Arma 3 Topology technique?

[Cunico 995]

I was just curious. I normally make a high poly model and bake onto a low poly model by deleting a ton of edges and vertices etc. Anyways, I was just wondering for arma 3 is it better to make a highpoly model and than retopologize over top of it? I just thought of this as I used to to this but its been awhile and wondering whats more effective I guess. I'm sure its just personal preference but figured I would ask. thanks.

[kiory 404]

Yep, it's personal preference, personally I make a high poly and then build over the top of it, I like having one continuous mesh for the most part, unless it has things like bolts or buttons poking out, moving parts etc, I would model those separately.

[Cunico 995]

Yep, it's personal preference, personally I make a high poly and then build over the top of it, I like having one continuous mesh for the most part, unless it has things like bolts or buttons poking out, moving parts etc, I would model those separately.

thanks for the fast response. Yeah I kinda figured it was personal preference. I guess I will just stick with deleting edges and such. I can see how making the low poly over top the high would be awesome in some cases tho. I guess it depends on the model to. I feel like deleting edges and stuff is a good way tho just because your model is already made and you have to go in and delete topology that isn't needed on your mesh. The pros always say "If it's not adding to your model, than its taking away".

[kiory 404]

Well, here's some advantages to consider for modelling over a high poly, using a continuous mesh:

1. You'll get better continuous UV islands that are easier to control and work with when doing texture work. Although bare in mind, 90 degree angles on your UV islands need to be split off, or you'll get vertex gradients which will cause shading errors. I use a UV to smoothing group work flow for this, basically make the UV's then apply a smoothing script over it, saves time and means I don't have to worry about smoothing groups. One work around for splitting off uv's would be to add some chamfering on the edges, 45 degree angles seem to be fine for Arma, but you can't really get away with doing this for everything as you'll most certainly ramp up the vertex count, so you'll have to split off UV's at some point.

2. The less model sections you have, the easier the bake is to set up, honestly my first bakes are generally over with the first or second time, with minimal clean up needed, if at all.

3. Less sections also mean less aliasing on the model (but you must also consider your edges on the high poly, tight edges also mean aliasing, looser edges not so much), as I understand it, where 2 edges meet that are composed of separate pieces basically means more aliasing, which is ugly.

An example of number 3 would be the MX rifle, I'm pretty sure they went with a HP/LP workflow as the normal maps appear baked (although I feel that the bake isn't very effective in light), but there's a section at the back near the buttstock, above the trigger and back a litte, it's composed of 2 low poly sections which causes all manner of aliasing etc, they could of got away with just a normal bake imo.

Now some cons:

1. Can cause some extreme ray calculation errors if you don't know what you're doing, warping and wiggly bakes and such, this method generally requires more supporting geometry but personally I find it fairly inexpensive for Arma 3, as it shits out vertices/triangles etc.

Well, that's it basically, some people would probably find the cons a deal breaker, but really speaking, the engine deals with vert counts very well, plus with lods and such, really isn't a problem.

Edited July 19, 2014 by Kiory

[Cunico 995]

Hey Kiory, thanks for the info. its much appreciated.