Low Poly to High Poly (or not??) / Workflow Question

YF_Sticks

Hey Polycount!

I am studying modeling and I get a bit confused about the workflow or process of Low Poly to High Poly (or baking).
My goal is to be an environment artist. So let's say I want to create an environment that features a military vehicle and then building the environment around that. 
Would you first do every part of the vehicle in low-poly, then do a high poly version of it and bake it after? Or would you directly do the high poly and do UV's?
Let me know what you think. For me it's a really confusing topic and I'm trying to bring it all together. 

Axi5

I think you've misread something or are confused.

Make your high poly first, if you even need one, maybe try focusing on getting a decent model first.
Then if your model is too high density for realtime, create a lower-poly version of it, while still closely matching the shape of the high.

Bake the high poly normals and ambient occlusion onto the low poly.

I always recommend that beginners try their hand at very simple geometry before they give themselves headaches with large models. It'll give you the practise you need.

YF_Sticks

Axi5 said:

I think you've misread something or are confused.

Make your high poly first, if you even need one, maybe try focusing on getting a decent model first.
Then if your model is too high density for realtime, create a lower-poly version of it, while still closely matching the shape of the high.

Bake the high poly normals and ambient occlusion onto the low poly.

I always recommend that beginners try their hand at very simple geometry before they give themselves headaches with large models. It'll give you the practise you need.

Thank you! Yes, I am confused .

I'm using Maya. So what I'm doing now is: doing the low-poly, duplicate it and then create the high poly using edge loops and bevel (to get the sharpness/smoothness in the edges). 

Yes, I'm trying to do it with simpler objects at the moment. There is so much information out there, you kinda want to understand it all haha.

Joao Sapiro

I would advise to start from the highpoly since when you do the lowpoly you are "locking" the highpoly to its proportions\extrusions, it might seem good but you usually will change stuff on the highpoly that you think looks better and then you have to go back and forth between high\low etc. Wouldnt recomend.

YF_Sticks

Joao Sapiro said:

I would advise to start from the highpoly since when you do the lowpoly you are "locking" the highpoly to its proportions\extrusions, it might seem good but you usually will change stuff on the highpoly that you think looks better and then you have to go back and forth between high\low etc. Wouldnt recomend.

Yeah, I'll try that. Makes sense. Thanks!

Neox

hehe and now comes me telling you that the lowpoly to highpoly is also a very valid approach. +1 to confusion

but no it can actually work, it really depends on the style and the final product. We had productions where it was exactly like this due to budget constraints there was no time to retopo and set up bakes classically. So we cut down the time needed by going the other way around. But arguably the highpolies have been pretty simple.

here are a few example of the approach







it involved little to no sculpting and was basically just lowpolies prepared for doublesmooth a few minor details on top and quickly baked down. Very quick turnaround time.

YF_Sticks

Neox said:

hehe and now comes me telling you that the lowpoly to highpoly is also a very valid approach. +1 to confusion

Really good example, thank you! I mostly do hard surface stuff, maybe there it's different.
I will try both methods. Guess it also depends on what you're working on.

CheeseOnToast

Another thing to consider is that real-world objects are often made of loads of multiple pieces. Generally, you'll want to replicate that in your highpoly model, but the lowpoly should often combine multiple parts into a single shell. There's a few reasons for that including optimisation, ease of weighting/rigging, baking etc. By not sharing the same underlying topology, you're free to build the highpoly as you see fit.

As you get a bit of experience, you'll become better at planning which parts of a model you can reduce / modify  from the highpoly to get your lowpoly and which parts will need to be retopologized manually.