- 27 11月, 2012 3 次提交
-
-
由 alteredq 提交于
These replace frontFaceDirection parameter strings in renderer.setFaceCulling( cullFace, frontFaceDirection ): "cw" => THREE.FrontFaceDirectionCW "ccw" => THREE.FrontFaceDirectionCCW See related cullFace constants changes in #2683
-
由 alteredq 提交于
Default value THREE.CullFaceFront, other valid value THREE.CullFaceBack. These constants also go into renderer.setFaceCulling( cullFace, frontFace ) where they replace strings: "back" => THREE.CullFaceBack "front" => THREE.CullFaceFront "frontback" => THREE.CullFaceFrontBack ""/0/undefined => THREE.CullFaceNone See discussion in #2683
-
由 alteredq 提交于
-
- 26 11月, 2012 12 次提交
-
-
由 alteredq 提交于
Default value is THREE.BackSide (use backfaces of objects when rendering depth map), other valid value is THREE.FrontSide (use frontfaces of objects when rendering depth map). See discussion in #2683
-
由 alteredq 提交于
Specular parameters will eventually need to go into G-buffer.
-
由 alteredq 提交于
Started with extraction of deferred shading code snippets into more reusable components.
-
由 Mr.doob 提交于
-
由 Mr.doob 提交于
-
由 alteredq 提交于
Before base class attributes that weren't in the constructors were not cloned properly in derived classes.
-
由 alteredq 提交于
This is needed for handling manually created non-power-of-2 or small RGB DataTextures. It corresponds to GL_UNPACK_ALIGNMENT parameter in glPixelStorei: http://www.khronos.org/opengles/sdk/docs/man/xhtml/glPixelStorei.xml
-
由 alteredq 提交于
Warning: this doesn't work with RGB textures (3 bytes per pixel) because of default 4 byte alignment requirement deeper in WebGL / OpenGL layer. Uploading smaller mipmap levels gets these GL errors: "ArrayBufferView not big enough for request with UNPACK_ALIGNMENT > 1" It works ok with RGBA textures (4 bytes per pixel). Seems this could also explain why 1 pixel RGB DataTextures were failing sometimes.
-
由 alteredq 提交于
-
由 alteredq 提交于
See #2681
-
由 alteredq 提交于
-
由 alteredq 提交于
Fixes #2683
-
- 25 11月, 2012 2 次提交
-
-
由 Ben Adams 提交于
-
由 zz85 提交于
THREE.Path THREE.Shape THREE.CurvePath THREE.Curve THREE.SubdivisionModifier THREE.ExtrudeGeometry @sole you might be interested in the script i used to help generate this. p = new THREE.Path() for (i in p) { p.hasOwnProperty(i) && console.log('\n<h3>.' + i + '</h3>'); } console.log('-------------') for (i in p) { var pp = p[i].toString().replace(/function/,''); THREE.Path.prototype.hasOwnProperty(i) && console.log('\n<h3>.' + i + pp.substring(0, pp.indexOf('{')-1).replace(/\s+/g, ' ') + '</h3>\n<div>todo</div>' ); }
-
- 24 11月, 2012 9 次提交
- 23 11月, 2012 4 次提交
-
-
由 zz85 提交于
-
-
由 MiiBond 提交于
Right now, if the DDS data is a cubemap, the code is hardcoded to load 6 faces. This could be made more robust by checking DDSCAPS2_CUBEMAP_POSITIVEX, etc. to verify that all 6 faces exist before reading the data from the buffer. I also did a quick update of the webgl_materials_texture_compressed example to load a cube DDS.
-
由 alteredq 提交于
Now morph buffers behave the same like other buffers and do not attempt to get set when materials do not use corresponding attributes. Not having this was crashing deferred rendering when shuffling around materials in G-buffer passes.
-
- 22 11月, 2012 10 次提交
-
-
由 Mr.doob 提交于
-
由 gero3 提交于
fixes #2666
-
由 alteredq 提交于
Gained more performance, used it for more bling.
-
由 Mr.doob 提交于
-
由 alteredq 提交于
Up to 95 fps.
-
由 alteredq 提交于
Small performance improvement (88 fps => 91 fps).
-
由 alteredq 提交于
Much faster with "discard" (56 fps => 88 fps).
-
由 alteredq 提交于
Also fixed light colors to be in linear color space to match the rest of the pipeline.
-
由 alteredq 提交于
-
由 alteredq 提交于
- changed model to Lee Perry - changed light pass material to use RGB-weighted wrap-around lighting (with deferred can't have easily together multiple material types :S) - changed deferred rendering to use linear space lighting
-