How Heroes Over Europe, the sequel to Heroes of the Pacific, blew its budget, slipped 18 months and sank a company
- Justin Halliday
What? 8 weeks for creation of a single plane asset. Seriously. Can you elaborate why it takes ~320 man hours per plane. That seems insane.
From memory it went something like:
* Basic plane itself: a week or two
* Variant skins/features: from memory, all aircraft variants were actually modeled correctly, as opposed to just consisting of a skin change. That meant new work on variants to include any extra antennas, extra gun ports, etc.
* Divide plane into various "segments" for destruction (which varied on hit location - bit hazy on this, but I think that the explosions and "cameras" for each aircraft were set up separately to be pre-canned, and as such needed individual work per plane)
* Creations of high-detail "Ace Kill" points (stuff like engines, ammo stores in wings etc often had to have INTERNAL elements modeled to support the Ace Kill feature - because you could shoot an ammo store in a wing and cause it to explode, for example, that store needed to be modeled because you saw it up close and in slow motion)
* Landing gear setup/animation rigging
* Probably a few other things...
All added up pretty hugely.
I am not sure where Justin got 8 weeks from. It actually took 6 weeks to make a plane from start to finish.
2 weeks research and modelling.
2 weeks texturing.
2 weeks LODing, creating internal and damage elements and damage texture, animation, setup and export.
The planes for HOE where infinitely more complex and much higher res than the planes for HOTP. 15,000 verts for fighters (bombers 20,000 verts) with hundreds of nodes for the top LOD, as opposed to 5000 verts and much more simplified hierarchy for the HOTP planes.There were 5 LOD's, with the internal components included for the damage system and Ace Kill.
To do all this, six weeks was pretty tight in the end. I am very proud of the planes in HOE I think that the fact that they were not outsourced shows in the high quality of the finished assets. The planes also ended up being the smoothest and least troubled part of project and were always delivered on time.
I never made planes on the project, but the attention to detail was astounding, and a real testament to the quality of the artists working on them.
I think that in hindsight, planes became the target of outsourcing maybe because they went so smoothly, and were *somewhat* independant of the pipeline (in comparion to say the terrains, etc). If you had to package up anything and send it off, planes always seemed to be a candidate.
I guess it depends whether you actual want to retain a team of senior artists, or whether you'd prefer them to be outsource managers, and see how long they stick around.
In the current climate in games, there seems to be little scope for hd projects that value a high level of artistic quality, or where decent senior artists are actual valued, or have the sort of work that really shows off their talent, which is really sad. Good art costs time and money, and if you're making a plane game, I would suspect that planes would be a good place to consider spending that effort ;)
Thanks to Justin for the great talk, and respect out to all the Transmission Games guys, from all departments.
Good talk from Justin - and nice job summing up the issues - it is a real testament to the hard work of that team that the project was finally delivered given the challenges.
In regards to Justin's thoughts on outsourcing I disagree. We keep talking about AAA titles in Melbourne, and for anyone that went to the IGDA reboot and saw Paul Callaghan's excellent and honest appraisal of the Melbourne industry, it is obvious that we're not hitting that level, nor have been for a while.
The planes in Heroes Over Europe are at that level, and having worked with various outsourcing companies, and some good ones at that, I've not seen that sort of quality on an asset of that complexity delivered complete and correct within a 6 week period for the cost of 1 artist. The management overhead, re-work & feedback, training in visual style and efficient creation & export, cultural/language miscommunication & finally quality... it only looks better on paper...
Outsourcing is a great solution for many things given the scale and expectation of current gen titles, but I honestly don't feel you will ever get AAA out of 100% outsource, especially on your core game assets.
This session was the highlight of the conference me. An absolute eye opener. Extraordinary.
I'm certain this footage will be the *must-see* for all CEO's and upper management staff for all developers in the local industry. Huge kudas must go to Justin Halliday for explaining the whole situation so thoroughly. It was unfortunate that there was such little time left for questions and answers since I'm certain that that could have filled another half hour, at least.
This was a superb, even entertaining, breakdown of what was obviously a very complex and highly challenging (and confusing) process. It was honest yet never condemning. Justing apportioned cause and even responsibility but never blame. This is an essential study from which studios (and publishers) could and should learn a great deal. Definitely the event's highlight.
I'm reading the post-mortem at Gamasutra on Torus's Scooby-Doo! First Frights, and I think these sorts of insights would be great if they appeared more often at GCAP. Perhaps having at least one post-mortem presentation could be a regular thing at the conference. I can understand that there may be issues with publishers, and no one wants to air out the major problems at their studio (and in any case, post-mortems aren't usually as straight-to-the-point / blunt as the Heroes one), but for the sake of learning and even getting feedback on solving some of the problems presented, I think it's worth it.
For example, I don't think there was one person in the conference room whose jaw did not drop when Justin showed the amount of image stamping required by artists in the London map for Heroes over Europe. That was completely *insane* (even if later on they could use groups of stamps, it's still quite a large amount of work). But how does a problem like this get solved, particularly by other studios who do these very large scaled maps that are sized in the tens of kilometres? I know games like Quake Wars was delayed a year, and the mega-texture technology for that was partly to blame (giving artists the freedom to stamp unique textures everywhere means a mammoth amount of extra work - id's Rage has got to be going through the same issue), but with maps in these days getting even larger and larger and more detailed, how does this issue get resolved more efficiently than throwing a mega-tonne time and artists to manually stamping these images everywhere.
I don't think the megatexture stuff added much extra time to Quake Wars at all, in face most of the terrain textures were all rebuilt not too long before release relatively quickly. It's one of those things where you can just push the generate terrain button like with older style methods and be done with it, and then add more details if you have time.
Now with the all encompassing method being used by Rage, that is probably an issue. Although you can still just texture things conventionally and have all the megatexture be generated off that adding no extra work.
More information about text formats
SUBMIT NEWS | CONTACT TSUMEA | TSUMEA WIKI | SITE MAP | OzScene - The Australian Amiga Demo Scene Archive
tsumea rss - The Cluster, Serviced Offices, Virtual Offices, Coworking