RVG Interviews – Fred Gill.
RVG is please to announce our latest interview with Fred Gill, Fred is the co-founder of Attention to Detail.
Attention to Detail, also known as ATD, is a software developer based in the UK. ATD was formed in September, 1998 by a group of graduates from Birmingham university, as well as working on games they also created some development tools and designed electronic parts for arcade machines.
The original five founding members are; Chris Gibbs, Fred Gill, Martin Green, Jon Steele and Jim Torjussen.
Attention to Detail developed many games that have gained a cult status and had significantly more success in Europe than it ever did in America.
Some of the game developed by ATD:
Fred Gill: I need to caveat all responses – this is how I remember everything; apologies to anyone that I have forgotten or misrepresented – it was a while ago!!!
How did you get Attention to Detail started? How was the company’s name chosen?
It was started by 5 graduates from Birmingham University. Jon Steele, Martin Green, Chris Gibbs, Jim Torjussen and me. Jon, Chris and Martin were on the same Computer Science course and Jim and I were on the Joint Honours Electronic Engineering and Computer Science course.
We all met in the first year at University – I believe due to spending long hours in the Computer Science lab, programming. We were being taught C, and I seem to remember most of us copying games, i.e. Space Invaders, to exercise these newly-learned skills. We all became friends, and indeed in the 2nd Year and 3rd Year, Chris, Martin and I lived together in university self-catering accommodation.
In the summer between the second and third year, another Computer Science friend, Nalin Sharma, who had already written several games (as had Jon Steele) created a small team to work on the port of SuperSprint for the Atari ST for Electric Dreams (Jon Dean was his main contact there); the team comprised Nalin, Martin, Chris (graphics), Jon and Mark Tisdale who did the audio programming (who was in his final year of the Computer Science course). I think they had somewhere around 9-12 weeks to port, and with no source code at all. They would go to the arcade and take photos and record the audio with a big blanket over them and the machine. This wasn’t a perfect system, and indeed, a couple of months in, they realised the music was wrong – they’d captured not only the Super Sprint arcade machine, but the arcade machine next door! I do remember also that the hard disk failed with the final master candidate on, and the team only had a backup from the day before. They hand-patched the final changes in with an assembler as they couldn’t recover the source in time to make the submission.
I worked on my own ZX Spectrum game, Octan (based upon Arian Mission, which I had become addicted to at university) over that summer, and that continued through to Christmas before my finals, as BT’s Firebird label offered me a contract around September/October that year.
As we approached our Final exams, Martin, Chris, Jimmy, Jon and I decided we’d like to create our own software company (Nalin went off to a ‘normal’ job at PwC), and so as soon as the exams were finished we went off to talk to our bank manager about overdraft facilities; Chris cut his hair specially, and I think the first time in the three years that I’d known him. We met with Jon Dean (ex Electric Dreams) and formed a loose arrangement with Jon to work together should he find any work that needed a development team. We also all managed to get onto the Enterprise Allowance scheme, which gave us £40 a week, whilst we worked on demos and ideas over the summer of 1988.
We had long lists and lists of potential names for the company. I don’t remember any of the others, but it did take several weeks to get to ATD. Whilst we all had experience at programming games, we also had an interest in hardware and high performance. We didn’t want to pick anything that precluded games, but also wasn’t ‘too gamey’ just in case something exciting that wasn’t a game came along.
We wanted to ensure the name reflected our passion – we were all sticklers for ‘attention to detail’, having seen the positive benefits of caring about getting the smallest details right.
Can you tell us of your involvement in the development of the Konix Multi-system?
We had just formed ATD, and our friend and mentor, Jon Dean got us some Atari VCS work – I think it was for Acclaim. We started the prototype (the people that programmed those were geniuses – what they could cram into 2Kb and such restricted hardware continues to amaze me), but a couple of weeks in it was cancelled.
