next irrlicht release requestes

Discuss about anything related to the Irrlicht Engine, or read announcements about any significant features or usage changes.
kburkhart84
Posts: 277
Joined: Thu Dec 15, 2005 6:11 pm

Re: normal mapping on animated meshes

Post by kburkhart84 »

esaptonor wrote:
buhatkj wrote:i would like to see normal mapping on animated meshes. in addition to shadows on non-animated meshes.
basically i want to be able to have normal mapping, animation, and shadows, or any combination thereof, on any mesh.
i fully agree, It would allow some awesome next generation style games, and irrlicht needs some good looking stuff like this for its screenshots and tech demo. But is it hard to do all this?
That would be nice. I wonder though, due to the animation, wouldn't you need to calculate the binormals/tangents/whatever they are every frame of animation. I'm sure it could be done, but that would be a good reason not to use it on animated models.
belfegor
Posts: 383
Joined: Mon Sep 18, 2006 7:22 pm
Location: Serbia

irrEdit Viewport Camera movement

Post by belfegor »

it would be very nice to add 3dsMAX like camera movement (alt + middle mouse) in irrEdit i it has better control over scene viewing/creation.
EmmanuelD
Posts: 25
Joined: Thu Aug 24, 2006 8:34 am
Location: Right in front of Dagoth Ur
Contact:

Re: normal mapping on animated meshes

Post by EmmanuelD »

kburkhart84 wrote:
esaptonor wrote:
buhatkj wrote:i would like to see normal mapping on animated meshes. in addition to shadows on non-animated meshes.
basically i want to be able to have normal mapping, animation, and shadows, or any combination thereof, on any mesh.
i fully agree, It would allow some awesome next generation style games, and irrlicht needs some good looking stuff like this for its screenshots and tech demo. But is it hard to do all this?
That would be nice. I wonder though, due to the animation, wouldn't you need to calculate the binormals/tangents/whatever they are every frame of animation. I'm sure it could be done, but that would be a good reason not to use it on animated models.
Unless I'm totally mistaken (something that can be true actually - my math are a bit rusty, and my internal GPU is kind of flaky too), I assumed that the normal map is a local perturbation of the surface normal at a particular point. Transforming the surface don't change the normal map, unless you are badly stretching the surface normals (which, admitedly, can happen when the animated mesh is not well designed but should not happen when using carefully designed meshes). I don't think that next-gen games recompute anything while playing an animation - it would be just too time consuming. Am I wrong?
-- Emmanuel D.
esaptonor
Posts: 145
Joined: Sat May 06, 2006 11:59 pm

Re: normal mapping on animated meshes

Post by esaptonor »

EmmanuelD wrote:
kburkhart84 wrote:
esaptonor wrote: i fully agree, It would allow some awesome next generation style games, and irrlicht needs some good looking stuff like this for its screenshots and tech demo. But is it hard to do all this?
That would be nice. I wonder though, due to the animation, wouldn't you need to calculate the binormals/tangents/whatever they are every frame of animation. I'm sure it could be done, but that would be a good reason not to use it on animated models.
Unless I'm totally mistaken (something that can be true actually - my math are a bit rusty, and my internal GPU is kind of flaky too), I assumed that the normal map is a local perturbation of the surface normal at a particular point. Transforming the surface don't change the normal map, unless you are badly stretching the surface normals (which, admitedly, can happen when the animated mesh is not well designed but should not happen when using carefully designed meshes). I don't think that next-gen games recompute anything while playing an animation - it would be just too time consuming. Am I wrong?
i don't know if this is true or not..and i am just a beginner, but how hard would it be to just hack up a new animated mesh type that supported a single tangent thing created at the start? I would really love for this to be in, its really a key feature -> your main characters have to look cool. Can someone offer advice or ideas on how to do it?
buhatkj
Posts: 444
Joined: Fri Dec 12, 2003 4:53 am
Contact:

