Nice one. ;D That would no doubt be something original, and would inspire the story as you mentioned.
[quote]In an ideal scenario the phys could avoid a lot of duplicated calculation by feeding render with ‘easy’ data.
Train of thought: Objects are in an octree. The Phys moves all objects, checks collisions and feeds ‘ticked’ objects with proximity lists for the AI. It also feeds the renderer with a culled list, maybe not as tight as individual polys (but the phys will need these for collision checks anyway) but pretty tight. The renderer would then need to call the relevant objects with a getColour(textureID, u, v) call if procedural textures are used. Certainly with software rendering there won’t be CPU enough to feed both phys & render with raw data.
Speculative question: Can we build a Physrenderer? Code which can take raw model data, move it and render it in one fell swoop?
[/quote]
Hmmm. I am not sure. Most of the time during play, poly-poly collision is not neccissary becuase very few near collisions actually occur (except docked ships = turn off check). Every rendering frame though, a subset of ships / parts of ships need to be selected based on the user’s camera position. So you need fast approximate fustum culling added to the post collision pipeline. This results in a list of ship posisions within the camera angle BUT probably not their mesh ordinates in absolute space because the collision checks probably didn’t calculate them. Even when the poly-poly collision check runs it only checks a subset of all the mesh ordinates.
The physics engine uses mesh data totally different to how the gfx does. So all you save is the memory, not calculations except in very rare circumstances. I say get a renderer off the self. There must be an almost infinite scope for us to slow down graphics processing through a custom implementation. e.g. bad fustrum culling, inability to split meshes up according to fustrum, innefficient checking the polygon is facing toward the camera etc.
More realistic scenario: …The system administrator intern of one of the largest orbital research plants caused a extraordinary long STW garbage collection in the main flight control computers zetabyte file system. He used an arcane 32bit browser draggable java plugin for remote debuging on his yPhone? which caused a buffer overflow in the ultra low pause “garbage first” garbage collector. The emergency system (don’t panic MARK II) initiated a course correction to the last known maintenance safe point: CERN the current location of the system administrator intern. The last known log message on the interns yPhone? was “Full GC: promotion failed” short before the research satellite crashed directly into
6 of 7 of all incarnations of the company “Orange” (TM in parallel universe 1) released almost immediate after “the day on which the intern died” security updates for all 32 bit draggable java plugins right before all 32bit systems where prohibited by law…
I’m not convinced… Frustrum culling et al are all well-known and fairly :
simple operations (+no far plane!). We surely have the talent to write an efficient software renderer tailored to the specific scenario of space, don’t we?
After all, if we want to impress people we need to impress them. The game’s got to look amazing in software rendered - and that’s a problem we got to solve…
Depending on the target view port size a real time ray tracing render is probably viable for web resolutions.
With a real time ray tracer it is pretty easy to implement impressive features without adding much complexity and since such a render’s embarrassingly parallel nature it can easily use the extra horse power of multicore processers.
for single core processors expensive eye candy (HDR, soft shadows, etc) can be turned off.
I like the LHC idea, the game could be called something bold like “The Higgs Boson”. Keep the creative stuff going its cool.
[quote]I’m not convinced… Frustrum culling et al are all well-known and fairly Roll Eyes simple operations (+no far plane!). We surely have the talent to write an efficient software renderer tailored to the specific scenario of space, don’t we?
After all, if we want to impress people we need to impress them. The game’s got to look amazing in software rendered - and that’s a problem we got to solve…
[/quote]
We might have the talent but we might not have the time or the patience of the commitment.
Anyway, all I was saying is that you don’t gain much in calculation times by linking the renderer with the physics because poly-poly collision checks use a subset of mesh points which will almost 99% of the time will not be the same mesh points you need to render. You havn’t negated that point.
Whether we build a tailored software renderer or find one is a different decision. What we lose in 0.5% of performance by not having the physics strongly linked to to the renderer we gain by all the time another group have people have spent optimizing their gfx engine (hopefully).
[quote](+no far plane!)
[/quote]
If you get rid of a far plane (good idea) you better have a stratergy for turning a high polygon procedurally generated capitol ship into a lower polgygon model inversely proportional to distance (squared). Otherwise when a ship hyperspaces into a system, and looks in the direction of the space port, all ships + station + docked stations must be rendered at the same time and the system grounds to a halt.
Oh yeah, ray tracing does indeed solve this kinda thing. Maybe I am thinking too openGL. I forget how ray tracing works exactly. You shoot a ray for each pixel on the screen. See who it collides with first and where for a lookup on the texture. If you want shadows and lighting then you need to join that collision point with every possible light source in the scene (or sample based on distance) to work out the final color?
I suppose in a space game 70% of the screen the rays will collide with nothing. The worst case would be approaching a space station for docking where 100% of the rays collide with a ship and need lookups. Hmmm. In that case most of the rays will collide with the same polygons and use similar light resources so perhaps you can save alot of time then as well. Dunno. Well if anyone is a ray tracing pro then maybe its fine doing this. I have no experience.
The physics engine should be happy to do the ray tracing collision detection. We don’t need fustrum culling in that case for the first stage of sending out rays. Culling light sources by a spatial sphere is easy by physics engine for the second stage. All the clever stuff to prevent aliasing and produce soft shadows in a reasonable time through intelligent sampling is alot of work though for someone to do.
