Previously: RealiMation the Datapath Years, Part 4

Looking back I now realise that RealiMation never sat well within Datapath. Their graphics card business had moved on from 3D to multi-screen displays and in their eyes we were a distraction. Brett had plans for RealiMation and didn’t want to see it shut down. I was unaware of all this until Brett broke the news in August 1999 that he was buying out the other directors share of RealiMation and going it alone. The next couple of months were tense as we couldn’t tell anyone else what was going on.

Eventually the rights to RealiMation were bought from the other two directors and a new company was set up. We moved into some rather grotty offices further down Alfreton Road – I won’t tell you where for fear of legal action – you can never be too careful ;). Rob Love decided that this would be a good opportunity to take some time out and do some travelling. He eventually ended up in Australia working for one of our clients – Pineapplehead Computer Animation. The rest of us, Greg, Mike, Larry and me stayed and formed the core RealiMation team.

We had contract from Real3D to develop software visualising the 3D data they were extracting from multiple 2D photos and an ongoing one from DaimlerChrysler for developing their factory viewer. These two formed the main initial income flow of the company but we also had contracts from some of our defence contractor clients to either upgrade existing systems or develop new ones. I was developing a system that read real time data from an external source and kept the display in sync with the physical equipment.

Unfortunately the Real3D work petered out as they ran out of money, but the DaimlerChrysler project would eventually become mantra4D after many iterations and changes in direction, both in technology and name, but only after we were taken over by Criterion.

This press release shows what Rob ended up doing in Australia. The work had started before the split – as indicated by the date of the release, but only came to fruition after Rob reached Australia at the end of his travels:

Green Reader Interactive Brings Golfers Perspective to Tournament Coverage

RealiMation Chosen Tool for Green Simulation Application

March 1999, Derby England – Pineapplehead Computer Animation, of Melbourne, Australia have selected RealiMation as the development tools for “Green Reader Interactive” – their 3D golf course visualisation program for broadcast TV.

Peter Lamb of Pineapplehead said: “Green Reader represents a giant leap forward and has already been used at a number of major tournaments around the world. For the first time in the history of golf broadcasting, the viewer can now see exactly what each player is confronted with when they reach the green.”

Green Reader Interactive displays accurate computer generated models of each green from the golfer’s viewpoint. The green can be rotated through 360 degrees in real time while commentators highlight the individual borrows and contours of the green, to give the viewers a better idea of the skills involved in making the putt.

Pineapplehead’s programming team visits each championship green and meticulously map and convert them to a compatible 3D model. The green along with the camera controls are enhanced using RealiMation, and are then output from the RealiMation application directly to TV output via a high performance 3D graphics card.

Pineapplehead develop a wide range of 3D applications including 3D visualisations, web pages and broadcast system. Green Reader Interactive is already in use around the world by leading broadcasters such as ABC USA, Seven Network Australia, Network Ten Australia, FOX Sports and many others are set to come on-stream throughout 1999.

I think this might have been used by Sky Sports, but as I’ve never been a subscriber I never saw it.

Rob returned from Australia in the late spring of 2000 and was offered his old job back – which he accepted.

It was at this time that I started development on the passive stereo and software edge blending technologies. The plan was to replace expensive custom built filters with software. At first we thought it would be easy, but unfortunately the human eye is so good at seeing edges it sometimes sees them when they’re not really there. The first attempt was to put a semi transparent (50% alpha channel) 2D polygon over the bit of each channel that overlapped. While in theory this would give 100% brightness over the entire screen – instead of the 200% brightness over the overlap, you could still see an edge where the overlap ended, especially if the projectors weren’t exactly aligned or the image was skewed in some way. We tried several schemes, but the one that worked the best was a series of polygons going from fully opaque to fully transparent in a sine wave pattern. However, as the polygons were generated by the master system it did increase the data that had to be sent across the network and hence had a detrimental effect on the frame rate. We never got round to generating the polygons locally – which would have been the logical thing to do – as there was always something more important to do.