Re: normal mapping on animated meshes

Post by buhatkj »

esaptonor wrote:
EmmanuelD wrote:
kburkhart84 wrote: That would be nice. I wonder though, due to the animation, wouldn't you need to calculate the binormals/tangents/whatever they are every frame of animation. I'm sure it could be done, but that would be a good reason not to use it on animated models.
Unless I'm totally mistaken (something that can be true actually - my math are a bit rusty, and my internal GPU is kind of flaky too), I assumed that the normal map is a local perturbation of the surface normal at a particular point. Transforming the surface don't change the normal map, unless you are badly stretching the surface normals (which, admitedly, can happen when the animated mesh is not well designed but should not happen when using carefully designed meshes). I don't think that next-gen games recompute anything while playing an animation - it would be just too time consuming. Am I wrong?
i don't know if this is true or not..and i am just a beginner, but how hard would it be to just hack up a new animated mesh type that supported a single tangent thing created at the start? I would really love for this to be in, its really a key feature -> your main characters have to look cool. Can someone offer advice or ideas on how to do it?
Well I doubt it would be necessary to precompute the normals for each frame. It should be possible to transform the normals the same way that the vertices are transformed and interpolated for animation. although, all those transforms might be really expensive too...
My irrlicht-based projects have gone underground for now, but if you want, check out my webcomic instead! http://brokenboomerang.net
hybrid
Admin
Posts: 14143
Joined: Wed Apr 19, 2006 9:20 pm
Location: Oldenburg(Oldb), Germany
Contact:

Post by hybrid »

Normals are of course already transformed. It was said that binormals and tangents cannot be transformed that easily. I neither know why you have to use binormals and tangents nor why they cannot be transformed that easy. Otherwise it would be rather simply (just another transformVect(binormal)).
EmmanuelD
Posts: 25
Joined: Thu Aug 24, 2006 8:34 am
Location: Right in front of Dagoth Ur
Contact:

Post by EmmanuelD »

hybrid wrote:Normals are of course already transformed. It was said that binormals and tangents cannot be transformed that easily. I neither know why you have to use binormals and tangents nor why they cannot be transformed that easy. Otherwise it would be rather simply (just another transformVect(binormal)).
I think there's some kind of misconceptions about what normal or tangent maps are. For those who have some problem to understand them, wikipedia has some correct introductory articles.

I also don't understand why someone would use binormals and tangents in the rendering process. Did I miss something obvious?
-- Emmanuel D.
hybrid
Admin
Posts: 14143
Joined: Wed Apr 19, 2006 9:20 pm
Location: Oldenburg(Oldb), Germany
Contact:

Post by hybrid »

Binormals and Tangents are required (per vertex) for the current normal and parallax shaders because these calculate their transformations in tangent space. The results without those vectors are not really convincing (and calculating those vectors every render cycle is too time consuming).
But since all three (normals, binormals, and tangents) are vectors and the matrix transformations used in animations are usually not scaling vectors they should all be transformable.
This is at least my understanding of the whole thing, though not being into computer graphics for too long :wink:
EmmanuelD
Posts: 25
Joined: Thu Aug 24, 2006 8:34 am
Location: Right in front of Dagoth Ur
Contact:

Post by EmmanuelD »

hybrid wrote:Binormals and Tangents are required (per vertex) for the current normal and parallax shaders because these calculate their transformations in tangent space. The results without those vectors are not really convincing (and calculating those vectors every render cycle is too time consuming).
But since all three (normals, binormals, and tangents) are vectors and the matrix transformations used in animations are usually not scaling vectors they should all be transformable.
This is at least my understanding of the whole thing, though not being into computer graphics for too long :wink:
Doo! Forgot about these! :oops:
-- Emmanuel D.
juliusctw
Posts: 392
Joined: Fri Apr 21, 2006 6:56 am
Contact:

fly straight animator

Post by juliusctw »

I would like the fly straight animator to add another method called

resetAnimator( same as parameter of constructor )
recalculateImidiateValues();


with this we could now create fire balls that track a target as it flys and it is only 2 lines to add .
:D
Spintz
Posts: 1688
Joined: Thu Nov 04, 2004 3:25 pm

Post by Spintz »

The fly straight animator is actually overkill for what you want to do. Look into dead reckoning type algorithms, and you'll find much simpler stuff.
Image
3ddev
Posts: 169
Joined: Tue Sep 19, 2006 6:51 am
Contact:

Better 2d support needed!

Post by 3ddev »

I think Irrlicht should have some more 2d support. Currently, the 2d drawing is so slow as well as basic. I am not suggesting that Irrlicht becomes a 2d engine; but things like gif animation support, pixel drawing(advanced), different surfaces(like sdl), resizing, color palettes, image animators and most of all, image collision detectors would be great. :lol:
Post Reply