Jon then approached us about the KMS. Konix had this new console, designed by an ex-Sinclair team that needed software and tools if it were going to be successful; it had a novel controller, which could transform between a steering wheel, aircraft yoke and/or motorbike handlebars (I believe Konix licensed the design from a talented schoolboy), and rather than release as purely a controller, they wanted to create a console.
We worked with Konix Engineers on the operating system, and worked on demos and development tools to make development as easy as possible and to attract other developers to work on the system. We wrote systems that were as flexible as possible, for example the art package had a scripting system so you could export in whatever custom format you wanted/needed to, and it has several standard exporters included; this was several years ahead of its time.
What were the challenges faced during the development of the machine?
There were many, but they were good challenges as we felt we were pioneering, and the KMS was something special…
The hardware was still being developed. The original KMS used the guts of a Flare 1, and not only were the custom chips changing, but the main processor too.
Konix had a very good reputation for quality joysticks at the time, but there were just too many shortcuts required to get pricing at a level that consumers could/would afford, i.e. the motors for the ‘chair’ (reliable motors were just too expensive) – this feeds back into #1 above, and a changing platform is much harder to develop for…
I remember being at one trade show and working on the controller input and chair until 5am; the public weren’t playing on real hardware as it was so far behind.
I remember Jon Dean was badly burned by this experience. He was owed a lot of money, but Konix simply didn’t have enough to pay, and unfortunately he didn’t have any leverage. ATD was owed £2,000; I remember we exchanged the master source code for the disk copy protection system for the £2,000 in a car park in Coventry. It was shortly before the demise of Konix. It was a real shame Wynn Holloway was a charismatic and passionate man. Ultimately he didn’t have the money to bring the console to market, and had too much pride to partner with anyone else.
ATD developed three games for the Atari Jaguar/CD: Blue Lightning, Cybermorph, and Battlemorph. Can you tell us of the challenges in making these games for the system?
We got hardware and I seem to remember that we had 10 days to prove ourselves to Atari. I did a ‘flying carpet’ demo, which used hold-and-modify (‘free’ hardware fill, allowing for fast polygon drawing) based on David Braben’s Zarch game, and that along with our Konix history got us the first Cybermorph gig with Atari (they originally just wanted some demos).
I touch on Cybermorph and hardware debugging below…
We completely underestimated how much work Blue Lightning was going to be, and I don’t think we fully understood the gameplay when we started recreating it for the Jaguar; I remember squeezing the graphics in was painful (weird shapes, lots of space wasted – a very novel fitting algorithm, which is now known as texture atlassing); we just couldn’t the performance we wanted with the barrel roll. And I am pretty sure the CD drive was a late addition to the project/contract, which compounded our development problems even more.
i.e. there was an issue with the seeks; if you requested a particular sector, the hardware was only guaranteed to deliver to within 6 sectors, and the controller software ran in one of the DSPs we used for audio, so everything had to be combined. We had to incorporate additional checksum and retry code as we found that the ‘within 6 sectors’ was unreliable too!
Cybermorph was released in two different versions, a 1 MB and a 2 MB cartridge version. What was the reason for having these two different releases?
It was a sad day when we got the call asking if we could optimise the 2Mb game to get it into 1Mb cartridge. The reason was purely cost. Atari needed to cut the price of the Jaguar to boost sales and a 2Mb cartridge didn’t give them enough margin, and indeed they may have even made a loss. It took approx. 2hrs to remove enough animation data and Skylar speech sample variations to get it to fit; whilst an interesting engineering exercise, it was painful ‘mutilating’ something that we’d crafted over such a long time.
Was Battlemorph developed at the request of Atari or did your team see the opportunity of using the Jaguar CD as a way to improve upon the original Cybermorph concept and approach Atari?
We pitched the idea to Atari. When you finish any game, there is always a long list of things you would have liked to put in the first game, but couldn’t. There can be many reasons why – time, feature cost, consumers/market not ready, etc.
We definitely felt that we could, and should make a sequel that contained many of the features we wanted to include in the original. I don’t remember whether the CD drive was part of the pitch – that may have been a requirement after the contract was signed.
