Betting sounds cool but I think it might limit the custom content.
First, how would tyou aquire it in the first place, bet the maker of the content? Like TF2 you'd have to download it to see it, so you'd already have it on your comp.
If you bet the maker and won it they would still have it, not like you'd have the only one, but you'd have to earn it. What if you couldn't find anyone to bet the item you want.
This also brings to mind achiement servers OK, bet me the ferrari, then just don't cross the finish line, I'll win and have it.
----
Second, at the rate it's going there will probably be so many custom items once the game is released that if all races are 8 people (I think there are already 8 stock cars- I have 2 working already, there are probably 10 more in the thread) there isn't alot of chance you'll see the same car on track unless there are just favorites [see earlier post about who gets to play sock monkey ] )
On top of that each car has 1-6 colors to customize, so even if 8 people chose the same car chances are pretty good none will look alike.
Then add character to equation (about 6 right now)
Accessories (With the stuff that's finished now it's around 25 I think)
Hats (12+)
So I think the point of wanting to have something unique is covered.
I do see marsh's concern about porno be legit. This happens in TF2 every once in awhile. I think for the most part it would be up to servers to lock out content, kick players.
Would be nice to have some kind of official rating 'tag' that items could get somehow.
Maybe there could be a setting in game that defaults to a random generated character if another player has an item you don't have. No forced DL's. You go to sites that host the files, look at pics, vids, see ratings and download based on that. The Steam User group could be a great place to discuss this stuff maybe.
Then if you don't want the porno car you wouldn't be forced to download it and only the people playing it would see it. Players without it would just see a random 'stock' car.
Being a stock car would eliminate any problems with obscene content, it would also help avoid clipping issues with wacky content.
Randomly generated would help keep it fresh even if you personally never DL'ed anything. (It would suck if all players defaulted to blue race car/blue helmet, hard to tell apart)
randomly generating stock items for people playing with custom items where the player has turned the visibility of custom items off would be a good idea I think.
Personally, I like the 2nd method more - I think it is more elegant and would streamline the use of items more.
But the biggest factor with it, is that it would be a lot more work to build and to support. It would also be a pretty big choke point for service for the game, I would hate for something to happen where either the database goes down, or we can no longer support it, and it pretty much kills the game. We value self-sufficiency a great deal. The most realistic option, at least in the short term seems to be to proceed with method 2, and try and fit as many of the features of method 1 in as we can, especially over time.
About the betting thing, that sounds like inventing a whole different game inside the game - we have thought about doing something simple to limit which character you can grab from, like you can only grab items from people you finished above - or only the winning team gets to grab from the losing team or something like that. Stuff like that will be something we just try and see if it feels ok.
ok, I have re-worked my model SDK into all .obj files and created a wiki page for it. I exported a few more reference files, including some bounding box limits. It can be seen here:
I have spent the last few hours thinking about appropriate polycount limits for each item, trying to factor in what my own models are, as well as where I want the game to fall in terms of graphical standard vs. performance, and these are what I have written down:
from lowest to highest:
wheel: 1,000 tris
hats: 2,000 tris
accessories: 2,000 tris
kart: 3,000 tris
character: 4,000 tris
with this allocation, one player's total maximum tri count if every item they were wearing had the max number of triangles - would be 15,000 tris. (4x wheels) I would imagine a maximum number of characters you could conceivably see onscreen at once would be around 16.
I do not have a carnal knowledge of pc hardware, but it seems that these amounts should be a decent balance of giving modelers enough room to play, while not having to be a triangle crunching master modeler in order to make something, while still being low enough for good performance on hardware from the last 4 years.
I am also hoping to have some auto generated lod models which would significantly reduce triangle count on karts not in your immediate vicinity. Also I think it will help that we aren't doing anything yet outside of fixed function effects, so no crazy shaders that take lots of passes.
What do you guys think about those numbers? I would like to hear your opinions more from a technical performance perspective, as most of us here would probably agree they are generous enough from an artist perspective.
Hmmm, too much talk in this thread and not enough OVERSIZED CLONE TROOPER HELMETS!
That was a fun little project. On a side note, I have everything set up in metres, and imported the right files, but the export was too big. I had to play around to get it the right size.
Hmmm, too much talk in this thread and not enough OVERSIZED CLONE TROOPER HELMETS!
