Just a quick question: I cannot decide whether I should use simple box or plane with transparent texture or modelled structure as railing (see image). View attachment 10681 Textured plane is definitely easier to handle and would make a lot less polygons but then I need to use transparency and textured shadows. It would also look very flat/thin. If I model the whole thing it would make a lot polygons but then I could use solid shadow caster and it would also look much more natural. On my 4 km track I would need these in every 200 meters, approx. What do you think? Is the higher polygon count a bigger problem than textured shadows and bad looks?
personally for looks, I'd go structured model. A quick look at the model I would go 2 x long rails (top and bottom) with the other rails (vertical) in the middle and on the ends. I wouldn't go as far as welding the pieces together, just have the pieces inside each other which in-turn will reduce the poly count. Get my drift ?
I think you're fine going with poly objects. Polycount is much less important these days as it was back in rF1 days (in terms of what hardware is capable of at least), and these look like they aren't going to be very high poly, anyway. The total polycount of the scene doesn't matter that much, either, as only what's in view will actually be drawn. To add to what Radar suggested, don't forget to delete any end-cap polygons that are completely inside of other polys (i.e. the top and bottom polys on the rungs).
Ah yes I should have mentioned this, open ended rails inside the top and bottom (vertical). Cheers for that blakboks and very good point.
I agree that rF2 can handle much, much bigger amount of polys than rF1, but we still have to remember that we have to save some polys for cars also. Car models are nowdays way more complex than rF1 days. In this case (imo) it depends on viewing distance. If rails can be seen very close distance, go for structured model. If they are seen far away, I would go for transparency approach. Either way, model itself has to be full box to be seen rF2 lightning engine correctly. Cheers!
Thanks to all of you! I feel much more confident now to stick with the modelled one and it looks so much better, especially as these are always along the race track which unveil the thinny textured one. And yes, all pieces are simple boxes, not welded together. Thanks for the tip to remove invisible cap ends!! That will reduce the vertex count by 12 for each vertical rail!!!
My suggestion: - make sure your fences are not made up of countless small objects - merge multiple segments of fence into one GMT file, to make sure graphics card can render all these polygons in one batch. - use modeled fences - provide flat versions for second level of detail (at far distance) - this only makes sense if you followed first advice - provide separate collision model - a simple box for each segment In case of collision model, it may be beneficial to cut your fence into short segments in separate GMT files. I haven't tested it, though.
Polygons. You can always use LOD and a plane outside of 100m. No need to set the polygons as the collision model, use the planes for that. Textures are the killer these days. Size of each, total size, and overall number. To drive the point home, someone made a wonderful demonstration video of a massive rotating bazillion-gon that hung in the air over a moving car in rF1... barely affected framerate.