![]() |
![]() ![]() ![]() ![]() ![]()
|
![]() |
next newest topic | next oldest topic |
Author | Topic: free but lousy way of doing a normal map... but its a nifty way to unwrap! |
KingofDaveness![]() Posts: 218 |
![]() ![]() ![]() ![]() ![]() If you guys are skint theres a little trick to creating your bump map thats not pleasant, but it IS free. 1: build your low poly model. Ok now the clever bit. 7: Apply a tesselate to the morph, then paint a displacement map to put
in Like I said, its not ideal, but it gets you through if you want to spend 0$. IP: Logged |
tpe Model ![]() Freak Posts: 2620 |
![]() ![]() ![]() ![]() ![]() Wow thats great, reminds me of the morph UV mapping technique. If you are into lightwave then check this out, a free script that makes
the standard lightwave do what polybump does (make normal bump maps) not only that it was done on request by Arnie Cachelin a member of lightwave support for one of their stinky snotty end users (me). For some reason i cant link the original URL so here is the script To use it, apply the shader to the relevant surfaces, set diffuse to 0
and luminous to 100%, and add baker.
@version 2.3 color; create init flags process: sa
IP: Logged |
KingofDaveness![]() Posts: 218 |
![]() ![]() ![]() ![]() ![]() Ohhh, I just got another one! You can render the max file using colored lights to get the normal map and it looks better than heightmap conversions. Oooooh, those pretty colors. You set max values thus... Ambient 128,128,128 Render that baby and voila.... it works with the cheap version above too... Damn. its full of stars..... IP: Logged |
Astro Zombie Sick Boy ![]() Posts: 1791 |
![]() ![]() ![]() ![]() ![]() Can I request that you do a tut complete with pictures showing the process and the end result? This sounds cool but I'd like to see it in action before I try it. IP: Logged |
Whargoul Electronic Arts Lead Character Artist ![]() Posts: 567 |
![]() ![]() ![]() ![]() ![]() The problem with both of those techniques is that the normals are encoded in world-space, not local space to each face (usually called "basis-space" I think, since it's based on the S-T vectors for each face that I believe NVidia came up with for transforming lights into texture space). So if you took your model and deformed it with bones, the normals won't
transform with it and it will look messed up. I've been looking into this
and that's why I know quite a bit about it IP: Logged |
The_SP PolY SlAv3 ![]() Posts: 132 |
![]() ![]() ![]() ![]() ![]() Yummy yummy Mel in my tummy ![]() ![]() Cheers, ThE SP IP: Logged |
tpe Model ![]() Freak Posts: 2620 |
![]() ![]() ![]() ![]() ![]() Whargoul the Lightwave one is relative to the normal of the poly. Because of the way baker works it is a reasonaly trivial task, and a shame to pay so much for so little. Do you mean your task will work out the normal of the high poly on the low poly? If not you will still have problems when changing mesh res (progressivly worse with LODs)? tpe IP: Logged |
KingofDaveness![]() Posts: 218 |
![]() ![]() ![]() ![]() ![]() Er, yeah Warghoul.... Whatever you say dude... Erm. Ok, look at it this way, its a morph map which is displaced and rendered. Nothing to do with bones. Therefore world space or not, it works fine. IP: Logged |
Nahaz Junior Member Posts: 9 |
![]() ![]() ![]() ![]() ![]() Whargoul, the first technique encodes the normals in UVW space. I have tried it on a deforming mesh in a pixel shader enabled app. The second technique seems to do as you say, but obviously CryTek must have worked away around that? IP: Logged |
tpe Model ![]() Freak Posts: 2620 |
![]() ![]() ![]() ![]() ![]() One un-avoidable problem with the crytek implmentation is that it projects pixel data from the high res model onto the nearest position on the low res uv, therefore there is a slight distortion proportional to the tangent of the plane of the high res with respect to that of the low res that will give some small errors. these errors will be present in all resolutions but will be greater on any lods that are no proximity calculations are applied. I think the easiest way round it is to go in a d edit the uvs for each model, althoug how prominant it will be with the LODs i dont know, hopefully they will be too far away to see the errors. Whatever, the Lightwave version (like the other two) does not take the res of the low poly model into account at all, all corrections have to be done manually. Incedently i tryed putting the LW version and the crytek version side
to side, and the major and only differences i could see with my eye were
the presence of the watermark and the lack of resolution in the crytek,
but i guess that is because i dont want to pay for the crytek version tpe tpe [This message has been edited by tpe (edited 07-29-2002).] IP: Logged |
Whargoul Electronic Arts Lead Character Artist ![]() Posts: 567 |
![]() ![]() ![]() ![]() ![]() For both of those techniques, you need to have matching UV's across both objects right? IP: Logged |
Nahaz Junior Member Posts: 9 |
![]() ![]() ![]() ![]() ![]() Don't have to match UVs with CryTek's plugin. IP: Logged |
KingofDaveness![]() Posts: 218 |
![]() ![]() ![]() ![]() ![]() Dammit, wheres Id when you need them ![]() IP: Logged |
Nahaz Junior Member Posts: 9 |
![]() ![]() ![]() ![]() ![]() Just dawned on me that id Software must be using LightWave to capture their normals. I remember seeing them using LightWave in that recent doom3 video. IP: Logged |
tpe Model ![]() Freak Posts: 2620 |
![]() ![]() ![]() ![]() ![]() I have been using the controll mesh to map the uvs on my high res version, this way both sets of UVs are identicle for both resolution meshes so matching UVs is not needed. However the low poly mesh does need tweeking after to fit the high poly well, this would be necessary anyway whatever package is used. When using lightwave (I think the modeling app used at id, not that it makes much difference as file formats are easier to convert for models than scenes) it is very easy to make the controll mesh UV map it and then apply a subpatch modifyer so the UV matching is not really an issue. If one were not using UV colour/bump/specularity maps on the high poly version and only procedural textures it would be an advantage not to have identicle uvs as atomatic mapping could be used on the high poly version, otherwise it does not take any longer or require any work to make both meshes have the same UVs.
tpe IP: Logged |
All times are PT (US) |
next newest topic | next oldest topic |
![]() ![]() |
i t ' s a c o n s t a n t c o s t u m e p a r t y