Low Poly Assets Research

Cdawilliams · Nov 2011

Hi there, I'm relatively new to PolyCount so wasn't sure if this was the right place for my question - if not then feel free to move it.

I'm a student currently writing a research document on 'How would one make low poly game assets effective within a contemporary game engine in today’s increasing demanding market?'

Could anyone briefly (or more in depth if you like) outline their workflow & approach, from modelling to importing into the games engine, that you would take to do such a thing - Or even just point me in the direction of some useful reading resources.

I've already read a load of tutorials regarding baking normals, collisions, difuse & specular maps and different low poly techniques I just wanted some first hand advice on it all.

Any help would be greatly appreciated.[FONT=&quot]

[/FONT]

Cdawilliams · Nov 2011

Realised that I was quite vague. I'm not really looking into how its done as I have a fairly good working knowledge of 3DS Max - I'm more interested in the order professionals would go about doing it and why.

CrackRockSteady · Nov 2011

I'll just give you a very brief synopsis of my process:

- model unoptimized "mid poly" asset.
- make 2 copies. One copy gets optimized (collapse edges, weld verts, etc) and UVed, the other becomes my high-poly mesh for baking.
- depending on the asset, either create high-poly model using 3ds max, or export the model for detailing in zbrush (or both).
- Bake normal map and AO using 3ds max or Xnormal
- Use baked normal map and AO map as a base for texturing in photoshop.
- create texture using photoshop. Export model and textures, import into UDK or whatever engine I'm using.

Not really sure what aspects of the process you're interested in but if you have any questions about specifics I'll do the best I can to answer.

Cdawilliams · Nov 2011

CrackRockSteady wrote: »

I'll just give you a very brief synopsis of my process:

- model unoptimized "mid poly" asset.
- make 2 copies. One copy gets optimized (collapse edges, weld verts, etc) and UVed, the other becomes my high-poly mesh for baking.
- depending on the asset, either create high-poly model using 3ds max, or export the model for detailing in zbrush (or both).
- Bake normal map and AO using 3ds max or Xnormal
- Use baked normal map and AO map as a base for texturing in photoshop.
- create texture using photoshop. Export model and textures, import into UDK or whatever engine I'm using.

Not really sure what aspects of the process you're interested in but if you have any questions about specifics I'll do the best I can to answer.

Thanks for the input; thats helped clear up the order people like to approach work with but if you have a second could you explain why you would go about making a mid-poly asset first.

I've only read a lot of tutorials but come January I'll be producing 5 game assets and I read some that some like to produce a low poly, UV, duplicate, add detail to the duplicate then start with the normals. But never really explained the importance/benefit of approaching it this way.

As I said, I appreciate the help; the more opinions/reasoning I can get behind the way things are done the better I hope to understand it!

EarthQuake · Nov 2011

CrackRockSteady wrote: »

I'll just give you a very brief synopsis of my process:

- model unoptimized "mid poly" asset.
- make 2 copies. One copy gets optimized (collapse edges, weld verts, etc) and UVed, the other becomes my high-poly mesh for baking.
- depending on the asset, either create high-poly model using 3ds max, or export the model for detailing in zbrush (or both).
- Bake normal map and AO using 3ds max or Xnormal
- Use baked normal map and AO map as a base for texturing in photoshop.
- create texture using photoshop. Export model and textures, import into UDK or whatever engine I'm using.

Not really sure what aspects of the process you're interested in but if you have any questions about specifics I'll do the best I can to answer.

This is generally, with perhaps some small variation, what most people do.

I also assign colors to the various highpoly bits, RGB or RGBCMYW so I can bake out a diffuse texture and create masks from in photoshop. This really speeds up the texturing process.

CrackRockSteady · Nov 2011

+1 to EQ's color assignments for masking in photoshop.

As far as the reason I start with a mid-poly first, it really is just personal preference. I know some people who make their entire High Poly mesh first and then go back and build their low-poly over top of it. A long time ago I used to do as you described, create a low poly mesh, UV, then copy and build more detail on top.