That was a fun little project. On a side note, I have everything set up in metres, and imported the right files, but the export was too big. I had to play around to get it the right size.
Hope I didnt break too many copyrights
hmm, are there any settings in the ogre exporter you are using where you can specify what unit size to use?
Just went through and tested the size of all my items zipped with normal compression, and these are the size limits I am thinking about using:
Some of my items didn't fit those numbers when zipped up, but for those that were over I was able to bring them under the limit by making sure my colormasks were saved as grayscale as well as converting some of the lum .png files to .jpgs
3k triangle count for a kart,- is that certain for now?,- because if so I could add looots of xtra details on my batmobile and propably the others but most of them are already UV layouted.
I think that's looking pretty good, will have to zip my items seperately, they run the gamot from animated karts to small static mustaches.
The poly count is awesome, I was hoping for that much, more is overkill I think. The textures are small enough that it's hard to pack tons of details into them.
I have even used a 8x8 pixel solid color for 2 colors on one object. I just gave each object it's own material and used that same 8x8 for the colormask.
Smirnoffka,
That's awesome. I uploaded some correct size 3ds files to the Zero Gear forum. I'm gonna put them on the Wiki here shortly, will probably resize all the bounding box.obj's first.
Seems to me that the .obj files imported into Max are about 2x the size of correctly exported Max objects.
Its not really that much of a problem to be honest marsh, just took a little fiddling. Now that I have a reference that works for me, it will me much easier. I checked the exporter, and everything is in metres, so it must just be my computer for some reason, some thing I should have checked but didnt.
Edit:
Just saw your post baddcogg, and it seems that that was the same problem I was having. Must have been a max issue. No worries anyway, I have it figured out now and can make a shitload of cool hats.
wish I could figure out a solid way people could import my .obj files and keep the scale intact all the way into the game engine. It seems like everyone with Max is having the same issues.
Also cog, it is much cheaper for the engine to have models that use only one texture. For every separate texture that is used on a model the engine separates it out as a separate model group and has to render that batch of triangles separately.
Its not really that much of a problem to be honest marsh, just took a little fiddling. Now that I have a reference that works for me, it will me much easier. I checked the exporter, and everything is in metres, so it must just be my computer for some reason, some thing I should have checked but didnt.
Edit:
Just saw your post baddcogg, and it seems that that was the same problem I was having. Must have been a max issue. No worries anyway, I have it figured out now and can make a shitload of cool hats.
AWESOME helmet, i oly have two TINY TINY TINY little nitpicks with it...
1. the nub on top is the middel can't be colored (when making a tie pilot the helmet is black.)
2. there is a model seam down the back that could be welded shut .
Also cog, it is much cheaper for the engine to have models that use only one texture. For every separate texture that is used on a model the engine separates it out as a separate model group and has to render that batch of triangles separately.
So you mean each material in the mtr file is counted as a seperate tex.
That's a bummer. I can try to minimize that alot but somethings have to be split (on my T-Bucket) the double sided springs have color and transparency so that needs to be one, then there's the paint, then there's the chrome alpha which is invisible if used with the same shader as the color/alpha, then I wanted no-enviro for a few things. I could live without the last but that's still 3.
I only have one colormask though, so that's fewer resources that need loaded then if I had one shader and 3 colormasks. I wonder if that helps at all, probably at game start but not during game?
My fro uses 2, one for the pick and one for hair (I don't want shiny hair). But only uses a tiny 8x8 pix colormask for both color changes.
*double edit*
lol, just noticed that badcog himself posted that
Yeah, and what I got there was that if it is a non-animated (non-skinned) object X-Form should work to reset scale, but I couldn't get it to.
I had to make a box, attch the mesh to it, delete the box and scale was reset. They said you have to do it this way for a skinned object, but I think that would mess up the vert wieghts.
So like I suggested on the Wiki: model, scale, attach/delete box, skin to one bone, import test.
Once scale is right then rig the whole thing and animate.
I made the mistake of skinning to several bones and animating, then trying to fix scale and had to redo it all.
So you mean each material in the mtr file is counted as a seperate tex.
