couldn’t use a NEOS mouse and a light gun together, but you could read a CBM 1531 mouse on either port, except if you wanted to use the light gun, in which case the CBM mouse had to be in port 2, whilst the light gun was in port 1. All these technicalities weren’t discussed with the developers when the deal was done. For the most part, the Spectrum and Amstrad versions didn’t have too many issues with the peripherals, the 64 on the other hand could be a temperamental beast when it wanted to be. To that end there was a lot of horrid logic in the controller setup that had to toggle peripheral selections depending on which player was using what peripheral on what port. Suffice to say what we would call “User Experience” these days was not considered; it just had to work with all the peripherals in some combination, even if the setup up was pernickety it still fulfilled the terms of the contract.

A Perfect Storm

All in all, the lacklustre product was the sum of its parts. It could be easily argued that given the same four month timeframe a lot of the aforementioned issues could’ve been ironed out, and in many cases that is a perfectly reasonable conjecture. It is still hard, though clearly not an intractable problem, to do the pseduo 3D better. Given the extra time the flow of the game and its attack waves could’ve been honed into a much more faithful representation of the arcade. The peripheral support would still be messy due to the hardware limitations, but the configuration could be so much cleaner. With hindsight and the time, the original conversion should’ve started with Colin’s Operation Wolf source code - after all that was a tried and true finished product that covered 70% of the mechanics of the sequel. This would’ve been a great leg up for “Roy”, leaving a lot more time to deal with the pseudo 3D and shunting things around to add in a second player simultaneously. *There was no option to decline the mission! It had to be done :o)

External Links

Games that Weren’t - Operation Thunderbolt V1 ZZap 64, Issue 57 - Operation Thunderbolt Review Mark R. Jones’ Ocean Software Ltd Facebook Group - the de-facto reference for all things Ocean
Behind the Curtain
Jon Dunn and Steve Wahid playing the arcade machine Photo Credit: Mark R. Jones
Operation Thunderbolt Magazine Advert Roy Adams is back!