• Announcements

    • khawk

      Download the Game Design and Indie Game Marketing Freebook   07/19/17

      GameDev.net and CRC Press have teamed up to bring a free ebook of content curated from top titles published by CRC Press. The freebook, Practices of Game Design & Indie Game Marketing, includes chapters from The Art of Game Design: A Book of Lenses, A Practical Guide to Indie Game Marketing, and An Architectural Approach to Level Design. The GameDev.net FreeBook is relevant to game designers, developers, and those interested in learning more about the challenges in game development. We know game development can be a tough discipline and business, so we picked several chapters from CRC Press titles that we thought would be of interest to you, the GameDev.net audience, in your journey to design, develop, and market your next game. The free ebook is available through CRC Press by clicking here. The Curated Books The Art of Game Design: A Book of Lenses, Second Edition, by Jesse Schell Presents 100+ sets of questions, or different lenses, for viewing a game’s design, encompassing diverse fields such as psychology, architecture, music, film, software engineering, theme park design, mathematics, anthropology, and more. Written by one of the world's top game designers, this book describes the deepest and most fundamental principles of game design, demonstrating how tactics used in board, card, and athletic games also work in video games. It provides practical instruction on creating world-class games that will be played again and again. View it here. A Practical Guide to Indie Game Marketing, by Joel Dreskin Marketing is an essential but too frequently overlooked or minimized component of the release plan for indie games. A Practical Guide to Indie Game Marketing provides you with the tools needed to build visibility and sell your indie games. With special focus on those developers with small budgets and limited staff and resources, this book is packed with tangible recommendations and techniques that you can put to use immediately. As a seasoned professional of the indie game arena, author Joel Dreskin gives you insight into practical, real-world experiences of marketing numerous successful games and also provides stories of the failures. View it here. An Architectural Approach to Level Design This is one of the first books to integrate architectural and spatial design theory with the field of level design. The book presents architectural techniques and theories for level designers to use in their own work. It connects architecture and level design in different ways that address the practical elements of how designers construct space and the experiential elements of how and why humans interact with this space. Throughout the text, readers learn skills for spatial layout, evoking emotion through gamespaces, and creating better levels through architectural theory. View it here. Learn more and download the ebook by clicking here. Did you know? GameDev.net and CRC Press also recently teamed up to bring GDNet+ Members up to a 20% discount on all CRC Press books. Learn more about this and other benefits here.
Sign in to follow this  
Followers 0
AzraelilMeraz

[SOLVED] Hair simulation. Verlet/Constraints

2 posts in this topic

EDIT2: SOLVED, there is nothing wrong with the algorithm - I accidentally ran the constraint generation kernel every frame

Hello!
I've been toying with OpenGL and OpenCL trying to simulate Hair on the GPU.
However I've run into issues that are not OpenGL/OpenCL specific - I seem to be incapable of writing a correct solving algorithm for distance constraints.
I'm not sure where the issue is - maybe I'm generating the constraints incorrectly, maybe it's the solving.
That's how I'm generating the constraints:
[source lang="cpp"]// create constraints from original positions
__kernel void buildConstraints(__global const float4* in, __global DistanceConstraint* out, const int verticesPerHair)
{
int id = get_global_id(0);
if((id % verticesPerHair) != 0)
{
out[id].id1 = id-1;
out[id].id2 = id;
out[id].d = length(in[id-1]-in[id]);
}
else
{
out[id].id1 = id;
out[id].id2 = id;
out[id].d = 0;
}
}[/source]

(id % verticesPerHair) != 0 means that the hair is a root. All but the root is constrained to the preceding vertex in the hair. The root is constrained to itself so it won't move

Prior to constraining the vertices the verlet integration is computed on every vertex except the root:

[source lang="cpp"]
// verlet
if(idl != 0)
{
float4 difference = localpos[idl] - oldpos[idx];
oldpos[idx] = localpos[idl];
localpos[idl] += (1-friction)*difference + force*pdt*pdt;
}[/source]

Then to satisfy the constraints I run fixed number of iterations of this code:

[source lang="cpp"]barrier(CLK_LOCAL_MEM_FENCE);
DistanceConstraint constraint = constraints[idx];
for(int i = 0; i < dcIter; i++)
{

if((idl%2) == 1)
{
distanceConstraint(localpos, constraint.id1 % verticesPerHair, constraint.id2 % verticesPerHair, constraint.d);
}
barrier(CLK_LOCAL_MEM_FENCE);

if((idl%2) == 0)
{
distanceConstraint(localpos, constraint.id1 % verticesPerHair, constraint.id2 % verticesPerHair, constraint.d);
}
barrier(CLK_LOCAL_MEM_FENCE);
}
[/source]

constraint is what was generated for this vertex in buildConstraints(). The barrier() function is there because the algorithm runs in parallel, so it solves the constraint for odd vertices then waits for all threads to complete and then solves the constraints for even vertices.

the function distanceConstraint() looks like this:
[source lang="cpp"]
// satisfy distance constraints
void distanceConstraint(__local HairVertex* pos, int idx1, int idx2, const float conDist)
{
// vector from vertex 1 to vertex 2
float4 difference = pos[idx2].position-pos[idx1].position;
// By what amount move which vertex
float2 cFactors = constraintFactors(idx1, idx2);
// length of the vector between the vertices
float len = length(difference);
// distance the vertices must move towards each other
float distance = len-conDist;
// no corruption through div by zero
len = max(len,1e-7f);

// move
pos[idx1].position += cFactors.x*distance/len*difference;
pos[idx2].position += cFactors.y*distance/len*difference;
}[/source]


constraintFactors() returns factors that determine how to handle vertices depending on whether or not they are roots:
[source lang="cpp"]

// for given indices id1 and id2 return constraint factors
float2 constraintFactors(int id1, int id2)
{
if(id1 != 0)
{
// first vertex is not a root
if(id2 != 0)
{
// both vertices not a root
return (float2)(0.5f, -0.5f);
}
else
{
// first vertex is not root, second is root
return (float2)(0.0f, 0.0f);
}
}
else
{
// first vertex is root
if(id2 != 0)
{
// first vertex is root, second is not
return (float2)(0.0f, -1.0f);
}
else
{
// both vertices are roots (shouldn't happen)
return (float2)(0.0f, 0.0f);
}
}
}[/source]

I have looked through NVIDIAs DirectX implementation of the simulation (NVIDIAs SDK 11) and the only deviation from the constraint code was that when the first vertex is not root and the second is, the factors were (1.0f,0.0f). I am however not certain how the constraints were generated, so I predict that my mistake is somewhere there.

The actual problem is that after a while the simulation goes full spaghetti mode... To be a bit more precise - the control vertices seem to move from the root towards the tip and the tip seems to move away in the direction the force is applied... Here's what it looks like in point mode:

[url="http://sw-ores.de/projecthorizons/spaghetti.mkv"]http://sw-ores.de/pr...s/spaghetti.mkv[/url]

I have no clue what I'm doing wrong

[img]http://public.gamedev.net//public/style_emoticons/default/sad.png[/img] Edited by AzraelilMeraz
0

Share this post


Link to post
Share on other sites
I thought that maybe it's because of the parallel execution of the Distance constraints in the loop, so created a serialized version, where only one thread solves the constraints:
[source lang="cpp"]
barrier(CLK_LOCAL_MEM_FENCE);
if(idl == 0)
{
for(int i = 0; i < dcIter; i++)
{
for(int i = 1; i < verticesPerHair; i++)
{
DistanceConstraint constraint = localconst[i];
distanceConstraint(localpos, constraint.id1 % verticesPerHair, constraint.id2 % verticesPerHair, constraint.d);
}
}
}
barrier(CLK_LOCAL_MEM_FENCE);[/source]

This yielded the same results however :/. I've looked into several resources on the topic:

[url="http://www.connellybarnes.com/documents/hair/#Verlet_Integration"]http://www.connellyb...let_Integration[/url]
[url="http://dl.acm.org/citation.cfm?id=1837101.1837102"]http://dl.acm.org/ci...1837101.1837102[/url]
[url="http://en.wikipedia.org/wiki/Verlet_integration"]http://en.wikipedia....let_integration[/url]
[url="http://www.gamasutra.com/resource_guide/20030121/jacobson_pfv.htm"]http://www.gamasutra...acobson_pfv.htm[/url]

But I see nothing on the issue at hand. I'm using 20 iterations for the constraints, but the same happens if I do 200 or 2000 - just at a slower rate!
0

Share this post


Link to post
Share on other sites
Nevermind... figured it out - there is nothing wrong with the algorithm - I just accidentally ran the constraint generation kernel every iteration...
0

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0