I'm currently working on a project and, i'd be interested to hear the views of experienced texture makers as when they choose to use each or both, and if they care to share the techniques they use to create them from the basic texture map.

Use one of the two, not both. They're different methods to get a similar effect, each with their own advantages and disadvantages.

Using both is like putting both sugar and Nutrasweet in your coffee.  

I seem to see quite a lot of mats, that come with products I've bought, having both a bump and a displacement map attached.

I'd assumed these were being compiled for effect to a small degree perhaps... but more likely this was more so that if the end-user turned off displacement in the main render settings, there would still be some bump?

I use both. Bump for tiny details, displacement for the very noticeable, larger ones (but ONLY if those details are large enough to warrant it.) ;). Never had a problem using both, because both have completely different information. I have a lot of respect for stewer and what he says, but in this case I don't agree...hehe.

For a for instance, I'd use bump for skin pores, but I'd also use displacement for large scars if that helps.

Laurie

it's al about level of detail. Objects far away don't need either.

Bump is for objects further away, or for small detail nearby.
Displacement is for objects nearby of details very close.
For details even closer, you've got to model the effects into the mesh.

It does not harm to use much details futher away, it just costs resources and (render)time. In gaming, that's relevant. In large scale hires renders, it might not.

I'm not sure if this is technically correct in every detail, but I think of it in this way. Bump uses shadows to create an illusion that the surface has high or low points. Displacement gives the surface actual high or low points. The difference is most noticeable when a surface is viewed from a very acute angle.

The same map is used on both cubes in the above image. In the top cube the map is plugged into Bump, in the bottom cube it is plugged into Displacement.

It would be cool if there was a node that detected how far away from the active camera the mat zone it was loaded into was, eh? e.g. that you could use to vary LOD relevant effects...

Quote - It would be cool if there was a node that detected how far away from the active camera the mat zone it was loaded into was, eh? e.g. that you could use to vary LOD relevant effects...

Variables - Du, Dv, dPdu, dPdv.

Du and Dv tell you how long the edge of a micropolygon is in UV coordinates, dPdu and dPdv tell you how long it is in world coordinates.

Quote - I use both. Bump for tiny details, displacement for the very noticeable, larger ones (but ONLY if those details are large enough to warrant it.) ;). Never had a problem using both, because both have completely different information. I have a lot of respect for stewer and what he says, but in this case I don't agree...hehe.

One of the first things I learned about computer graphics: if it looks right, it is right.

Using bumps and displacements separately makes sense in renderers that don't use micropolygons, since there bumps are of a higher resolution than displacements. In FireFly however, small displacements are only slightly slower than bumps and you're not gaining anything by separating the two as opposed to using all displacements.

Attached Link: Stonemason's site

For techniques on texturing it's worth having a look at Stonemason's pdf on painting hard surface objects - from the link go to the Downloads section and it's on the second page. Really useful and gives good detail on creating data maps.

Both bump and displace maps do the same thing, they encode height information as grayscale data - each pixel has a value between 0 (black) and 255 (white). Both can be used for different effects at the same time if needed. Poser reads displacement as black = none and white = maximum (so to get gray as none, black as negative and white as positive involves the use of a math node as covered by a lot of other threads here). One of the main considerations will be whether you need to affect the silhouette of whatever you're texturing - as lesbentley shows a displacement map will alter the surface of the mesh while a bump map won't. Bit depth can be a consideration - an 8 bit grayscale map will have 256 levels between the extremes, so 256 steps from highest to lowest points. A 16 bit grayscale map however will have over 65,000 steps so is much higher quality but costs more in memory terms.

A straight grayscale conversion of a colour map is unlikely to give ideal results by itself but could make a good starting point. Bear in mind that reds tend to convert to lighter gray tones than blues, so altering the colours in the original map can have a big impact on the conversion (Photoshop's conversion dialog gives sliders for the different colour ranges). Then apart from altering the tonal balance with stuff like levels, contrast etc you could also paint onto the image, use light blurring to soften details and prevent artifacts, combine other textures - plus you can use Poser's procedural nodes like fbm, turbulence etc either by themselves or in combination with image maps.

As Laurie said pretty much, for my opinion on the subject. basically anything that will show as raised or depressed in profile, displacement... a good way to model geometry resolution independant details otherwise impossible (hence ZBrush generated displacement maps).

Good commentary and insight as usual, caisson.

"and if they care to share the techniques they use to create them from the basic texture map."

Well, if you painted the texture map from scratch and don't have to worry too much about burnt in specular, things will certainly go alot easier. Biggest problem I can see is eliminating lighting effects on the 2d map in order to create precise bump or displacement. In my case this would mostly be a concern with creating bump for close skin detail effects. Probably 100% painted maps are the most flexible route to go in particular for generating derivative greyscale maps... of course this is a tremendous amount of work and is probably best utilized for creature skins or alien, goth, cyborg, toon type skins for figures. Then you don't have to worry so much about the photorealism thing, i.e. people expect a pin-up character type skin to be derived from a merchant resource (as about 99.9% of them are), which in turn leads to the 3d.sk photoreference method. I'd say the toughest part of texture creation is double checking seams and cloning details and fudging to get things just right.

Don't discount the use of procedural methods and masks in combination with maps either... e.g. turbulence, granite, noise, spots, etc.

Another maybe obvious point would be to keep everything you create as separate layers, and I mean everything... including symmetry details. You'll end up with way more flexibility and can begin to create your own merchant resource as it were. Downside, gigantic .psd (or whatever) files...

Great example Les!