TrekMDCan you discuss the development of Blue Lightning for the Jaguar CD?
As above, I believe we misjudged the effort required to make a good game. This was made even worse as ATD was expanding to multiple teams, and we made a lot of mistakes with the way we expanded and how we ran the teams.
Were there any titles ATD had started to develop for the Atari Jaguar that were never completed/released?
No. By the time Battlemorph had shipped, Atari were in a lot of financial trouble, and the Playstation and Saturn killed off their markets.
Have you seen the news relating to a possible new console designed for retro gaming that uses the Atari Jaguar tooling? What do you make of this idea?
I had not – I’d love a Jaguar to play my copies of Cybermorph and Battlemorph on!
Can you divulge any stories from your time in and around both the Jaguar and Konix consoles?
Hopefully you see some colour in the responses above.
I liked the Tramiel family; I don’t remember meeting Jack, but Sam seemed sharp. Leonard was the ‘tech’ guy, having programmed algorithms for early calculators in the 1970’s. We were working onsite at Atari for long periods in 1992/1993/1994 and Sam let two of us have his season ticket seats for a San Jose Sharks match – a very memorable night.
It was Leonard that wanted us to texture map everything in BattleMorph, and it was Leonard that insisted everything was textured in Hover Strike with disastrous results. Something that had been quite playable at 15-20fps, was crippled down to an unplayable 5fps as a result (it was only later that the Iron Soldier guys discovered a ‘hack’ which allowed the texture palette to be a texture source, doubling the speed of texture mapping for small textures).
What was your greatest achievement in the games industry?
Creating and being part of ATD for nearly 15 years. We employed some great staff, and I think the environment we created allowed them to grow as artists, engineers and game makers. It wasn’t perfect, but we all learned and grew a lot. I moved into our parent company, Kaboom in 2000, but wished I had stayed in ATD.
What are your fondest memories working in the industry and why?
There are many! They are mostly around the talented and passionate people I’ve worked with over the years, especially those that I’ve finished games with – there’s nothing quite like ‘finaling a game’; everyone at ATD involved in Cybermorph, Battlemorph, Blast Chamber, Rollcage, Sydney 2000, Lego Racers II; everyone at Dice involved in Battlefield: Modern Combat, everyone at Swordfish involved in 50c BOTS, everyone EA or partners involved in Crysis2, Syndicate, Crysis3 and Titanfall, etc.
I also have fond memories of working with LucasArts on Night Shift and Indiana Jones; dining at SkyWalker ranch and having the full tour; getting the sweatshirt, and then throwing it away a few years later – all before I appreciated how special that time was!
If given the chance is there anything you would of changed in the production of the games you worked on?
With the knowledge and skills I have now yes, but at the time I didn’t have those skills.
I try to learn from every past mistake to ensure I don’t repeat them. I also try to learn from past successes to ensure I do repeat them. The latter can be harder than the former.
It must of been ground breaking to have worked on the Atari Lynx at the time, what was it like to program software for such a console?
I didn’t work on the Lynx.
How do you feel the industry compares now to what it was back in the early 90’s?
I still love being part of it. There’s still much to learn – new hardware, new techniques, new opportunities and markets via gen4 consoles, AR, VR, mobile, cloud, etc.
Many of the technical challenges have similar grounding, but more complexity, i.e:
How to parallelise on Jaguar is similar to multithreading on PS4, Xbox One and PC
Even now, performance is critical
Understanding resources and performance bottlenecks; there are always compromises to ensure the technology is affordable for gamers – what should be optimised and when, i.e. it’s still very easy to cripple performance by organising data poorly, etc.
Creating great tools for efficient workflows
Making engineering teams (more than 1 engineer) work well together
It’s still an exciting and pioneering industry to be part of.
What’s the story behind the inital screens of Blue Lightning on Jaguar CD? that were doing the rounds in the UK Press?. They looked far more impressive than anything in the final product, lot better texture-mapping etc. Were they mock-up’s or real in-game shots from the game as it was then?.