That's a bummer. I can try to minimize that alot but somethings have to be split (on my T-Bucket) the double sided springs have color and transparency so that needs to be one, then there's the paint, then there's the chrome alpha which is invisible if used with the same shader as the color/alpha, then I wanted no-enviro for a few things. I could live without the last but that's still 3.
I only have one colormask though, so that's fewer resources that need loaded then if I had one shader and 3 colormasks. I wonder if that helps at all, probably at game start but not during game?
My fro uses 2, one for the pick and one for hair (I don't want shiny hair). But only uses a tiny 8x8 pix colormask for both color changes.
Yeah, and what I got there was that if it is a non-animated (non-skinned) object X-Form should work to reset scale, but I couldn't get it to.
I had to make a box, attch the mesh to it, delete the box and scale was reset. They said you have to do it this way for a skinned object, but I think that would mess up the vert wieghts.
So like I suggested on the Wiki: model, scale, attach/delete box, skin to one bone, import test.
Once scale is right then rig the whole thing and animate.
I made the mistake of skinning to several bones and animating, then trying to fix scale and had to redo it all.
maybe then if we could get copies of all those .obj files as max files at the right scale it would save people all that bother?
About the LOD system : do you already have it working? The reason I ask is because I've yet to see a really good auto-lod generator that doesn't require a mesh to be made in a really specific way. I'd suggest that you allow artist-created LODs as well. The auto-lod could be made to kick in if it doesn't find a hand-made LOD, getting the best of both worlds maybe?
For example, say you want four levels of detail altogether. The artist could make all of them, just the highest level or the highest and second level. Whichever levels are missing are filled in by the auto LOD. This would give artist control over the most important LODs (the ones nearest the camera).
About the LOD system : do you already have it working? The reason I ask is because I've yet to see a really good auto-lod generator that doesn't require a mesh to be made in a really specific way. I'd suggest that you allow artist-created LODs as well. The auto-lod could be made to kick in if it doesn't find a hand-made LOD, getting the best of both worlds maybe?
For example, say you want four levels of detail altogether. The artist could make all of them, just the highest level or the highest and second level. Whichever levels are missing are filled in by the auto LOD. This would give artist control over the most important LODs (the ones nearest the camera).
No, we don't have an lod system working yet. I would have to figure out how the automatic system works before I made any decisions about that. The lod range would probably be pretty far away. If you didn't force people to make the lods, I doubt many people would make them besides the professional type guys on this forum.
thanks baddcog!
Put up a post on the 'ol blog - http://nimblebit.blogspot.com/2008/08/zero-gear-marks-1-year-of-development.html - has some shots of everyone's stuff in here. I just grabbed some things really quickly, sorry if I missed your stuff (though I did skip over a few for possible copyright reasons, don't want to use those to "advertise" the game) I also linked people to the thread so people could check out this epic project.
as for overrunning....
2999 triangles it was fighting for details till the end. Next is UV layouting and AO/Texture stuff. I might fix some minor clipping issues -but overall I am happy with this one.
That is very cool. Nice details. I like it. Now everyones gonna fight over batsockmonkeyman
---
badd time to post as everyone is drooling right now, doesn't do mine much justice
Anyway, what I worked on today. Gotta have more than one truck in game so hwy not a classic step-side
1200 polys now without details.
very cool baddcog, I love those style of trucks. I think you might have some problems with the front wheel clipping through the wheel well when it angles to turn.
Yeah, I was gonna test it. Right now I think that front wheel might be turned to max, not sure.
It was really hard to squeeze the driver in between those wheel wells. I wanted the back wheel well behind the cab but that couldn't be done.
I did a test on the classic truck yesterday. I just applied 2 materials in max, the yellow and and transparent white for glass, skinned to one bone. Made a blank 15 frame anim (0-5 idle, 6-10 turn_left, 11-15 turn_right) and exported.
Added the enviro and rimlight effects to those materials and it looked fine in game. NO UV's!, the basic colors don't need em and the enviro and rim work off of verts.
So lets see 'em in game, no need to wait for uv map to be finished. It only takes about 5 minutes once you got it down. (Plus it's good to test size early)
makes me glad that so many people like it,
update UV layouted + AO passes rendered + started on the dirt texture
the car is a little bit light right now so that I can better see the AO and dirt for now
I will be interested to see if you have any difficulty fitting into the size limit I came up with becuase the mesh file will probably be pretty large. You might have to make sure your colormask files are greyscale and you might need to save your lum texture as a .jpg
looks AWESOME! looking forward to seeing it in game!
