Why displacement maps in A:M?

#1

hi,
one question.
what are displacement maps in A:M good for?
In my understanding the use of d-maps is to get a “light” mesh with lot of detail. well i tried it… the result was very dissapointing. to get something i would call “ok, but not great” i had to use a mesh that was far from beeing animatable. the density was so high i thought i was in a poly-program :frowning:
i don´t expect zbrush quality but d-maps in A:M are simply useless i would say.

to show you an example of what i mean http://www.herr-der-ringe-film.de/v2/de/specials/effekte/kreaturen-1_1.php
i don´t know a way to make such complex geometry without making an insane dense mesh.

please prove me wrong.

greetings
oliver

edit: i´m working with 10.5

#2

Nope…sorry, wish I could prove you wrong :hmm: but I can’t.

AM converts the models into polygons for rendering using adaptvie tesselation (basically adds as many polygons as needed to get a smooth looking surface). Unfortunately it doesn’t add extra polys to get all the details needed for you displacement maps to look nice…and if it did it would probably use up all your RAM and time.

Some renderers, like mentalRay and RenderMan, support sub-pixel displacement/tesselation, which is what you need to get results like the one in the link you posted. I’ve sent this as a feature request a long time ago to Hash, maybe more people should too. :slight_smile: It’s a whole lot faster than standard poly tesselation and has a lot more detail.

If it makes you feel any better, most other renderers don’t have this either, including for instance the Maya renderer. That doesn’t mean AM shouldn’t have it though. :rolleyes:

Ragnar

#3

In that troll example the 1st picture is the low-res mesh and the picture to the right of it is the high-res mesh that was scanned from the sculpture.
The picture below the 1st picture is what the displacement map looks like applied to the low-res mesh. The picture to the right of it is what it looks like with the color and bump applied also.

You can see that the displacement mapped image isn’t as detailed as the laser scanned model, but is much simpler to handle the low-res model with displacement.

Displacement maps in A:M work exactly the same as that example.

#4

hm the picture with displacement map and low-res model. how to make this in A:M? .

You can see that the displacement mapped image isn’t as detailed as the laser scanned model, but is much simpler to handle the low-res model with displacement.
i mean you have to use a LOT of patches to get this structure. so you can nearly model it.

i still don´t get the use of displacement in A:M. you have to make such a dense mesh the advantage of splines is negated. it´s like working with polys without the poly-tools.

in short displacement-maps doesn´t make any sense right now in A:M?

(i don´t want to flame A:M i just want to know if i´m wrong)

#5

the displacement map divides the patch 4x4 so that is 16 times the resolution of the patch alone…So the resolution of the displacement map does depend on the underlying patch resolution.
It will let you add detail, but you still have to model the basic shape…If you also add bump maps on top of the displacement map, you will have a very detailed surface that is still not too dense.

#6

That model in the example (if my German servers me right) is from LOTR and on it they used RenderMan to render. Wich means you got sub-pixel displacement maps going on there. Which basically means it’s dividing the surface into smaller polys than the image resolution (more than a poly per pixel). Which is probably more than you would get with AM. (Zandoria: Are you sure it’s 4x4, and not variable depending on how close to the camera the patches are?)

Secondly, they used a process for that film which involved comparing the high-res scanned surface of the model with the low-res model, to generate a perfect displacement map (something that would be extremely hard to simply paint in PhotoShop). There is a script available to do this in Maya, but I haven’t heard of any thing like it in AM. Of course, you would also need a sculpture of the figure and a laser scanner to achieve this.

Added to that you got people who spent ages on modeling this and tweaking it to get it just right. (e.g. the guy that did Gollums face spent about 1-2 years just doing Gollums facial expressions and face, of course it looks good).

Ragnar

#7

A:M’s displacement isn’t variable. It divides up each patch 4x4.

I didn’t mean to imply that it is “identical” to the method used to create the LOTR troll. I meant that the same principal of adding extra detail to the model with displacement maps is true in A:M.

#8

argh deleted my post :-/
ok once again…

thanks for the replies

@ragtag
sure the cgi in LOTR was build with a lot of manpower. i´m not so foolish to belive i can make gollum in a week :wink:
the trollskin was only an example. just imagine the cloud renderer in photoshop for a displacement skin. or imagine a dried out desert ground with the cracks in in it. how can you make this with d-maps in A:M? or even worse some well defined creases or a mechanical surface (typo, geometric objects,…). the result has such a high polycount… in polyprograms you get more or less used to a rather dense mesh. you have the tools to handel them with relative ease.

the first time i used d-maps in A:M was when i tried the pumpkintutorial from www.dvgarage.com (sorry no direkt link to the tutorial. i downloaded the quicktime a while ago). the steps where so easy to understand. the displacement map was quick made. then i applied it on the basic geometry. hm… something wrong. higher mesh, higer mesh, once again higer mesh… this is ridiculous now i have a mesh so dense i can model it all by hand (even with all the bumps and knots).
some basic pictures are following:

@zandoria
quote from an other website:
"Displacement/Fractal maps dynamically subdivide the patches they are applied to, limited to 16 subdivisions per patch. For more detail, large patches may have to be constructed out of many smaller patches.