Have you copies of the screens, as I cannot tell without seeing them? As per previous notes, we did have problems on Blue Lightning – delivering at the same time as CD hardware being finalised, underestimating the work, etc.
ATD had 12 weeks to debug and test the Jaguar hardware, yet it’s the bugged hardware that everyone mentions as the reason the Jaguar was such a nightmare to work with. Do you feel you had anything like enough time to bug test the chipsets?, what did you find and advise Atari needed changing/fixing? and what did they fix? Impression always seemed Atari simply had to launch the hardware ASAP, so simply didn’t have time to do the much needed fault finding and fixing on the hardware.
Flare were responsible for the hardware; they developed the custom chips over several years, and as with any chip design had sophisticated test simulations.
We found a crash in Cybermorph about 4-6 months from ship, which we believed was a hardware bug, not software bug.
It was very painful – I drove from Birmingham to Cambridge every day for two weeks to work with the Flare team to try and isolate the bug to as small a piece of code as possible. We eventually got it down to about 15 lines of assembler, at which point the Flare team were able to isolate it as a hardware bug; they managed to integrate a fix into the production run; any later and it could have been too late, or prohibitively expensive…
I was just glad it wasn’t a bug in my code, after all the time they’d spent looking at it. I don’t remember encountering many more bugs in the final released hardware (I am sure there were some that we had to work around, but they were not as catastrophic as the one we found via Cybermorph)
ATD seemed to change it’s mind on the Jaguar over the years, originally praising the hardware, saying it’d be a good few years before anyone got the best out of it, then feeling it was something of a ball and chain as it were, Atari wanting Jag CD games to be fully textured, so they could compete again’st 3DO/Saturn+PS1 games, even though that’d mean a frame rate of 3-5 FPS and by being contractually obliged to finish Battlemorph, you were going to be late getting started on development for PS1+Saturn Would you of liked to have just finished up Atari work once Cybermorph was finished and it was apparent Atari simply did’nt have the resources to compete, or support the Jaguar properly, so you could of started on development work on other systems sooner?.
We were caught up in the excitement (and love of Atari – when ATD first set up, our first development systems were based on Atari STs and we created tools and assemblers for the other platforms on them) before the Jaguar shipped. Reality hit very quickly with the realisation that this wasn’t going to be a multi-million seller that would revive Atari’s fortunes.
ATD was happy to sign contracts for Blue Lightning and Battlemorph after we finished Cybermorph; it provided a steady stream of income and allowed us to expand. We hoped that instead of a fast-seller, the Jaguar might be a slow-burn, but it wasn’t to be.
With the Sony and Sega announcements, it became clear that Sony and Sega had leap-frogged the Jaguar, and we were indeed backing the wrong horse.
Do you feel the Jaguar hardware was simply a victim of an ever changing industry at that time? It seemed ideally suited to Goraud shaded Polygons etc, yet the new benchmark was clearly texture-mapped 3D, as far as polygons were concerned. I read a quote from you saying Jaguar wouldn’t let you use the ‘tricks’ you might of, in terms of coding on PC and was around 10X slower to do texture-Mapping on, than Goraud shading and you’d hoped to use the DSP in areas to assist, but it was pretty much maxed-out.
I think the Jaguar was incremental, and that was ultimately its downfall; they hadn’t looked far enough ahead to see what was coming 3-5 years in the future, and by the time the hardware had been created and the console marketed, they’d eaten into a large slice of that 3-5 years. It feels like the Playstation in particular took that longer-term view and nailed it.
We didn’t perform that same critical analysis before working with Atari, as there was an element of us being ‘fanboys’ and hoping Atari could rekindle the flame!
One of your 1st projects for Atari was the conversion of Super Sprint to the ST, did Atari ever approach you to do any work on the Lynx at all?
Cybermorph started out on the Atari Panther console, can you talk us through how far along it got, your impressions of the Panther hardware and whether code was used from the Panther version on Jaguar Cybermorph.
