Questions about game-ready asset modeling workflows

Rokukkkk

Hi everyone,

I've been learning modelling for about half a year now, but I am still confused about workflow for game-ready hard surface modelling. I have seen different workflows like:

1) Perfect quad (low-poly) -> SubD (high-poly) -> bake (game-ready)

2) Boolean, quad, etc. (block-out) -> fixing & SubD / dynamesh / sculpting (high-poly)    -> topology fixing (low-poly) -> bake (game-ready)

3) CAD-like software (block-out) -> export mesh & tweaking (high-poly) -> manual reducing the polycount (low-poly) -> bake (game-ready)

I heard people say the 1st method is the best, but I feel it's too hard to keep the topology in mind when the model gets more details and it costs too much time. The 2nd one, I've seen many people do it that way on YouTube, but IDK if that's ok for something like a real project. I know for concept, unnoticeable shading issues are ok but for high-poly will be used in baking is that the same?

Btw, is there an industry standard now for game-ready modelling?
Thanks for any advice.

Benjammin

Its mostly #2 in my experience, but it depends very much on what your modelling. Something round without a lot of surface detail is gonna be mostly quads because that's often the easiest way to make it. If its something with a lot of booleans.... not so much  

Shading issues are only issues if they show up in engine, so there's no such thing as an 'unnoticeable shading issue' In my experience, you want to be careful to have a nice clean bake that doesn't require cleanup.

The industry standard is that it looks correct in engine, is within your polycount budget, and doesn't take forever to make. GPUs can eat insane amounts of triangles, but content creation is expensive, so its ultimately a balance between optimization and time. 

sacboi

There are a few more...however what you've listed, are still valid so just boils down to a specific use case.

"Btw, is there an industry standard now for game-ready modelling?"

Again, lot of variables packed into that query; Anyway personally for mechanical hard surface, senior people I know of mainly favor a non destructive approach both active boolean/dynamesh or CAD rounded edge shader/material level non 'high flow' modeling paradigm, especially over the last 5 or so years.  

Rokukkkk

Benjammin said:

Its mostly #2 in my experience, but it depends very much on what your modelling. Something round without a lot of surface detail is gonna be mostly quads because that's often the easiest way to make it. If its something with a lot of booleans.... not so much  

Shading issues are only issues if they show up in engine, so there's no such thing as an 'unnoticeable shading issue' In my experience, you want to be careful to have a nice clean bake that doesn't require cleanup.

The industry standard is that it looks correct in engine, is within your polycount budget, and doesn't take forever to make. GPUs can eat insane amounts of triangles, but content creation is expensive, so its ultimately a balance between optimization and time. 

Thank you for the advice!
I saw a lot of tutorial videos, and they said something like try to limit the shading issues in a small area, then that will be unnoticeable.
But most of them are doing concept art or design, so it confused me if it's the same for in-engine assets.

Rokukkkk

sacboi said:

There are a few more...however what you've listed, are still valid so just boils down to a specific use case.

"Btw, is there an industry standard now for game-ready modelling?"

Again, lot of variables packed into that query; Anyway personally for mechanical hard surface, senior people I know of mainly favor a non destructive approach both active boolean/dynamesh or CAD rounded edge shader/material level non 'high flow' modeling paradigm, especially over the last 5 or so years.  

I have heard that non-destructive modelling is nice because sometimes they might require you to change small details if you do modelling for job.
Thanks a lot!

dimwalker

Rokukkkk said:
I saw a lot of tutorial videos, and they said something like try to limit the shading issues in a small area, then that will be unnoticeable.

I think it would work better with flat area instead of small one. Flat surface can handle a lot of weird geometry scenarios without affecting shading.

gnoop

if  hard surface object is mostly rectangular shapes and cylinders  I  don't do hi res models at all . it's a waste of time.  Splitting to parts and re-naming  is a big part of it too.  

   I use either rounding (bevel) shader in offline renderer  to bake hard edge roundings into a normal map   or rather just a substance  that  edge bevels uv islands  and you adjust normal map intensity to seamless look  ( same as ancient way  from early 90th  when we just blur  opacity map on uv  islands in Photoshop  to make rounded edge normal map. )    Still works perfectly ok and quick as two clicks.   

 

if it's complex thing  it's CAD modelling for sure.  Way quicker and less brain stressful  than subD. You don't have to plan ahead so much.  Unwrapping just a packing click. 

pior

 Rokukkkk said:

I heard people say ...

If anything, the one habit to take is to check the crendentials of whoever "says" things. Check if they've worked at studios or as a freelancer on shipped titles that reflect what they are talking about before taking anything as gospel.