Tom
All this talk about rendering… Perhaps we can decide what the game should be about, and what should and shouldn’t be in the gameplay scope first? I really don’t care what method of rendering is chosen as long as the end product is fun. :persecutioncomplex:
Yeah, maybe we’re getting a bit technical - there won’t be a game without visuals though! I’ll branch the tech stuff off onto another thread.
@LHC:
Could it not have created a black hole (that grew over time) and thus swallowed the world?
Would give justification for:
- accelerated space research
- everyone leaving the planet (with all the possible factions)
- no earth (thus not having to model how the earth would look like in xhundred years)
Just a thought.
Perhaps we should start with a style/theme we would like to go for. I think it is very important to get this sorted up front and try to stick to it as much as possible. Here are a few sliders for people to consider:
Thematically:
Dark/Gritty============Clean/Advanced
(Aliens) (Star Trek)
Fantastic==============Realistic
(Star Wars) (2001)
Visually:
Realistic==============Stylistic
(Eve) (Darwinia)
I would personally vote for a dark/fantastic universe and visually highly stylized. Most space traders out there are of the realistic variety and it would be cool to do something a bit different and less boaring.
I tend to think that sticking to only one style for an entire game gets quite boring… e.g. quake 2. Having areas of different styles makes the game feel fresh for the entire game. e.g. half-life original.
So why not have all styles represented but have them segregated into different areas.
One style, “Fantastic” (Star Wars) is hardly restrictive though, so why should it get boring! But I see the point. There is no reason not to combine different thematical themes though. Different worlds, different stories.
However, visually, it would would be good to decide on a single concept imho. (And no, not talking about which renderer!!).
And last but not least, it should all seamlessly fit together somehow. I’d hate to go from system A to system B, and feel that I moved from one game to another. There needs to be some herring.
I like dark/gritty fantastic too. Also when thinking about developing a sector I was thinking about uylesses 31. The 80s cartoon retelling the story of the Greek hero Uylesses’s adventures – set in space! It was an awesome cartoon. Zeus would appear as a mildy transparent enormous bust in the spacedrop and say “Uylesses! you have angered the gods! You are banished to the kingdom of Hades for all eternity! which was just a sector of uncharted space” And each episode was Uylesses running into a planet and meeting other people who were punished by the gods. Of particular note was Cisiphus. He was found on a desert planet that was shaped like a funnel. He had to roll boulders down the funnel. When Uylesses arrived, ciciphus tricked him into doing his punishment for him. Uylesses prompty fell over and fell down the hole at the center of the planet. Here the boulders were crushed up by enormous hi tech equipment (cue uylesses dodging the crushing equipment). Then the rubbish from the boulders were put into a boulder mold, and then transported back upto the plannets surface! Haha. So much of the gods powers were mearly technological manifestations. Damn cool though.
Erm so yeah. This is not a suggestion. Just chat about my favorite space context. Its on youtube, its worth seeing if you have never heard of it. I was only about 3 when I watched it, so unless your 28+ its likely you will have never seen it.
http://www.youtube.com/results?search_query=ulysses%2031&search=Search&sa=X&oi=spell&resnum=0&spell=1
Dude, of course I saw Ulysses! Although I’m not sure writing the universe around the series is really viable, we definitely got some cool idea for unique systems/planets 
Why not? Spaceships vs Ancient Gods gets my vote!
(I watched ‘Ulysses 31’ fanatically as a kid, but I can’t remember any of it now. Except, oddly, the theme tune.)
Simon
puts on deep voice “Dishmoth! You have angered the gods!”
Lol! BTW If you want inspiring space animations - check this out.
Speaking of space ships and Gods. People here read Dan Simmons? Hyperion, Ilium, Olympus. Amazing books.
Back on topic… I started writing a strip of history in relation to plot and story. If I twist my head and come up with some backstory for earth that’s a little more original (that seemed to be the general consensus I think), do you want me to do it?
If you do, I ill definitely do it. However, this implies that what gets written is written in … clay(?). Not as permanent as written in stone, but harder to erase than on paper. The point is, if I take the time to write up more documentation, I’d like there to be some threshold to “overthrowing” it afterwards. I would try to split the writing, so that individual, less universal stories can/must be written for the different systems.
If you don’t, then that’s fine too, but someone needs to pick up the ball in this case. New 5-liner ideas from everyone isn’t going to write itself into anything
Cheers.
Yeah, we should work out how creative decisions are made. There is nothing worse than someone putting in time inventing great idea + fleshing them out into dialog for people just to decide that we don’t like the idea anymore.
Maybe we should call a vote for the proposed high level human race contexts (inc. a none of the above option if we decide we need more time to think). I think whoever puts forward an idea for the vote, should then be responsible for managing the production of story boards and history etc. if it is chosen (so Greek mythology is not a genuine suggestions unless anyone else wants to champion it). I think the creative manager’s job would be organizing a wiki of creative contents and/or launching votes.
Once that vote is finalized then thats that. So we should not rush to get the vote up or anything.
Might to time to transfer some of this to the wiki, maybe formalize agreement on some of the ideas here?
Kev
I’m not too sure about the wiki - why put stuff in a hard-to-get-to place? (to join this wiki first get a gmail address, then email that to &c, who’ll then &c) I’m also worried about all the (obviously newbie) posts on the wiki saying ‘let me join! I’ve got a CV to invent!’
We’ll need a CVS repository & a forum… but we’ve already got a forum here…