Anyone have any experience creating virtual environments using the UnrealEngine?

[Pixelhugger]

Been thinkin about getting this book and schooling myself on the creation of virtual environments. Anyone else do this? The ease of creating an interactive world by adding 3D arcade environments in )P('s 3DFE has me interested in taking it a step further to add more detail and speed. Am I talking crazy talk? My programming experience includes and is pretty much limited to 10 line programs on the PET 4K. Heh.

[Howard_Casto]

)p( and myself have already experimented with trying the arcade models on other engines (including unreal). The performance was horrible. 

Most of 3darcades models are extremely high quality, generally they have 5 times the polys of your average quake 3 player model.  So with that said, I'm not sure what you mean by higher quality models.  The quality of the model is only limited by the author of the model in 3darcade.  We have access to lightmaps, reflection maps, animations ect... about the only thing it can't do is bumpmapping, which is useless when modeling smooth arcade cabs. 

Plus to be blunt, unreal is not a front-end.  It'd be difficult getting it to do things like launching external apps, make gamelists, ect. 

[Pixelhugger]

Well, I'm not actually interested in using it as an FE. 3DFE already does so much more than I could ever hope for in that regard.

It was my understanding (and since I know virtually nothing about the UnrealEngine Runtime or Director for that matter.... possibly a misunderstaning) that the higher end game development tools allow for much greater speed than Director. I guess speed is a more accurate way of expressing what I'm after than detail, though that is the net result. I'm interested in something that can handle huge amounts of detail without the frame rate dropping to a stutter. I believe textures can be much larger in Unreal as well. Not sure.

As I said, I'm very new to all of this FPS modding and understand very little about the differences between Director based environments and those of other engines. So, I'm all ears.

[Howard_Casto]

Unreal only appears to be able to handle those sorts of things better because the models used are of a lower detail and things like shaders (mult-pass textureing) are used to "fake it".  Director does a  rather fantastic job of balancing speed and quality while gaming engines are bult for speed and the game developer uses a great deal of tweaks and tricks to keep things looking nice while still maintaining a good framerate. 

It would be possible, but all of the models would have to be re-done from scratch optimized for a game engine.  In other words it would be a lot of work. 

[Pixelhugger]

So do you think that with a good enough texture artist and efficient enough modeling (plus that lightmapper burner thing) that 3DFE could handle something like this at 640x480?



If so that's great news for me!! If I'm understanding you correctly 3DFE should be able to handle a scene of 100K + polygons. No? And the additional detail could be handled in the textures. In this image I'm guessing seams in the rocks are likely 2D photoshop trickery for the most part? And the highlights on the rocks appear more detailed due to lighting tricks like what can be done with the lightmapper?

*Crosses fingers*

[Howard_Casto]

Yeah that is a fairly flat enviornment.  It's mostly lighting and texturing that is making it look less-flat. 

I don't see why 3d arcade couldn't handle something similar, the only limitation being the lightmapper only having a few dozen spots available. 

[Lilwolf]

I also did some work with using those same models in opengl.  Worked GREAT on my develpoment system... then found that it worked like crap on the ATI card I have in my arcade machine... that sucked...

but anyway...

Many of the models aren't optimized.  They use some basic shapes and not just triangle strips (or whatever was the optimized ones... haven't looked for a while). 

Also, many are stored in somewhat protected formats.  So you have to use preexisting tools / api's to get and display the data... which leaves optimizing out of the equation...

What I was doing, I would replace the screen in a model, then replace it with a rotating cube that would rotate to show the next/previous game screenshot (and marquee)... then random would change the cube into a slot machine and it would spin until it would stop on a game... lots of fun little things.  But then I realized I would have to modify each model by hand to get rid of the screen... and it not working on MY arcade... I stopped playing with it... (plus I think I had a lighting issue that cropped in somewhere right towards the end)... anyway lost enthusiam and more likely lost all available time.  but I code for my own cabinet....

but go for it... if you have an idea it would be fun... but I wouldn't start off using someone elses full blown engine... I would start off with a language that you know, and play with a good API.

[Howard_Casto]

Yeah, just because I said the unreal engine couldn't do a better job doesn't mean that a pure language with a custom engine can't.  Dk uses dx8 and I am spear-heading the process of getting the proprietary w3d models converted to  a open source format.  I've had great success with speed, although admittedly I'm not interested in a full 3d world, just 3d objects.