Hello! I am new on this forum so please let my introduce myself really briefly.I am doing 3d graphic alongside with some small projects in UE4.I am doing this already for few years but it was always just for fun as my hobby so i can say that i am real begginer/amateur.In the past weeks i decided to start with 3d graphic little bit more serious and build skill which can help me in future maybe even find a job in this industry.Of course i know that ahead of me is long long way but i decided to start taking it more serious. I always liked creating enviroments and i would like to go this way.Right now i am trying to improve in props creation for video games and thats why i am here. I already read hundereds of articles and saw hundreds of tutorials but i thing that constructive feedback would be much more helpful for me. I just wanted to ask for your opinions about my workflow and if the way i am doing this is a good practise.
I will use my latest wrench model as an example and i will try to describe my whole workflow. I am using 3ds max, RizomUV, Substance painter.
1.I created High Poly model focusing on quad geormetry so i will avoid problems with turbosmooth and lighting artifacts during baking.(I am most comfortable with turbosmoothing in combination with smoothing groups and not crease or supporting lines to keep topology clean when i will be making low poly).
2.I made low poly from clone of my HP. deleting unnecessary edges, trying to preserve shape and siluethe as much as possible for later baking proces.In low poly i am already trying reduce geometry as much as possible therefore i am using lot of triangles.
3.Unwrapped model in RizomUV.
4.In Substance painter i baked details from HP to LP and textured the model
I ended up with 2272 polygons low poly model with one 4K texture set.HP model is 36422 polygons.
I would love to ask you few questions please.
1. Is it okay using turbosmooth with smoothing groups instead of support lines ?
2. Is it okay using triangles on my LP model ? (i am trying avoid n-gons completely on both LP and HP models)
3. Is it okay to make model from smaller parts and attach them in the end of process(not boolean just simple attach) instead of making one whole complicated model ? (i am baking parts separately)
4. Is this workflow suitable for creating Low poly assets and level props for games and enviroment design ?
Please keep in mind that i am aware of the fact that this is really amateur work but i am lost in insane amount of ways and 3d graphic workflows for different projects and i would just like to hear if i can stick to this way of crerating assets and improving in this workflow or should i change it.
Thank you very very much for any response and please provide me as much critical feedback as possible i think i really need it to move from place where i am little bit stuck now. I am attaching Low poly, high poly and final renders.
P.S. Quality of renders is not good i am aware of that and it is also field where i should improve a lot.
Replies
Anything goes with the high poly. As long as it bakes well can do anything you want with the HP model.
Yes, it'll need to be all triangles before you bake the normals and textures anyway. If the model does need to be iterated on later, you do want to be able to work with edge loops to make modeling changes easier, but you can be liberal with tris where you need them since it'll end up that way anyway. Also no need to avoid n-gons on the HP, they can subdivide well and make simpler and cleaner models.
Floating separate pieces of geo is fine and normal
It's alright, working from a high poly to create a low poly often creates uneven use of geometry, there's places on your low poly that could get away with less geo, and there's places that ideally need more.
Also when you export to fbx or whatever make sure you check the triangulate box (which turns polys into triangles) because if you let later programs like substance or a game engine do the triangulation it might differ and you'll get surface artifacts like the one in the picture
thanks for the wiki site. If i understand this correctly does this mean that i should bevel my edges even if real life object or reference is completely sharp ?
I will definitely do this with triangulation next time i am just assuring if i understood right the thing with smoothing groups on bigger surfaces. Should i use only one smoothing group for bigger surface unless there are no comples edges or something like that right ? This make sense for me.
Fabi_G said: so if i understand this correctly i should bevel edges on my HP at least little bit even if they are completely sharp in real life or on reference ?
Yes i understand that triangulation... i will definitely try it next time but about that smoothing groups ...does it mean that i should use only one smoothing group for bigger surfaces right ? that makes sense for my ....of course if there is not complexity which needs to be smooth grouped. Do i understand it right ?
For ex. If you have a large poly next to a small one and they are not coplanar (on the same plane) by default you will have an average smoothing or normals on those polys. That will create a large gradient in the normalmap and when compressed or just in 8bit will produce artifacts (looks bad). If you add a loop on the bigger poly or use face weighted normals, the big poly should be flat and the smaller poly should take all the change in normal direction and should produce better normalmap when baked. Really hard to explain in words just test it for yourself in both ways and youll see the difference...