Work as designed …"

sounds variable to me.
i can remember that between some versions of A:M the way d-maps are calculated was changed. but i´m not totaly shure with it…

#9

well, how about letting the user decide about
how many subdivisions are enough or best lookin for each displacement-map?

would that be a good feature request?

#10

That would be a great feature request, although it should definately come under the advanced features tab. You should also be able to set it for each model, so that you’re not rendering a higher poly count for the entire scene just to get that one pumpkin looking right.

Ragnar

#11

Here is a simple pumpkin shape displaced in A:M–using the example map you provided.

http://www.zandoria.com/images/tutorial/AM-displaced.jpg

#12

A:M’s displacement mapping is not very usable at the moment. And yet, good displacement & normal mapping would be incredibly beneficial for A:M users who wish to do high detail modelling. Splines alone are simply not suitable for high detail.

#13

I don’t see how you draw that conclusion…What is it that you don’t think is suitable?
The example I just showed demonstrates the extra detail (16 times!) that you can add to a mesh with a displacement map!
If someone were to recreate the LOTR troll, using A:M-- I believe that most of the people on this forum would still cry about how A:M is holding them back from creating any art.

If only A:M added the <fill-in-the-blank>, then it would be great:cry:

It is very disappointing…Time for me to take a break from this forum I guess.

#14

zandoria: Creating 16x the amount of detail with displacement maps isn’t enough. For an example of the kind of detail I’m talking about, take a look at the recent ZBrush displacement/normal-mapping thread. Don’t get me wrong… I love A:M. I just don’t think it is currently possible to use it to create extremely high-detail models. But that’s the thing about CG apps… They all have their weaknesses, right? Hence “Feature Requests”;).

#15

Ok Im a newbie so dont shoot me. But it looks like Zandoria showed that this was possible in AM-thanks by the way! But the past month Ive gone back and read alot of the older posts to learn as much on hash that I could. But it seems no matter what Hash does-they cant catch a break. Users said the render needs to be better-Hash did this. For me AM is stable. It shows they are listening! :> But I noticed that some people say one nice thing then say a bad thing to. Its like a weird vendetta or something. I hope you dont leave Zandoria-learned a lot from your posts. There are a lot of good people here-I hope no one leaves. :slight_smile:

#16

binder3d: You will notice that zandoria’s example is inferior to the other one. It’s A:M’s fault, though, not his. As long as you don’t need extremely detailed geometry, however, A:M is great:thumbsup:.

#17

@zandoria

ah come one. this thread doesn´t was started to flame A:M with the usual “seen new features in other programs -> why has A:M not this function -> it sucks…” my intention was not to rebuild LOTR, but create some easy pumpkin :wink:
my first intention was not that i was unable to create something in A:M. i wanted to use some tool in A:M and it didn´t behaved as i expected it and later on i didn´t seen the use of this tool in production at all.
ok i don´t want to start a flame war i really like the way A:M is developing and i don´t want to attack anybody.

thx for you effort to make the model.

but you can see on your model how hard the precise use of d-maps in A:M is. at the end of your punpkin the mesh gets naturaly denser. that means more detail and deeper creases (not THOSE creases :wink: compared to the middle. so to get even creases you have to crank up the mesh more and more (with the result of a much to dense mesh at the ends or a lot of hooks)
I don´t mean it´s impossible to get the result of the example by dvgarage, but it´s not really worth the effort. modelling it by hand and bumpmaps is a lot preciser and quicker.
for me i made the decision that i don´t use d-maps any more. maybe i´m wrong but the effort to get the expected result is not worth it. i take the classical way model it out.

at one point i was wrong. we now have the ability to control hight density meshes. thanks to weight mover by anzovin studio. i don´t know how good this program is but i sounds really impressive on paper.

so my rant about high complex meshes with d-maps may be outdated :wink:

#18

How will weight mover be able to help with d-maps in AM? Are you saying that d-maps might not be needed? Thanks-

#19

It’s not that weight mover helps with d-maps, but that it helps with higher density meshes, which allow d-maps to tesselate better. In other words a displacement map will look better on a 10,000 patch model then it would on a 1,000 patch model.

Weight mover lets you use a low patch proxy to more easily weight a complex model.

#20

Thanks Myfault now I understand.