Re: Problem with normalmap
res <[email protected]> Sun, 28 Jul 2013 15:06:58 +0200
| Newsgroups | gmane.comp.graphics.crystalspace.devel |
|---|---|
| Message-ID | <[email protected]> |
On 27.07.2013 01:49, Naman Gupta wrote: > res, was CS/data/shader-snippet/tcoffset2-perturb-normal.xml left > incomplete? > was it suppose to pass "z killed normal" in simple float3 ? I don't think so. Z is zeroed out in the tcoffset2-perturb-normal-cam snippet because a transformation from tangent to camera space will take place; with a non-zero Z this could result in unwanted (I think) extra distortions when the camera is tilted. > Phew, I finally figured why the normal maps in weaver was showing > such weird results. > > It's because normal,tangent and bitangents are NOT provided by the > watermesh plugin and I failed to see in the "outer" code as it is > triggered by coercion. > > The way I see it to work around with this problem is to alter > tcoffset snippet. or maybe... > Well, or you provide tangents and bitangents. Though that probably means getting them must move out of the coercion library… the position snippet looks like an appropriate place. There's some precedent for a value generated by the position snippet but consumed by coercion snippet; specifically, it's the object-to-world transform. For general meshes, it's passed in a shader var, but for instanced meshes it has to be computed per instance. So, the position snippet stores that transform in a global variable, and the coercion snippets needing object2world take it from there – and it doesn't matter whether the mesh is instanced or not. If the same was done for tangent & bitangent, you could write a “water” position snippet that computes them in the shader. -f.r. ------------------------------------------------------------------------------ See everything from the browser to the database with AppDynamics Get end-to-end visibility with application monitoring from AppDynamics Isolate bottlenecks and diagnose root cause in seconds. Start your free trial of AppDynamics Pro today! http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk _______________________________________________ Crystal-develop mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/crystal-develop
signature.asc
(application/pgp-signature, 263 B)
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.13 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlH1F3YACgkQNCTzcE/07exejQCfVAXK/yASKoCmlq2+aJsg9a1n wlEAniMZ1XQ9diI4tWQBrs14OIRD1NQT =f2At -----END PGP SIGNATURE-----