Well, it's the same..no matter what platform you're using. A subdivision is a subdivision... You can get subdivision with or without smoothing (nurbs/quads/HN etc..), is pretty the same for every 3D platform. BTW, as already said, before proceeding with other cuts, I think you can work to delete a lot of useless cuts/edges. More useless cuts you've, more inefficient the mesh and more difficult will be a final correction, when the model is full of detail, bodycuts, extrusions etc... Look at the sharkfin; This surface should be vertically flat...and horizontally linear, so you could solve it with few polys... At the contrary with all those polys you've a non-linear surface, with useless curves on the top side..and a lot of useless polygons. There is no need to subdivide a linear and flat surface... I'm just trying to help btw...I don't want to bother you too much, but these are things already said a lot of times... Perfect example...
It's not bothering, don't worry, it's helpful I fixed it a bit : View attachment 2998 so, for example, I'll make an extra edges loop around the front/rear wheel to make a small radius chamfer corner, right?
Here you go, I did this quickly to show you, you see the edges and how the poly's are used to sharpen the corners?
Check the shark fin top side. In the real deal is flat and 100% linear. No changes in the line from the start to the end. Talking about this; Yep. For the game engine we've to use as much as possible smoothing groups instead real cut subdivisions. You can create a smooth angle/corner with just few cuts, letting the smoothing group angle the rest of the job..
Ye, i see it. I thought those extra edges can be avoid using smoothing groups. So it's better to add them and then using 2 non consecutive SM to have an hard corner?
Oh you're referring to this part i thought you were talking about the small sharkfin near the front wheel. That one is flat now and I think it's ok. i'm gonna flat the up one too Thanks for screen eheh
Just one thing Tommy, before you export Your modell the car bottom need to be on the "ground" not the wheels.
Look at the mianiak example. Explain perfectly how to achieve the correct result with a handful of polys...
Excatly what I suggest in my tutorial aswell (for those who missed it, it's here). By the way - It's a shame programs like 3D Studio and Blender don't have any proper algorithms to generate normal vectors. When we design geometry, we do it with some shape in mind and normal vectors should describe that shape aswell. We usually have only two words of describing what we had in mind to our modelling software: "flat" and "smooth". For my own purposes, I have written my own algorightms: "flat" "equal" = unweighted average (so that's ordinary "smooth") "size" = weighted average based on polygon size "field" = weighted average based on polygon field "greater" = normal vectors always follow bigger polygon And of course there's a maximum smoothing angle. Unfortunately these can't be used for rFactor modding in current state. Perhaps some day when I get to create some tools for rF2, but it would be even better to see such options in GMT exporter, woudn't it?
Do you think the mesh is clean enough for now? View attachment 3016 Thanks for all tips above, really helpful
I'm so out of my depth here schematics and blueprints and hightech models make me feel stupid but it's cool that you guys are enjoying it, rather than you being a bunch of unhappy drones at EA or activision slaving away at ****ty games for a ****ty company for ****ty pay, you guys have my respect. it does look like a really cool car, I youtubed the car and holy cow I want it now.
I'm pushing for 120/140K tris for the entire max lod car (everything for exterior/interiors)...When I turn down the high detailed cockpit stuff and brakes/calipers/differential I get a 70k car with max lods. Actually I do not have any problem even putting 10/15 cars without low res lodding on the track...so I think we can push a little bit working more on the lodding side. Tested on both HD6870 and 6970...(considering actual ATIs drivers sucks for rF2 I'm pretty satisfied..) PS: of course we've to think at liveries and other car textures...but I'm confident there is a good range for optimization.
I've made a global triangulate to check total tris, and now I'm at 30k (only exterior + wheels). The mesh look like these now. View attachment 3024
Yeah. 30k just for the bodycar and tires sounds good...but you've to model a lot of stuff for interiors, suspensions, rollbars, diffuser, bottom of the car, wipers, pitot tubes, bolts, etc etc...so for now you're on the right way and I think you'll get more than 60K for a complex car like that Audi...
Just a little doubt; are you sure 30K for the entire simmetry? Seems you've very high detailed rims and in 3ds max you have to fight a little bit with modifiers to get the correct viewport simmetry polycount... BTW I don't know how lightwave works with simmetry polycount..
mmh i think it's right because there isn't modifiers and it's simple geometry mirror so it's pretty accurate