hellow, im trying to build some car LoD for my racing game on mobile, before I start I’d like to know if my workflow is fine, or is there better ways to approach it. anyone experienced with modeling
here is the LoD I need,
Lod0, high-poly, for CG
Lod1, 5-8k, for In Game Close Up
Lod2, 3-5k, for distance
and this is my assumped modeling workflow:
1, build something around 5-8k
2, Unwrap UVs
3, add some structural lines around the hard corner, and generate the high-poly using Turbosmooth
4, bake the normal.
5, now the Lod1 is done, Unwrap the UV for LoD0
6, remove lines from LoD1, and make it to LoD2.
Hi.
As a professional Modeller for 7 years I’d say this workflow is mostly correct.
My usual method, switches a little per project but its really down to preference whether or not you start with High or Low Poly.
1: Model my low poly in maya (can take varying amounts of time according to project requirements and detail).
This polycount can be under 2k for a mobile asset or as high as 25k for a Next Gen character it all depends here.
2: Unwrap my UV’s into 1 or multiple UV sets depending on the object requirements.
3: Make a High Poly from my Low Poly (usually use Zbrush for this but sometimes I use Maya)
4: Export the High Poly and low poly models from Zbrush or Maya (or whatever external tools).
5: Use XNormal to bake my normals (and also possibly my texture painted in Zbrush)
6: Apply the Normals to the object and see how it works.
Thats my usual work flow.
You’ve mentioned making a separate UV map for the LOD versions, but this isn’t necessary, simply take your low poly, and lower the polycount on this by removing the edges and welding the vertices manually.
The best way is manually as you can keep a solid eye on the quality, it doesn’t matter much if certain things become a little distorted or stretched as they will be far enough away from the camera for it to not be noticeable.
3-5k for distance isn’t really required honestly, unless its an animated object, if your object is only seen from a certain angle you can remove faces from areas that’re unseen, roughly 3k is good 5k seems a little extreme.
really realy thanks for this detailed post,
No problem, I hope that was a little useful to you.
I didn’t really mention it here but, sometimes artists start with the high poly and then create the low poly from this, usually with retopologising within Zbrush (or maya), though sometimes people export from Zbrush or Mudbox to Maya and then create a low poly mesh around their high poly, in this situation though people either export a lower subdivision or they have a lot of RAM to be able withstand a model with 10+ million polygons in maya on screen at once.
Good luck with your modelling, if you have a portfolio I’d love to see it
If you want to retopologise in maya/3ds/blender you can use the plugin “decimation master” to export a “lighter” version of your model (like 50k polys).
If you don’t paint the colors in zbrush you can use the same plug to decimate your high poly (200k-300k) for xnormal so the maps are faster to bake.
I would recommend use of this tool, it takes your models topology into account too, though I would recommend against using this methof if you wanted to export something and the animate it however.
yea really, for various reason I guess it’s better to start with lower poly for some mechanical stuff like cars. i dont really need that much details and the most important thing is i have to keep the wire nicely formed, otherwise the highlight would be very ugly without bumpmap(dark shadow in triangles)
retopology is cool for building natural things as human or animals.
Where you start is really all down to taste, some people like to get stuck in with the details whilst others like to establish the form first, whilst others just find it more natural to sculpt shapes in zbrush sometimes.
I usually start low as I like to get the overall feel for a model before I start adding endless details, plus with something non organic I usually just stick with Maya for the high poly too, as zbrush isn’t so great for hard surface sculpting (despite its new additions I don’t think much has changed).
I start with Low Poly for everything, characters, cars etc, the only thing I don’t start low poly with is something like terrain, as I like to be able to sculpt that a little more naturally without the jagged polygons.