Here's how I understand it:

The effect you get from a bump map is really just an illusion, a trick of the light. Highlights and shadows appear to be caused by projections and depressions in the surface, but the surface is actually flat. Quite often that's all you need.

A displacement map, on the other hand, really does move the surface in and out to create actual lumps and hollows. As Les's example shows, this affects the profile or silhouette of the object.

I generally try to get away with using bump first, because it is so much more efficient in terms of memory usage. I use displacement when I know bump won't look convincing, such as when the material will be seen as a silhouetted edge, or when sharp shadows will be cast onto the material from other objects.

For example, if a flag-pole cast a shadow onto Les's 2 cubes, its shadow would accurately follow the irregularities of the displaced surface, but the shadow would appear straight and smooth on the bump mapped surface, ruining the illusion of irregularities.

thanks chaps, lots of useful snippets and opinions in here so far, I suppose really you are right, the proof of the pudding, is if it looks right when it renders.......

Here's an example of one of the texture sets (face in this example) I've been working on which shows how I keep all of the painted details on separate transparent layers... all of the textures are 100% painted by myself. Part of a couple of character sets I've been working on here and there amongst other modeling projects or non-cg/Poser related endeavors. Still a ways off from being finished in the texturing dept. but the morphs are just about done and ready for INJ creation. Just having a blast drawing various tattoos and patterns, character specific motifs.

I've found that the most difficult thing with from scratch texture creation is F'ing eyebrows! Eyes, no problem, teeth and tongues, no problem... eyebrows, ahhhhg! Luckily, there's some ways around this with custom brushes and the freeform pen tool but it's still a difficult task. Fantasy eyebrows are a cinch but convincing everyday eyebrows are an entirely different thing. Anyway, my custom texture layered files weigh in at about 300 mb's. The separate bump and displacement layered files are still in the gestation stage so aren't quite as heavy... All of my skins are more toward the fantasy, toon type with a lot of elaborate tattoos and such so I'm not as concerned with the photo accuracy of broken blood vessels in a millimeter stretch of flesh at the corner of a nose... something that seems to give Poser forumites a hard-on for whatever reason.

All of this is very labor intensive but I foresee a great time saving in the long run, plus the ability to create things that you can say are 100% your own is a bonus!

One final note, regardless of how you go about doing this I can't stress enough the importance of looking at, learning from, and being inspired by the works of other artists. At the end of the day observation is probably your strongest ally, asking questions is fine but generally amounts to how articulate, the mood, and the willingness to share the answer giver is. Direct observation speaks for itself.

Displacement all the way, baby.  :)  

@ Primorge - your post highlights many of the reasons why I gave the hell up on texturing.  Even working from photos as a starting point, it's a royal pain in the arse.

For eyebrows, I used brush dynamics with a reducing brush size and a gradual fade.  Painted them pretty big then reduced to the texture scale.  That, or use a bloody photo.  :D  

 

Thanks for the tip Sam, I think that photo's are probably the simplest solution to eyebrows. I'm kinda on the 100% painted angle with this one so I'll follow your suggestion and fiddle with the dynamics a little more. I do have a huge collection of brushes but for some reason they are just giving me the most difficulty, I think it has something to do with rendering them onto transparent layers for flexibility and cross-texture usage. If you look at all of the merchant resources for eyebrows you'll notice that the brows are attached to the underlying skin, consequently you have to do  all kinds of adjustments in order to match them to your texture. Sigh.

"It would be cool if there was a node that detected how far away from the active camera the mat zone it was loaded into was, eh? e.g. that you could use to vary LOD relevant effects..."

 

Maybe it could somehow take into account the angle as well. Seems like a "smart" renderer could do a lot to optimize resources bu automatically choosing geometry detail, maps etc.

There's one thing for certain, I have a new respect for texture makers, not that I had no respect before you understand, but at first glance it's just a picture, errrr with many, many layers Mahoney, to all intents and purposes that "simple" texture map is going to make or break your model. I've decided to go the "hand made" route for my texture map, pretty much because I haven't the technical knowledge to create all my own procedurals from scratch, however the option of a good selection of mat zones then gives the end user plenty of scope to use more complex shaders should they wish. I also discovered the "fun", hours of joy?, that getting a Good UV  map entails, closely followed by weeks of texture shopping, rebuilds, remaps,reshop, re render, rebuild, remap ad infinitum...Lets do it again   LOL

Yeah, welcome to the wonderful, fascinating, FUN, world of texture creation! Especially if you're doing a body texture. It's running Poser and an art program at the same time. It's fiddle, fiddle, refresh textures in Poser, and re-render. Hmmmm, I'm reminded of a W.A.S.P. song from the '80's, "Scream Until You Like It"! Come on in Saxon3d, the water is boiling! :)

Quote - I'm currently working on a project and, i'd be interested to hear the views of experienced texture makers as when they choose to use each or both, and if they care to share the techniques they use to create them from the basic texture map.

What sort of mapping am I going to use? How much texture space do I have? What is the level of detail expected to be when rendering the object?

I would use bump for light surface features and textures, displacement for heavier ones. Sure, displacement will actually create geometry for the renderer to render, but that's not always needed and can sometimes slow things down. For something like skin being viewed from a couple of yards away, I don't need displacement. But, up close, I would prefer it.

Having both is not a crime. :D

Quote - I would use bump for light surface features and textures, displacement for heavier ones. Sure, displacement will actually create geometry for the renderer to render, but that's not always needed and can sometimes slow things down.

A micropolygon renderer will create that geometry anyway, you're not really saving much by using bump maps instead.