View Full Version : First Vehicle Modelling Attempt.
09 September 2006, 02:29 AM
Hi guys. Since i'm fairly new here i'll go into a little background that explains the motivation for this thread. In my other thread (where I was working on an industrial environment) someone suggested that the prefabricated MAX tree i'd placed in the scene was the weakest part of the whole thing. I was inclined to agree with this, but if I removed it i'd have had a big empty space leaving the scene unbalanced. The person who made the suggestion said I should just model a more realistic tree, but I thought that might be quite tricky, so I thought for a little while about what else I could place in that empty space. Then it hit me; why not place a machine there? I naively thought that it would be easier to model a machine (that is comprised of regular components) than a tree >_<.
Several weeks ago I got started on a low-polygon model for a digger/excavator. Although I thought I made decent progress, I just wasn't satisfied with the blocky look. In my other thread, I mentioned that one recruiter said my work was overly simplistic. I guess I took that to heart because I slowly began to refine the model I was making, and eventually it became rather more complex.
To cut my babbling short, I am nearly finished with the modelling side of things. As the thread title says, this is my first attempt at modelling a real vehicle of any sort so it may look a bit crappy, and I didn't have good reference materials at all so roughly 50% of what you see is improvised. It still needs texturing but that will obviously have to follow the modelling stage. Well, thanks for reading this far ^_^.
Here are a couple of in-progress renderings:
09 September 2006, 07:03 AM
no it looks good but it seems a little high poly what is the count? Can we see some wires?
09 September 2006, 03:32 PM
Thanks for the reply ^_^. As I mentioned, the model grew more complex than I had originally planned, so it became a case of going in to very blocky areas and refining them, as I wanted this model to be usable for close-up shots as well as distant ones. Despite that, it runs perfectly smoothly on my aging graphics processor, so I imagine that it would be able to run on a current PC or console too under DirectX. Since it is my first time attempting something like this, my wireframe is pretty ugly i'm afraid, so i'm sorry about that.
If I knew how to work in Sub-D, I probably should have used that technique. That method would generate thousands of useless polygons though (like on a flat surface).
09 September 2006, 03:38 PM
I don't think you've quite grasped the basic concept of game modelling. Just because a model renders fine in a graphics program doesn't mean it will work in a game engine. What you see in your viewport is just unshaded polygons. When it gets into a game, it will have a texture on it, and probably a normal map and a specular map, plus real time lighting, and that's in addition to all the other stuff the game is rendering and all the calculations the CPU is doing. There is no way any game engine is ever going to be able to handle that amount of polygons in real time.
Your best bet is using this model for a normal map, and making a low polygon model with the same dimensions that would actually go in the game.
09 September 2006, 04:49 PM
yes, i have to agree with TychoCelchuuu...
it's a very huge pro, so you could spend around 6k-8k trigons, but your model got much more!
use it as said, as a normalmapping source and build up another model, with the polycount i mentioned above.
but bsides that, it's a really nice model, very detailed, you can be proud of it!
09 September 2006, 06:04 PM
With all due respect, I am fully aware that this model has a much higher poly count than what one would find in a game, and as I have repeatedly said, I allowed the model to become far more complex than I would for actual deployment in an engine. I grasp the concept just fine, and have been working on low-polygon models prior to creating this one, so there's no need to patronise.
Incidentally, any 3D surface in order to be visible has to be shaded (in a viewport or otherwise) so I assume you meant diffuse, specular, bump etc. I quickly added a couple of shadow-mapped spots and a rough texture to the model just to see if it slowed down the framerate much and it didn't appear to.
I'll see if I can get someone to move this thread to a different sub-forum, since it doesn't look like the gaming folk think it belongs here >_<.
09 September 2006, 06:49 PM
Dude chill, you didn't specify what kind of game to begin with, and maybe people just thought you'd mean a basic shooter that's a bit old or something.
Don't turn this into the wrong/right kinda thread, no one was patronising you just yet, there are tons and tons of people who "grasp the concept" The only reall crit I can give based on the information you provided is this:
"It looks way more high poly than the wireframe makes it, or than it actually is, which I believe is a good thing. "
It's not to high to run in a game, whether or not that depends on the kind of game may or may not be another question. If you want clearer C&C from people you should give them preciser information to base their criticism on.
09 September 2006, 11:04 PM
As a high poly model, it looks pretty good. As a real-time model it's absolutely horrible.
For example, why is this cylinder segmented like so (red lines)? They do nothing for the silhouette or shading - they are in fact useless and wasteful.
Also, the grill on the back and sides should be normal mapped, not modelled. It's a huge waste of something that doesn't affect the profile and is too small to be noticeably a texture. And those beveled edges? Just use a specular map if you want the edges to shine a little, don't spend all those polys.
This model is way heavier than anything you will see in any game on a current console, especially for an incidental prop. I wouldn't spend more than 5000 polys on it.
09 September 2006, 11:22 PM
I guess I could edit those out. I cheated a bit and subdivided that component, to make the perpendicular tube connecting to it come out smoothly.
And, as I keep on saying, I don't expect this to realistically be in a realtime environment. You guys are just ignoring that to be antagonistic it seems, so i'll request that this thread be moved. I wouldn't have gone to so much effort to work detail into this thing if it were to end up in something like UE3.
09 September 2006, 11:34 PM
I have to say it looks great, I am sure you will normal map this with a much lower poly version.
09 September 2006, 11:34 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.