What is the purpose of a Normal Map and a Bump Map on the same character?

Jonstark

I'm just curious, I thought both Normal and Bump channels were for the same purpose, to fake bumpiness/displacement on the texture.  So why do so many character sets have both a bump map and a normal map?  Does the normal map serve some other purpose?  I would have thought it would be one or the other, or is there some advantage to using both at the same time?  This is probably a dumb question with an obvious answer, but I couldn't help wondering.

FSMCDesigns

In my experience normal maps are for the larger details and bump maps are for the finer ones.

Phoenix1966

This thread has a little more info.

 

Victor_B

Sorry, guys, but my English is not good, so I needed take a time to write this... Anyway.

Normal maps are usually created automatically in 3D software based on two versions of the model, low- and high-polygon. The low-poly model consists of a minimum of geometry (the basic shape of the object), and the high-poly model contains everything necessary for maximum detail. Then they are compared with each other, the difference is calculated and stored in a texture, called a normal map. After you created normal maps, you can also use the bump map (or height map) for very small parts, which even in the high-poly model can not be modeled (skin pores, other small depressions) and, therefore, they can not be in the normal map. But you can draw this small elements on the bump map in Photoshop, for example.

That's why we have so many character sets with both bump maps (height and normal). If I understand something in 3d World...

Jonstark

Thanks for the tips, sounds like there's a use for both map types, even on the same character I suppose, so I shouldn't assume for example that 'Normal maps are better' and just throw out the bump maps  :) 

FSMCDesigns

Jonstark said:

Thanks for the tips, sounds like there's a use for both map types, even on the same character I suppose, so I shouldn't assume for example that 'Normal maps are better' and just throw out the bump maps  :) 

Really depends. I develop for games and do mods and you rarely ever see bump maps in game engines. I think the reason we still see them in this community is just that it was the standard awhile back and now with increasing use of normal maps we are still seeing the bumps and some users are still finding uses for them. Try this link for more info

https://www.pluralsight.com/blog/film-games/bump-normal-and-displacement-maps

 

RawArt

I often use the different maps to add different types of details to a character. Each has their own particular strength and weakness.

Game engines are limited in their rendering capacity compared to D/S, so we can use all the different options to their best effect.

 

Victor_B

Jonstark said:

...so I shouldn't assume for example that 'Normal maps are better' and just throw out the bump maps  :) 

FSMCDesigns said:

... I develop for games and do mods and you rarely ever see bump maps in game engines.

Normal maps are really better. This is a new technology of bump mapping. 3D models with normal maps looks more realistic in comparison to 3D models with height/bump maps. We don't need very small parts in games because of productivity (and we even don't recognize them on the rather small props and characters in games especially in dynamic scenes), so that's why contemporary games uses normal maps mostly. But, again, you can use both maps for big and more detailed objects.

wolf359

Normal maps are the preferred method for any model that will be moving
(animated)and needs the Appearance of surface detail without CPU choking high polygon counts.

JonnyRay

Another reason game engines prefer normal maps to bump maps is that normal maps basically have calculations built into them. When you use a bump map, the rendering engine has to calculate how the surface normal would be changed based on the new height information. A normal map directly tells the engine how much to change the normal in X, Y, and Z axes without the engine having to calculate the differences. This saves on GPU cycles and therefore makes the engine more responsive.

If you're rendering a static image where the "cost" of calculating the changes due to a bump map are negligible, then there's nothing inherently "better" about normal maps over bump maps. In the end it's a matter of the quality of the model's base mesh and the creation of the texture map not the technology.

Personally, I've always preferred displacement maps since they actually change the geometry of the surface, so you get self-shadowing and profile changes that you wouldn't get with normal or bump maps. However, Iray's implementation of displacement requires a fairly dense model to work with and so the usefulness is somewhat limited in that engine compared to how 3Delight implements it.

DigitalSteam

Victor_B said:

Sorry, guys, but my English is not good, so I needed take a time to write this... Anyway.

Normal maps are usually created automatically in 3D software based on two versions of the model, low- and high-polygon. The low-poly model consists of a minimum of geometry (the basic shape of the object), and the high-poly model contains everything necessary for maximum detail. Then they are compared with each other, the difference is calculated and stored in a texture, called a normal map. After you created normal maps, you can also use the bump map (or height map) for very small parts, which even in the high-poly model can not be modeled (skin pores, other small depressions) and, therefore, they can not be in the normal map. But you can draw this small elements on the bump map in Photoshop, for example.

That's why we have so many character sets with both bump maps (height and normal). If I understand something in 3d World...

That was a very clear and helpful way to put it.  My 3D experience is a 10-20 years old now, and coming back to it, i wasn't sure about how Normals worked, or if they were an addition or a replacement. This helped perfectly. Thank you :)

And compliments on your English, too - I think it's quite good, especiallly for a second (or more) language) :)

DigitalSteam

JonnyRay said:

Another reason game engines prefer normal maps to bump maps is that normal maps basically have calculations built into them. When you use a bump map, the rendering engine has to calculate how the surface normal would be changed based on the new height information. A normal map directly tells the engine how much to change the normal in X, Y, and Z axes without the engine having to calculate the differences. This saves on GPU cycles and therefore makes the engine more responsive.

If you're rendering a static image where the "cost" of calculating the changes due to a bump map are negligible, then there's nothing inherently "better" about normal maps over bump maps. In the end it's a matter of the quality of the model's base mesh and the creation of the texture map not the technology.

Personally, I've always preferred displacement maps since they actually change the geometry of the surface, so you get self-shadowing and profile changes that you wouldn't get with normal or bump maps. However, Iray's implementation of displacement requires a fairly dense model to work with and so the usefulness is somewhat limited in that engine compared to how 3Delight implements it.

Also very helpful, thank you! 

(This forum really needs a like / upvote / helpful / thanks button :) )