Hmmm, interesting question. From memory, and it’s patchy here, I think Flare were working on 32-bit versions of their custom chips, which were Panther, and 64-bit versions that were Jaguar. I think Atari realised they could only support one release, and so plumped for the 64-bit chipset. I am pretty sure that Panther was dead when we started on Jaguar, or very soon after; I don’t remember the differences, but as the main CPU was 68000 in both, a lot of the common gameplay code would ‘just have worked’.
Cybermorph is one of my favourite Jaguar games! How did your company get the job of making the original pack-in game for the system, and what did you think of the hardware?
See above for how we got the Atari contract.
It wasn’t chosen as pack-in until the very last minute; Atari were going to pick Trevor McFur, which had been created by their internal team. I cannot remember when the decision was taken that Cybermorph would be the pack-in game, but I was onsite at Atari finishing the game with Stuart Tilley (designer). We were absolutely delighted, and also thought, naïvely that we ‘were made’!
I loved the hardware, and we could see many opportunities to make the game a lot faster if given time; we used those techniques on Battlemorph, and used the extra CPU time it freed up to make the game better.
How did you come up with the idea for the computer guide Skylar and the different things she would say?
We brainstormed a lot with the Atari team; I cannot remember, but Sean Patten (Atari Producer) was a big influence and so we probably came up with a list of about 50 things and then as reality kicked in, reduced it to the final 10-15 that we ultimately used (I honestly cannot remember how many are in the game now).
What did you think of the infamous Diehard Gamefan review of Cybermorph, and did you ever talk to anyone at the magazine after that?
If this is the one with Game Of the Month and scores of 99%, 96%, 98% and 98%? If so, then yes – Chris Gibbs met one of the team when promoting Rollcage a number of years later – I think they had offices in a Playboy building at the time. Apparently they had been a little high, and the psychedelic colouring influenced their scores…
Rollcage is a really fun racing game on the PS1! Where did the idea of being able to drive on the walls and ceiling come from?
That came from Steve Bennett (with a little brainstorming with Stuart Tilley) – they both loved Wipeout, and wanted you to drive cars that were so fast the downforce would keep you glued to any reasonable surface. I am sure that I’ve read somewhere that F1 could theoretically ‘do a Rollcage’, but I am not sure if that’s something we put out a long time ago…
How were you able to get all the musicians on board for the soundtrack of Rollcage?
That was with the help of Sony’s production team and marketing departments; the team we worked closely with to bring the game to market were as passionate as we were, and made it happen.
How come ATD closed in 2003, and what are you up to these days?
Financial troubles at our parent company, Kaboom Studios, contributed to by ATD, Silicon Dreams and Audiomotion.
On reflection, I think that at ATD we always cared about quality; most of our staff also cared about quality (it wasn’t 100%, and should have been; that was a flaw in our recruitment). We over-reached on Sydney 2000, which hurt us on Salt Lake 2002 – whilst we earned a BAFTA for Sydney 2000, SLC2002 was too small for the market. We had a troubled relationship with Lego, which I’d prefer to not discuss. I think the result of this was that in the delicate balance of quality versus time (it’s a triumvirate with cost/resource), quality lost out – and it’s a downward spiral when the most important factor is delivering on-time.
In publishing: everyone remembers a poor game delivered on time. No one remembers a great game delivered late.
For gamers: no one remembers a poor game delivered on time. Everyone remembers a great game delivered late.
When Kaboom collapsed, I set up Gusto with a number of people from Silicon Dreams, and became Managing Director. I left eleven months in as I missed technology, and joined EA’s North West studio – within a couple of months we were working on Battlefield: Modern Combat, and 11 months after starting it was on the shelves, with the North West studio having created the single player campaign to accompany Dice’s MP. I then joined Swordfish and worked on 50c Blood on the Sand, before rejoining EA in 2008 to work in their Partners team, which helps bring 3rd-party independent studio products to market, i.e. Crysis, Titanfall, etc.
Note: I still have our first ATD brochure here, along with copies of the Atari/Activision and Electric Dreams letters for Super Sprint. I can scan if you’d like…