Here's the second custom AI in ZG
I just recycled him from another project I did about a year ago. He never got finished tex wise so I gotta do that.
Unfortunately I had already done all his anims and he was set up nice. But I had cloned/mirrored the bones for his right leg and arm. That's a nono. During anims in ZG the right side would bend from the left coords. Hopefully I can recycle those anims, but after making new bones and repositioning everything I'm not so sure.
This was his celebration anim.
This is where I stop tonight. He is in game, the anim (left_turn test) works right, his position is correct (although hat bone needs to go foward a hair).
He's a little picky on hats and accessories. A gargoyle can't be seen in just anything ya know.
The Afro, Samuri hat, Birds, Halo all work pretty good. Hopefully the 3d glasses will work good. I might have to make a few accessories that just fit him good, I don't know.
I have been spending a lot of time polishing up the tracks in our game in preparation for the race mode that Brian is working on. One of the fun parts is adding particles to the level, here is a shot of the mine from Spaghetti West with lots of glowy particles.
Brian wrote a new blog post today titled "The Secret of Zero Gear". An excerpt -
I'll tell you a secret, Zero Gear is not a kart racing game."
in non-blog-posted news, I am currently polishing our 3 race tracks - Spaghetti West, Champion Circuit, and Chill Mountain. And hope to have some fly/drive through video of each of them soon.
Brian is toiling away on the race mode game rules and system, as well as tweaking some of the vehicle physics setup as we come to better grips with the different types of terrain people will run into.
That sounds awesome, Marsh! The possibilities are endless. And with all the people obviously interested in this game, you should have plenty of people coming up with games. I mean, look, your pimp thread is more than half of the Low Poly art thead in posts. That's got to be a first.
Replies
First, how would tyou aquire it in the first place, bet the maker of the content? Like TF2 you'd have to download it to see it, so you'd already have it on your comp.
If you bet the maker and won it they would still have it, not like you'd have the only one, but you'd have to earn it. What if you couldn't find anyone to bet the item you want.
This also brings to mind achiement servers
----
Second, at the rate it's going there will probably be so many custom items once the game is released that if all races are 8 people (I think there are already 8 stock cars- I have 2 working already, there are probably 10 more in the thread) there isn't alot of chance you'll see the same car on track unless there are just favorites [see earlier post about who gets to play sock monkey
On top of that each car has 1-6 colors to customize, so even if 8 people chose the same car chances are pretty good none will look alike.
Then add character to equation (about 6 right now)
Accessories (With the stuff that's finished now it's around 25 I think)
Hats (12+)
So I think the point of wanting to have something unique is covered.
I do see marsh's concern about porno be legit. This happens in TF2 every once in awhile. I think for the most part it would be up to servers to lock out content, kick players.
Would be nice to have some kind of official rating 'tag' that items could get somehow.
Maybe there could be a setting in game that defaults to a random generated character if another player has an item you don't have. No forced DL's. You go to sites that host the files, look at pics, vids, see ratings and download based on that. The Steam User group could be a great place to discuss this stuff maybe.
Then if you don't want the porno car you wouldn't be forced to download it and only the people playing it would see it. Players without it would just see a random 'stock' car.
Being a stock car would eliminate any problems with obscene content, it would also help avoid clipping issues with wacky content.
Randomly generated would help keep it fresh even if you personally never DL'ed anything. (It would suck if all players defaulted to blue race car/blue helmet, hard to tell apart)
Personally, I like the 2nd method more - I think it is more elegant and would streamline the use of items more.
But the biggest factor with it, is that it would be a lot more work to build and to support. It would also be a pretty big choke point for service for the game, I would hate for something to happen where either the database goes down, or we can no longer support it, and it pretty much kills the game. We value self-sufficiency a great deal. The most realistic option, at least in the short term seems to be to proceed with method 2, and try and fit as many of the features of method 1 in as we can, especially over time.
About the betting thing, that sounds like inventing a whole different game inside the game - we have thought about doing something simple to limit which character you can grab from, like you can only grab items from people you finished above - or only the winning team gets to grab from the losing team or something like that. Stuff like that will be something we just try and see if it feels ok.