A new hire, Simon Cooper, who Greg and I worked with at PAFEC, worked on a module for Railtrack to model railway lines and position signals – this was after an accident caused by poorly sighted signals, so they were looking for a training aid and (if I remember correctly) system to help them place signals in better positions. The system was demonstrated but never went live.

We installed a multi channel system at the University of East Anglia (three channel mono) and one at the University of Derby (two channel passive stereo). We would go on to use this latter system to test edge blending and some of the tracker devices we had to develop interfaces for. As it was the same configuration as the systems installed at DaimlerChrysler it was a useful test bed.

December 2000 saw the release of version 5.0.

Other projects

• Mike’s target drone telemetry simulator – Jindivik. Developed the display of 2D labels to show information about the drone on top of the 3D image. This code was later reused in mantra4D.
• Rob Love prototyped real time shadows, but this never developed further and I no longer have the code.
• City of Rennes – a virtual copy of Rennes in France. This was the spur for the first attempt at paging data using references and multi-threading. It wasn’t a great success, but was adequate for the size of model involved here.
• SGI based air traffic control simulator for a company in France, whose name I now can’t recall. This was a continuation of work started whilst we were still at Datapath.
• “Space to face”. We were talking to the US defence industry about a system where the user could zoom in from space to a highly detailed model. DaimlerChrysler wanted something similar to link all their factories spread across the globe. This never came off, but was why we were all so interested in (and jealous of) Google Earth when it launched in 2005.

Towards the end of the year we were bought by Criterion Software. Looking back I think this was always the long term goal that Brett had in mind when he bought out RealiMation from Datapath. We had a long term relationship with Criterion, we used their Renderware system as the fallback software render in RealiMation for example. Brett approached them with the ideas and prototypes we had for RealiMation Enterprise (called at that time Navigator if I remember) and they were interested enough to first consider buying a license, but then deciding that they liked product so much they ought to do a “Remington” on us and buy the company.

The deal was effectively done by the September of 2000 – we were invited on a Criterion weekend away in Malaga, Spain. We flew out separately to the rest of Criterion on the Friday from Luton Airport on EasyJet – an interesting experience! We were supposed to get a minibus to the airport, but I ended up driving (as I had the 6-seat Fiat Multipla by then) and worrying about getting fuel on our return as it was in the midst of the fuel duty protests of that year. The Friday was excellent as we arrived mid-afternoon and spent the rest of the day drinking by the pool in the warm late summer sun. Everybody else arrived late on the Friday night after we had eaten. The Saturday was spent on various team building activities, which wasn’t that much fun and made worse as it was very hot. Though we were close to the pool so cooling down was easy!  The party on the Saturday night had free booze, but I didn’t overindulge. Perhaps I should have, but I didn’t want to be hung over during the flight back to Luton and I had to drive home too. I ended up going to bed relatively early (about 1 am) and as a result I missed the shenanigans that went on into the early hours of the morning. We left earlier than everyone else on the Sunday to catch our flight. I managed to find a petrol station back in Nottingham that sold me £10 of diesel so I was OK for commuting for the next week or so. Fuel was so much cheaper in 2000!

Summary of improvements to the API:

• Very large image generation. This used the multi channel algorithms to tile a view of the required resolution from many smaller views. Each was rendered (off screen), captured and then all the images were stitched together to produce the final image. This allowed DaimlerChrysler to produce poster sized screen shots for printing. This also highlighted a long standing bug in the multi channel code that had been affecting the display very subtly but we had put down to projector alignment.
• I added PNG image support; both read for textures and write for screen grabs and printing.
• Created of a suite of user interface components, for example a RealiBase file dialog that read the file header for the version, comments and snapshot, which were then displayed in custom fields.

Personnel Changes – Larry Binks (left), Rob Love (developer, returned), Simon Brentnall (developer), Andy Greaves (sales), Claire Webb (PA), Simon Cooper (developer).

Next time: [The Criterion Years]

© 2015, Chris. All rights reserved. If you republish this post can you please link back to the original post.