I want your bodies

tpe

Hi Guys this is a mixed post, request, 3d and GD.

the quick version...

Has anyone any experience with producing content for many platforms at the same time?

What went wrong? and how could it have been done better?

The long version...

The reason being that publishers are wanting more and more out of the character depts. They want simultainius release on all platforms. That can mean making the same content many times over, shunting up production costs, especially now as currently produced games can be aimed at gameboy (down to 150 tris per character) PS2 (a thousant or so etc), PSP (a little less), Xbox (normal mapping starts here so millions of tris), Xbox 360, PS3, PC (for a single release a pc game may have to suit just as many different specs, to cover middle to high end as the rest together). etc. etc.

Are they asking too much?

So i would like to put together and test a method for piplining workflow for a current game project and see if there are any ways of streamlining it.

Starting from the high poly down to PS2. so a way of being able to produce game content easily from the high poly that will make it easy to get characters of whatever resolution needed without having to make the same character too many times.

I have been thinking about using a standard mesh that can easily be changed to match the high poly from quads that can have its polygon count changed easily (strangly enough quads work well for strips when you put the edges in) to avoid having to uv map too many times and then use complete mapping from kaldera or something. it would also heelp with the riggers so they dont get too many unexpected verts flying all over the place.

Otherwise a form of auto lod for anywher betwwn 10k tris and 150, that respects UV seams but this would have to be tested too. Microsoft were working on something, has anyone heard of any comercial pacages that are worth anything?

So the request, if there is anyone who would send me their high poly models it would be of great help to me so i can see if it will work on different characters. That is if you dont mind parting with a .max or .obj to "timothye at ioi dot dk" I cant really offere anything in return except sharing any results (if) i get.

many thanks

tpe

KDR_11k

That's usually set up in tiers, there's the GBA/DS tier and the current gen tier. Usually the main team will make only one version (the home console version that is aimed at the weakest system included and given some additional effects for the bigger systems) and the portable version is outsourced. For the Splintercell games the XB/PC version is made by the primary team while the PS2/GC version gets made by another one. With the next gen upcoming perhaps you'll see another team working on a next gen version but I've neverseen a game where one team did all the versions though Rick sounded a bit like all GTA versions (including GTA Advance) were handled by their team.

tpe

Per, Yes making one or two low poly versions is pretty trivial, but if you have 140 characters in a game at a day each and need them at 10k tris 2.5k tris 1ktris and 500k tris to cover the two xbox and 3 PS consoles and the pc then you are still looking at nearly 2 years work, so it isn't totally trivial.

I have seen one that was quite impressive some microsoft guys were showing off at an expo but didnt get the details at the time, also i beleive that bothe the Unreal 2 and 3 engines use auto LOD and the characters look pretty good so it isnt impossible.

It is also now that we should be looking ahead to changing the production line method, and as things improve it will be necessary to cread a divice independent production tool. not necessarily for this year, but it is on the cards.

KDR, we have been using various studios for various pilot projects and had mixed results. It may be that soon a development studio that is able to compete against outsourcing will be no more than a high tec design team that coordinates various sub studious or model banks (the equivilants of actor agencies or special effects houses in films) for resources, and that appears the logical end and the way it is going, but i beleive there is still some time left where development studios will still be making their own content, and to be able to cope with the number and diversity of end user divices the amunt of resources they have to coordinate is growning very quickly. Anyone who isnt able to change gear is going to be left behind or loose out to china. BTW we dO all the hitman and Ffreedom fighters on three platforms our selves (atm ).

tpe

Yup precisly thanks, thats what i was trying to get at , any and all ideas welcome.

I have been looking at various cages, with geometry created to be easily reduced but still rigged and textured easily for the various platforms, within each platform there is still a need for optimization tho (i,e just because you have a character into a platform does not mean that it nolonger needs LOD), and i am trying to find different methods for that two, thats why i was asking what other experience was, where they went wrong and how they could improve it, i didnt think that we were the only ones that took on such large projects tho...
They are all simple bipedal humans, hence the cage could be easily modified, small things like ammo pouches could conceivably be added later. I am expecting long jackets to be one of the more difficult things to teal with, that and womens clothing that varies very much(you see where i have been going too now ?;)).
That is also why i was asking if anyone could provide me with some high poly models to try it out on, the ones i have are all a bit too similar and broken up into segments such as torso etc and it is not a good real world test, where i expect the pipeline to have to work on characters produced by artists that have followed a concept that is not specifically designed to be easily remodeled at different resolutions.