Here is a small project I started on today, I don't know if I will be able to finish, but I'll try
nice Elise
http://myzerogear.com/wiki/index.php5?title=Zero_Gear_Model_SDK
from lowest to highest:
wheel: 1,000 tris
hats: 2,000 tris
accessories: 2,000 tris
kart: 3,000 tris
character: 4,000 tris
with this allocation, one player's total maximum tri count if every item they were wearing had the max number of triangles - would be 15,000 tris. (4x wheels) I would imagine a maximum number of characters you could conceivably see onscreen at once would be around 16.
I do not have a carnal knowledge of pc hardware, but it seems that these amounts should be a decent balance of giving modelers enough room to play, while not having to be a triangle crunching master modeler in order to make something, while still being low enough for good performance on hardware from the last 4 years.
I am also hoping to have some auto generated lod models which would significantly reduce triangle count on karts not in your immediate vicinity. Also I think it will help that we aren't doing anything yet outside of fixed function effects, so no crazy shaders that take lots of passes.
What do you guys think about those numbers? I would like to hear your opinions more from a technical performance perspective, as most of us here would probably agree they are generous enough from an artist perspective.
That was a fun little project. On a side note, I have everything set up in metres, and imported the right files, but the export was too big. I had to play around to get it the right size.
Hope I didnt break too many copyrights
hmm, are there any settings in the ogre exporter you are using where you can specify what unit size to use?
Just went through and tested the size of all my items zipped with normal compression, and these are the size limits I am thinking about using:
Characters: 512k
Karts: 512k
Hats: 128k
Accessories: 128k
Wheels: 128k
Some of my items didn't fit those numbers when zipped up, but for those that were over I was able to bring them under the limit by making sure my colormasks were saved as grayscale as well as converting some of the lum .png files to .jpgs
The poly count is awesome, I was hoping for that much, more is overkill I think. The textures are small enough that it's hard to pack tons of details into them.
I have even used a 8x8 pixel solid color for 2 colors on one object. I just gave each object it's own material and used that same 8x8 for the colormask.
Smirnoffka,
That's awesome. I uploaded some correct size 3ds files to the Zero Gear forum. I'm gonna put them on the Wiki here shortly, will probably resize all the bounding box.obj's first.
Seems to me that the .obj files imported into Max are about 2x the size of correctly exported Max objects.
Edit:
Just saw your post baddcogg, and it seems that that was the same problem I was having. Must have been a max issue. No worries anyway, I have it figured out now and can make a shitload of cool hats.
Anyway, files, do with them what you will.
http://myfreefilehosting.com/f/ec0e1f8603_0.23MB
Also cog, it is much cheaper for the engine to have models that use only one texture. For every separate texture that is used on a model the engine separates it out as a separate model group and has to render that batch of triangles separately.
*EDIT* was searching the Ogre3d forums for issues with Max exporters and scale problems, this sounds exactly like what is happening, right?
http://www.ogre3d.org/phpBB2addons/viewtopic.php?p=45692&sid=ce193664e1d3d7c4af509e6f4e2718c6
*double edit*
lol, just noticed that badcog himself posted that
AWESOME helmet, i oly have two TINY TINY TINY little nitpicks with it...
1. the nub on top is the middel can't be colored (when making a tie pilot the helmet is black.)
2. there is a model seam down the back that could be welded shut .
even with those i'm still gonna use it :V
also..
COOOBRRAAAAA
So you mean each material in the mtr file is counted as a seperate tex.
That's a bummer. I can try to minimize that alot but somethings have to be split (on my T-Bucket) the double sided springs have color and transparency so that needs to be one, then there's the paint, then there's the chrome alpha which is invisible if used with the same shader as the color/alpha, then I wanted no-enviro for a few things. I could live without the last but that's still 3.
I only have one colormask though, so that's fewer resources that need loaded then if I had one shader and 3 colormasks. I wonder if that helps at all, probably at game start but not during game?
My fro uses 2, one for the pick and one for hair (I don't want shiny hair). But only uses a tiny 8x8 pix colormask for both color changes.
Yeah, and what I got there was that if it is a non-animated (non-skinned) object X-Form should work to reset scale, but I couldn't get it to.
I had to make a box, attch the mesh to it, delete the box and scale was reset. They said you have to do it this way for a skinned object, but I think that would mess up the vert wieghts.
