View Full Version : Displacement mapping vs. detailed meshes
what are the pros and cons with with displacement mapping vs. detailed meshes
It depends, If your object is to be a static object, you can use a detailed mesh, or normal maps if you want to consume less resources.
If your model is a character that you will animate, you shoud model the details and then use another program, such as ORB or XNormal, to bake the details into a texture, and then apply that texture to the lowres model as a displacement map.
Blender MultiResolution modelling does not support armature animation, the details vanish or just go crazy.
Here's a thread I started with a little tutorial on how to bake displacement maps with ORB
http://blenderartists.org/forum/showthread.php?t=96746#
Here's another thread that is about the same issue.
http://blenderartists.org/forum/showthread.php?t=97994
Jeremy Ray
17-Jun-07, 01:46
Since Blender doesn't have micro-poly displacement you have to crank the subsurf up crazy high to get anything approaching decent results out of Blender's displacement so you might as well not even think about it 99% of the time.
Lord of the Rings Junkie
17-Jun-07, 02:02
Since Blender doesn't have micro-poly displacement you have to crank the subsurf up crazy high to get anything approaching decent results out of Blender's displacement so you might as well not even think about it 99% of the time.Yeah, a displaced model might even have more polygons than one where everything is modeled...:(
Absolutely, in blender displacement is restricted to subsurf levels, but I really think that you can achieve great and really professional results without micro-poly displacement.
It'd great that Blender had that feature, but with a little extra thinking and planning and some normal maps for small details you can do some great stuff.
Jeremy Ray
17-Jun-07, 04:59
but with a little extra thinking and planning and some normal maps for small details you can do some great stuff.
See, that's the fundamental problem with the volunteer/open source software model. If there is a twenty-hoop way to do something, we call it done. Meanwhile, in the prorietary/paid laborer world, they have a thing called "useability." They produce apps that are really slick, intuitive, and time-efficient to use.
When you are trying to do work for real, it counts.
I give Ton credit for being the best of the open source software developers, of the programs I've tried. And he's come up with great interface ideas which proprietary companies would do well to imitate. But let's be honest about where we're at. Blender may be as much as a decade behind in its displacement technology. What Blender has - sucks.
The volunteer model is a problem. We need to hire programmers to work on Blender full time. I'm hoping XTIN project will help solve the money problem with the tithing model. If we can get a few successful Open Movies going and establish the tithing model for use of Blender in commercial products, maybe we can get Blender where it needs to be.
Felix_Kütt
17-Jun-07, 05:14
Jeremy dont worry, havent you noticed that theres a blender institute now(or will be, and the interesting thing is that it has some thing to do with the year 2012[EU grant?]) in there they have artists working - geting payed? oh housing and umm traveling costs payed[EU Grant?] and some 2-3 programmers actively working(getting payed?[EU Grant?]) :eek: well, something sure seems kinda hmm.. ah never mind... :p
edit:
http://www.blender.org/blenderorg/blender-foundation/2007-plans/blender-institute/
Research/development
Permanent development or research positions, 1 - 3 people.
A little off topic but 2-3 programmers for a program as large and diverse as blender really isn't much. Now these guys can be more like managers to work with the community of volunteers then it would work, but just having these guys as coders - you would need more people.
That's a wonderful thought, but there are all those vounteers donating time and effort, too. You should check the recent release logs. The 2-3 paid programmers would be more like whipped-cream, or gravy, depending on what food metaphor we're using.
Felix_Kütt
17-Jun-07, 12:44
yet those programmers can work on blender all day, and are chosen by their skills 'nstuff, so i think its better than nothing ;) sure it would be cool if there could be more but..
