Jump to content
  • Advertisement

Archived

This topic is now archived and is closed to further replies.

Captain Insanity

Pushing the limits

This topic is 6188 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Okay, I really need help on this one. Some of you may remember from my posts that I''m trying to make a first person adventure in a combination of 2D panorama and real time 3D. The plan was to have a panoramic pre rendered background to the game, and all interactive objects and characters rendered in realtime 3D clipped around an invisible mesh synchronised to line up with the rendered background in a processor saving maneuver like alone in the dark. Here''s the problem - is it possible with current realtime 3D engines (and the hardware most consumers have) to pump out enough polygons to make the objects and characters detailed enough to be indestinguishable (or at least believably so) from the pre rendered background? I mean, with only a few objects to render per scene, can you make the meshes of those objects roughly as complex as those used in pre rendering, ie high poly count? Also, can current 3d hardware and engines support all the features one would need such as specularity, bump mapping, hardness and shine, and most importantly, oversampling and antialiasing? In my experience, all realtime 3d engines out there simply cannot hide those unsightly pixelated edges to polygons, which pre rendering programs mix with the background to smooth them out, and the most heavily bump mapped polygons still look completely flat from most angles. I ask all these basic questions because I am merely the designer, not the programmer, and know little of such magical things. All I really need to know is: do you lot think it''s a feasable plan? Should I change it in any way? What is certain is that I can''t do the whole thing in realtime 3D at the level of detail that I need. There aint enough processing power in any single home user''s machine to do that.

Share this post


Link to post
Share on other sites
Advertisement
you will have to compromise somewhere. if you keep the number of objects down then you can get quite detailed models. a good example of this is Space Channel 5, while the rendered backgrounds (full motion video actually) looked a bit better then the realtime rendered stuff, it still fit smoothly. the trick is to ensure your art design fits what the hardware can handle. it is just not feasible to get the quality of a software rendering app like 3ds max due to all the complex per pixel lighting calculations, shadows, radiosity, recursive reflections, refraction, etc in todays hardware (even on the geforce3/radeon 8500).

oversampling can be done, but there is a fill rate hit. you would be surprised at the quality level that can be acheived on todays hardware if you are careful, and know how to model your objects/worlds. specular lighting is supported on must cards, but unfortunatly the quality is kinda bad unless you have high poly models or use pixel shaders.

as to polygons looking flat at most angles even with bump mapping. well polygons ARE flat. bumpmapping helps aleiviate this by adding "depth cues" to the texture so when lights pass by it shows by affecting the "bumps" on the poly. you could look into the radeon 8500s deformation mapping (i think thats what its called) it allows better mapping of bumpy surfaces. though is ONLY supported on that card currently thus limiting your audience greatly.

also as a side note, what resolution do you run your games at? at 1024x768 and above its not too bad (assuming you are running on a decent moniter). you could though turn on forced full screen anti aliasing if your hardware supports it.


now my take on your idea. first off, making a first person shooter like this is stupid. you limit movement by using pre rendered animations for the backround. unless this is a rail shooter you are handling this wrong. why does everything have to be so detailed? is it because you suck at modeling with low polygon counts? or are you trying to do something that commercial game developers are not doing (and for good reason i might add)?

Share this post


Link to post
Share on other sites
I just KNEW this would happen.

This is great advice and I''m immensely grateful for it, but you''ve misread my post.

I said "First Person ADVENTURE", not "first person shooter".
You know, like RealMYST.

This is more proof for my hypothesis that the whole development world is obsessed with shooters and nothing else! (no offence intended, this is just an observation)

Share this post


Link to post
Share on other sites
Resident evil is some sort of this. 3d and 2d both mixed. Also Alone in the dark 4 work in the same ways.

Pre-render scene is common. But pre-render characters, of course not!.



-----------------------------------------------
<marquee></marquee>
"Cuando se es peon, la unica salida es la revolución"

Share this post


Link to post
Share on other sites

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!