tpe

p.s. this is interesting (http://www.multires.caltech.edu/pubs/)

tpe

Thanks per, i am trying to make a prototype for it now, do you think that a relative endo morph will be able to displace a mesh that has been edited after the construction of the morph well enough, in lightwave this is easy enough, but how is it in other packages, is there one that is easier than the rest and seem to remember that it can mess up max morphs(I dont have to use any specific package or finished work, just make an example of the concept)? I only have the weekend to make a preliminary investigation so i am a bit pressed for time...

tpe

tpe

thanks a lot. Yes, I have a hangover so know how you feel, go bedering (i oslo?)

tpe

Rick Stirling

I have no idea who did GTA adnvance - its wasn't R* North. The PSP game is a join venture between us and R* Leeds, but nearly everything was done by them as far as I am aware. So thats 3 studios.

The previous Xbox GTA games (3 and VC) were handled by R* Vienna.

Eric Chadwick

I haven't read this yet, but it might offer some ideas.

The Scalability Problem
http://www.acmqueue.com/modules.php?name=Content&pa=showpage&pid=118

tpe

Thanks very much for the help all.

Eric, He puts in some interesting ideas about cloth trees and fluids and how they should be allotted different processor threads for diferent specs, he also makes some nice generalisations and provides references. So i have a couple more ideas to throw in if need be, thats always good.

So far i have a reasonably half baked prototype using our old work, which is a bit of a shame as i (and everyone else) knew more or less it would work anyway. I have also some ramblings about workflow and general stratagies to be able to wing it a bit, but most i have a reasonable basis to find out what i am missing, so its a good start in that respect. Still needs a lot of sorting out, a harch round of questions and answers should help finish me or it off tho...

thanks again, and if anyone ends up doing this sort of thing please let me know (or if you have any good ideas or experience where it went wrong or right), this is only the very start of what is going to be a hell of a lot of hard work for a good time to come, so all input collaboration is greatly apreciated.

tpe

btw rick if you need any older (lower poly concept) i can probably send some your way, ill have a look to see if there is anything at work that you could use on monday)

tpe

Many thanks Per, once the turbo mesh is off it it looks like it may not be too difficult (wishfull thinking), i am going to try and stick it through the xsi one which preserves much better than multires, but if it doesnt get anywhere at least i can use it as an example of what we shouldnt try , nice model to have lying around, ahhhhhh i wish we had the chance to do monsters instead of guys in suits...

I thought you were in the US or UK or something, if in Oslo are you at funcom (we have a few of them here), or do you know the guys at artplant (hope they are still going)or anything? whatever hope its fun...

tpe

PaK

Multiple teams is a good way to go. A Sony, MS and Nintendo branch for each one, share certain aspects of the development process that don't change all that much (some parts of art and the overal game design ect ect)

1) a core team in each dept that would be 'shared'. Like for art, shared character designers and modelers.

2) hire technical artists and some extras for each branch to fill out the system spacific needs.

3) Snych every so often to make sure everyone's on the right track, so the shared personelle know what the branch guys are doing with their stuff, and vice versa the branch guys know where their raw files are comming from.

This way you can share as much data as possible and still give each platform the atention it deserves.

-R



...That's what I think anyways.

-R

okkun

Polygon Cruncher for Max. My trial ran out so I couldn't bring it back to max for the render. It does a much better job than Multires IMO. Check out the nice tesselation on the hose. This is at about 10% of the original polycount.



Polygon Cruncher page

Rick Stirling

Polygon Cruncher is superb for making LODs in crunch - it keeps UVs and weighting, and you can choose loops that you dont want effected

okkun

Yeah someone download the trial, bake it and post a pic. I can't get the thing back into max without buying poly cruncher (And i would if I had need for it)

tpe

That looks pretty good, turn a couple of edges and it would probably be pretty cool. For IGI2 we actually managed to use multires a bit, it kept UVs too, this looks far nicer tho. Per i am sorry i havnt got back, and it is not a good time for me to explain how it went. although bits went very well and are going to be used, it got bodgged down in another point of discussion about changing apps, and personally atm i am rather PO with loosing a fair deal of work because someone got a free drink and signed up on a beta test at e3, so shouldnt really comment on anything at the moment for fear of getting my self in trouble and then there are all the people who want holls and.. etc ... etc so embarassingly little to report back .

tpe