The volunteer model is a problem. We need to hire programmers to work on Blender full time. I'm hoping XTIN project will help solve the money problem with the tithing model. If we can get a few successful Open Movies going and establish the tithing model for use of Blender in commercial products, maybe we can get Blender where it needs to be.I noticed that whenever the topic of donating to the Blender Foundation comes up, the thread suddenly halts. Very few are interested in helping out. I don't even call it donating. I call it a quid pro quo. You give something and you get something in return like buying BF's new book, Essential Blender. Even the so-called conferences or user group meetings are not geared towards the development of Blender. Some guy even calls them "parties". If these parties would just make a point of adding a donation box at some corner of the place, well. It may not be much but at least you'd trigger a spark of interest in it.
Jeremy Ray
17-Jun-07, 17:23
Felix_Kütt - Then I'm not the only one who sees the obvious :)
2-3 Full time programmers is a great start.
snelleeddy
17-Jun-07, 17:34
On toppic, displacement maps are good to add some volume to large objects, like water or to make soil look less flat. For the minute precise thing you should model it or use Nor maps imo.
Off toppic, every bit helps.... and I think that the institute will be able to move some decent amount of work.
Quick note, just in case this needs clarification: I'm not actually saying 2-3 full-time programmers would be useless because of the volunteers, I was just disagreeing with Mr Rage that they'd be useless because there'd only be a few.
Felix_Kütt
18-Jun-07, 05:35
Quick note, just in case this needs clarification: I'm not actually saying 2-3 full-time programmers would be useless because of the volunteers, I was just disagreeing with Mr Rage that they'd be useless because there'd only be a few.
ah, sorry a little misunderstanding on my part. though i guess my earlier post still goes as an answer to Mr Rage :o :eyebrowlift2:
edit:
Felix_Kütt - Then I'm not the only one who sees the obvious :)
2-3 Full time programmers is a great start.
yes, yes it is! and even more than just a great start! ;)
[QUOTE=Jeremy Ray;891974]See, that's the fundamental problem with the volunteer/open source software model. If there is a twenty-hoop way to do something, we call it done. Meanwhile, in the prorietary/paid laborer world, they have a thing called "useability." They produce apps that are really slick, intuitive, and time-efficient to use.
QUOTE]
Sorry to disagree with you. But that's not true at all. We has fake SSS for some time now, but look what happened in blender 2.44? Also, artists could fake blurry reflection with bump-maps, but broken just implemented a 'real' blurry reflection/refraction path in blender. Being an open software, I think blender is actually more open to ideas from other commercial softwares, we can literally copy the features, micropoly, in this case (re-writing it ourselves, of course) and not having to worry about legal issues.
The thing with open source is that when they come out with something good and new it gets copied and rarely on the commercial side of things programs that are considered to be an industry standard becomes pirated to an insane degree and thus they have to increase the price to an excessive amount which causes even more problems, legal issues, and such...with programs like photoshop it became too expensive for the average person to go out and buy it so someone made a program that achieves similar results and was kind enough to just give it to people for free. Also, programmers can help out in any way they see fit to help said programs become better.
Open Source also tends to support old and new hardware more so than
closed source software does. I mean you basically need a new pc to run max and the program will cost you alot more than that to buy.
High resolution polys are good for lighting because shadow mapping doesn't always take account of displacement maps. Similarly for environment maps.
High polygons use a lot of Ram though and if you are raytracing, you could easily start paging your drive.
Displacement maps are good if you need to layer displacement effects though or need to adjust them easily. Think of two dinosaurs with rough scales on their skin. Are you really going to model two sets of fine scales or just paint two maps and use the same basic mesh?
Not to mention, animated displacements can be done with maps too.
But then if you have characters that interact with the displaced mesh, polys are easier to use there.
Projects in computer graphics are problem solving exercises and you use whatever technique gets the results you want and this judgement cannot be made generally. This is why applications like Blender can be used to produce good work. Blender has enough flexibility with Python and now composite nodes to solve problems.
I agree that it is lacking in relation to some commercial packages in a few areas but in others it is better. The more that Blender moves towards a modular/programmable structure regarding nodes and shaders, the more flexible it will be and therefore easier to overcome problems.