The problem I would run into with that method is that often times after a mesh had been optimized and had edges collapsed and welded, the topology would not be very conducive to subdivision. It is much easier to add in extra edge loops in preparation for something like TurboSmooth if the model hasn't had a bunch of edges collapsed and welded into triangles all over the place.

Hopefully that makes sense? If not I can try to find some images that explain things better.

Cdawilliams · Nov 2011

Yes, it makes sense - thanks. Sadly, at this stage I am just conducting research to write the document then produce the artefacts. If this wasn't the case I'd of happily spent hours trying different techniques, toolsets, and generally playing about till I found a workflow that suited me.

But the feedback is greatly appreciated

cryrid · Nov 2011

but if you have a second could you explain why you would go about making a mid-poly asset first.

From my own experience, it's just a fast way to get the ball rolling. There's no worry about topology, final quality, or optimization; it's just mashing together various shapes while pushing and pulling large clumps of verts to quickly work out the overall form and proportions you're after. Once you have that sorted out, it's easier to retopologize it towards whatever end.

Cdawilliams · Dec 2011

cryrid wrote: »

From my own experience, it's just a fast way to get the ball rolling. There's no worry about topology, final quality, or optimization; it's just mashing together various shapes while pushing and pulling large clumps of verts to quickly work out the overall form and proportions you're after. Once you have that sorted out, it's easier to retopologize it towards whatever end.

Thanks - I'll take this into consideration when I get to the practical side of things.

One final thing: When did use of Normal Mapping become the norm for video games? And what was used prior to it (if any)? Would help with my research - thanks

Cdawilliams · Dec 2011

I've found out the idea for normal mapping onto low poly models was introduced in a paper in 1998 - prior to that was there just no such thing as normal mapping?

Were models just kept simple enough to allow for smooth performance but complex enough for the gamer to actually tell what it was?

Not sure if anyone that uses there performance were involved in the games industry 13 years ago but I couldn't find much more information on it

CrackRockSteady · Dec 2011

AFAIK normal maps did not become 'standard' until the current generation of consoles. Prior to xbox360/PS3 there were some PC games which took advantage of the technology (Doom III and Far Cry come to mind) but I don't believe it was very widely used before that.

Prior to that, (I could be wrong here, someone else can confirm/deny) models were textured with diffuse maps only, and any extra details that were desired were baked into the diffuse map.

Cdawilliams · Dec 2011

CrackRockSteady wrote: »

AFAIK normal maps did not become 'standard' until the current generation of consoles. Prior to xbox360/PS3 there were some PC games which took advantage of the technology (Doom III and Far Cry come to mind) but I don't believe it was very widely used before that.

Prior to that, (I could be wrong here, someone else can confirm/deny) models were textured with diffuse maps only, and any extra details that were desired were baked into the diffuse map.

Wikipedia states:

Basic normal mapping can be implemented in any hardware that supports palettized textures. The first game console to have specialized normal mapping hardware was the Sega Dreamcast. However, Microsoft's Xbox was the first console to widely use the effect in retail games. Out of the sixth generation consoles, only the PlayStation 2's GPU lacks built-in normal mapping support. Games for the Xbox 360 and the PlayStation 3 rely heavily on normal mapping and are beginning to implement parallax mapping. The Nintendo 3DS has been shown to support normal mapping, as demonstrated by Resident Evil Revelations and Metal Gear Solid: Snake Eater.

But nothing regarding how it was done prior to that - but I'll take what you just said and see what reliable google turns up

cryrid · Dec 2011

Were models just kept simple enough to allow for smooth performance but complex enough for the gamer to actually tell what it was?

Just take a look at some N64 / PS1 / PC games from that era. Very polygonal, with most of the detail coming from low res color textures.

http://www.gamegoldies.org/old_game_files/2009/05/unreal-1-screenshot-yard.jpg
http://www.blogcdn.com/news.bigdownload.com/media/2009/04/1999feature99.jpg
http://images.wikia.com/finalfantasy/images/f/fc/FFVIIbattleexample.jpg
http://www.gameguidedog.com/game-walkthrough/wp-content/uploads/2011/04/halflife1.jpg

Low Poly Assets Research

Replies