To normal map or not to normal map

Andy Horns

I apologise upfront if this seems like a noobish question, thing is i am a software dev by day and spend my nights with my passion for 3d, i am aware of what normal maps are and how they work, i have always just used them as they help with shading and adding details to low poly mesh. My question therefore is do i use normal maps too often, i see games like team fortress try not to use normal maps unless its for details that cant be easily modelled into a low poly, they end up using smoothing groups. I started thinking maybe i just use normal maps as they are convenient and maybe i could get away with not using them. Is there a point where an asset just doesn't take up enough screen resolution to warrant a normal map or how do you guys as professionals go about determining when to use a normal map? This is also in the context of mainly hard surface game objects not characters etc. Sorry for the long ramble...

Garasin

My understanding of normal maps is you use them to convey detail when it would be to tri intensive to do by actually modelling the details in. This normally comes in the form of small surface details such as bolts, rivets, embossed patters, etc.

It also depends on the look you’re going for in the case of TF2 they use them very sparing because the look they’re going for doesn't always need them. Whereas a game like Left 4 Dead uses them a lot more because the visual style they want requires a lot more surface detail that would probably take them over there polygon budget if they didn't use normal maps.

Andy Horns wrote: »

Is there a point where an asset just doesn't take up enough screen resolution to warrant a normal map

I’m still a student so I could be totally off the mark here! So please forgive me if I am. I imagine that you’d want to always make your assets look as good as possible so if using a normal map will do that and keep you within your memory budget then go right ahead and use one. If it turns out that the normal maps on the smaller assets are hogging too much memory then maybe get rid of them then.

Hope this helps!

Sammy101

Normal maps are not applied sometimes when you are generating a lot of something, such as zombies. Valve did this in that case and had no normals for the zombies, to allow for a lot more to be on screen.

zetheros

Normal maps are (generally) a lot of work, because they introduce a high-poly sculpt, retopology, and tinkering with normal map settings to your work flow. 3rd and first person games often use normal maps because the player camera is really close to the model(s), and isometric and top-down usually don't.

WoW is a good example of a game that doesn't have normal maps, yet still looks great. And any next-gen first person shooter is bound to have normal maps; it all really depends on what you're making, what kind of hardware you're using (or your audience is using), and how much detail you think is required for your project.

Andy Horns

Thanks for the replies ...

Normal maps are (generally) a lot of work, because they introduce a high-poly sculpt, retopology, and tinkering with normal map settings to your work flow.

Indeed, this is exactly what has brought me to this question, I can create a half decent low poly model yet a normal map always helps, but its like you said, depends on camera etc. In my case it will be top down so i think i might just try doing a version with and without. Normal map has just always been part of my workflow and as i learn more i have started asking if its always necessary