Micropolygon rendering would be nice for more reasons than displacement (no more tessellation artifacts ever again) but they already tried to put it in and they obviously had technical issues. It's quite a fundamental change to the way the core rendering works. If they can't optimize it enough then it slows renders down unnecessarily for people who don't take advantage of it.
It's the same with texture rendering. Other renderers render textures sharper than Blender but Blender is quicker. Turn on full OSA and Blender's texture quality goes up but it is much slower. If you can get away with the lower quality then the performance increase is a benefit. Blurry raytraced shadows are nicer but if you can get away with shadow-maps, it's better to use them instead.
Isn't Micropolygons in the development pipeline now? Or is it not the same thing.
http://blenderartists.org/forum/showthread.php?t=119216.
http://www.graphicall.org/builds/builds/showbuild.php?action=show&id=620
Felix_Kütt
10-May-08, 08:16
Isn't Micropolygons in the development pipeline now? Or is it not the same thing.
http://blenderartists.org/forum/showthread.php?t=119216.
http://www.graphicall.org/builds/builds/showbuild.php?action=show&id=620
Its not actually quite same, but its not a bad thing to have either, definitely has its use. ;)
on-topic:
And one thing to think about is that you can use both a displacement map to define the general shape,
and a normal map to define the fine detail.
renderdemon
10-May-08, 09:46
on-topic:
And one thing to think about is that you can use both a displacement map to define the general shape,
and a normal map to define the fine detail.
This isn't completely true,as displacement is really limited in allowing extreme variations with the base mesh,displacement work only on normal vector,base mesh and displacement generated mesh should be very similar in shape to work.
Michael W
10-May-08, 11:00
This isn't completely true,as displacement is really limited in allowing extreme variations with the base mesh,displacement work only on normal vector,base mesh and displacement generated mesh should be very similar in shape to work.
True, but it's pretty common to prepare data in just this way
Eg i n z-brush you might produce a displacement map of the differences between the cage and subdiv5 and a normal map of the differences from subd5 to 7 for example (where 7 is the most refined/detailed level in this example)
This is still commonplace, even though now there's re-topologizing going on more often than not to complicate things further!
On the programmers thing, 3 fulltime working as a team is pretty good really, even in commercial products, yeah the team as a whole in commercial product may be ten times that in the credits, but the core is usually pretty small.
See, that's the fundamental problem with the volunteer/open source software model. If there is a twenty-hoop way to do something, we call it done. Meanwhile, in the prorietary/paid laborer world, they have a thing called "useability." They produce apps that are really slick, intuitive, and time-efficient to use.
To be fair, the internal renderers of most other 3D apps can't handle displacements well enough to be worth using either.
The only commercial renderers that have what I would consider "usable" displacement are Pixar's Renderman and other Reyes-based renderers (like RenderDotC, 3Delight, Houdini's internal renderer, etc.).
I played with MentalRay quite a bit when I was in school, and their displacements are slow, as well as time consuming to tweak properly. I wouldn't want to use its displacements in a production environment. I suspect the same is true of almost any non-Reyes renderer.
On the other hand, there are stand-alone open source renderers that do displacements really well (namely Pixie and Aqsis). So I don't really think the open-source world is as far behind in that particular feature as you think.
displacement work only on normal vector,base mesh and displacement generated mesh should be very similar in shape to work.
Not by definition. It depends on what software you're using. Renderman compliant renderers, for example, can displace meshes arbitrarily, and aren't limited to displacing along the normal vector.
renderdemon
10-May-08, 19:10
I know what renderman compliant render can do,but programs that do displacement baking(high res to lowres)basically find a scalar value in the normal direction,without changing the baking code(in Blender,zbrush,xNormal and so on)you can't use with advantage vector displacement(obviously I'm not speaking about procedural,but mainly about sculpting displacement)
vBulletin® v3.7.3, Copyright ©2000-2008, Jelsoft Enterprises Ltd.