So like I suggested on the Wiki: model, scale, attach/delete box, skin to one bone, import test.
Once scale is right then rig the whole thing and animate.
I made the mistake of skinning to several bones and animating, then trying to fix scale and had to redo it all.
requested
maybe then if we could get copies of all those .obj files as max files at the right scale it would save people all that bother?
About the LOD system : do you already have it working? The reason I ask is because I've yet to see a really good auto-lod generator that doesn't require a mesh to be made in a really specific way. I'd suggest that you allow artist-created LODs as well. The auto-lod could be made to kick in if it doesn't find a hand-made LOD, getting the best of both worlds maybe?
For example, say you want four levels of detail altogether. The artist could make all of them, just the highest level or the highest and second level. Whichever levels are missing are filled in by the auto LOD. This would give artist control over the most important LODs (the ones nearest the camera).
I've also figured out the pivot issues. Just need to reset scale and pivot in the hiearchy tab before rigging/export.
I have filled out a Max/O Fusion export guide on the zero gear Wiki
http://myzerogear.com/wiki/index.php5?title=Model_Export_Guide
No, we don't have an lod system working yet. I would have to figure out how the automatic system works before I made any decisions about that. The lod range would probably be pretty far away. If you didn't force people to make the lods, I doubt many people would make them besides the professional type guys on this forum.
thanks baddcog!
Put up a post on the 'ol blog - http://nimblebit.blogspot.com/2008/08/zero-gear-marks-1-year-of-development.html - has some shots of everyone's stuff in here. I just grabbed some things really quickly, sorry if I missed your stuff (though I did skip over a few for possible copyright reasons, don't want to use those to "advertise" the game) I also linked people to the thread so people could check out this epic project.
if i get any ideas tonight you can bet they'll be ingame.. i HAVE an idea.. just gotta figure out how i wanna do it.
2999 triangles
no words . . .
coolest kart . . .
ever.
---
badd time to post as everyone is drooling right now, doesn't do mine much justice
Anyway, what I worked on today. Gotta have more than one truck in game so hwy not a classic step-side
1200 polys now without details.
It was really hard to squeeze the driver in between those wheel wells. I wanted the back wheel well behind the cab but that couldn't be done.
I was gonna do those
Nice work though
Good job on the X-Ray Glasses.
Now that i'm back I still have to do some adjusting. I saw the post of the finalized tricounts, so i'll work in more detail.
I did a test on the classic truck yesterday. I just applied 2 materials in max, the yellow and and transparent white for glass, skinned to one bone. Made a blank 15 frame anim (0-5 idle, 6-10 turn_left, 11-15 turn_right) and exported.
Added the enviro and rimlight effects to those materials and it looked fine in game. NO UV's!, the basic colors don't need em and the enviro and rim work off of verts.
So lets see 'em in game, no need to wait for uv map to be finished. It only takes about 5 minutes once you got it down. (Plus it's good to test size early)
Congrats on the interview, though short
update UV layouted + AO passes rendered + started on the dirt texture
the car is a little bit light right now so that I can better see the AO and dirt for now
looks AWESOME! looking forward to seeing it in game!
I just recycled him from another project I did about a year ago. He never got finished tex wise so I gotta do that.
Unfortunately I had already done all his anims and he was set up nice. But I had cloned/mirrored the bones for his right leg and arm. That's a nono. During anims in ZG the right side would bend from the left coords. Hopefully I can recycle those anims, but after making new bones and repositioning everything I'm not so sure.
This was his celebration anim.
This is where I stop tonight. He is in game, the anim (left_turn test) works right, his position is correct (although hat bone needs to go foward a hair).
He's a little picky on hats and accessories. A gargoyle can't be seen in just anything ya know.
The Afro, Samuri hat, Birds, Halo all work pretty good. Hopefully the 3d glasses will work good. I might have to make a few accessories that just fit him good, I don't know.
in non-blog-posted news, I am currently polishing our 3 race tracks - Spaghetti West, Champion Circuit, and Chill Mountain. And hope to have some fly/drive through video of each of them soon.
Brian is toiling away on the race mode game rules and system, as well as tweaking some of the vehicle physics setup as we come to better grips with the different types of terrain people will run into.