Sculpt Branch:

svn merge https://svn.blender.org/svnroot/bf-blender/trunk/blender -r24889:25180

13 files changed, 283 insertions, 171 deletions

diff --git a/source/blender/render/SConscript b/source/blender/render/SConscript
index 63ef83a0cfe..2642d31bd35 100644
--- a/source/blender/render/SConscript
+++ b/source/blender/render/SConscript
@@ -1,18 +1,21 @@
 #!/usr/bin/python
 Import ('env')
 
-if env['OURPLATFORM'] in ('win32-vc', 'win64-vc', 'win32-mingw'):
-	# FIXME: need to set the appropriate flags for msvc, otherwise we get warnings
-	cflags = []
-	cxxflags = []
+cflags = env['CCFLAGS']
+cxxflags = env['CXXFLAGS']
+
+if env['OURPLATFORM'] in ('win32-vc', 'win64-vc'):
+    if env['WITH_BF_RAYOPTIMIZATION']:
+        cflags = env['CCFLAGS'] + ['/arch:SSE']
+
+if env['OURPLATFORM'] == 'win32-mingw':
+    if env['WITH_BF_RAYOPTIMIZATION']:
+        cflags = env['CCFLAGS'] + ['-mfpmath=sse']
 
 if env['OURPLATFORM'] == 'darwin':
-	if env['MACOSX_ARCHITECTURE'] in ('i386', 'x86_64'):
-		cflags = env['CFLAGS'] + ['-mfpmath=sse']
-		cxxflags = env['CXXFLAGS'] + ['-mfpmath=sse']
-	else:
-		cflags = env['CFLAGS']
-		cxxflags = env['CXXFLAGS']
+    if env['MACOSX_ARCHITECTURE'] in ('i386', 'x86_64') and env['WITH_BF_RAYOPTIMIZATION']:
+        cflags = env['CFLAGS'] + ['-mfpmath=sse']
+        cxxflags = env['CXXFLAGS'] + ['-mfpmath=sse']
 
 sources = env.Glob('intern/source/*.c')
 raysources = env.Glob('intern/raytrace/*.cpp')
@@ -31,15 +34,19 @@ if env['WITH_BF_OPENEXR']:
     defs.append('WITH_OPENEXR')
 
 if env['OURPLATFORM'] == 'linux2':
-#	SSE is NOT safe all the time on linux, plus that ignores users compile flags and therefore no no
+#    SSE is NOT safe all the time on linux, plus that ignores users compile flags and therefore no no
 #    cflags = ['-O2','-msse2','-mfpmath=sse', '-pthread']
 #    cxxflags = ['-O2','-msse2','-mfpmath=sse', '-pthread']
-    cflags = env['CCFLAGS']
-    cxxflags = env['CXXFLAGS']
+    if env['WITH_BF_RAYOPTIMIZATION']:
+        cflags = env['CCFLAGS'] +  ['-O2','-msse2','-mfpmath=sse', '-pthread']
+        cxxflags = env['CXXFLAGS'] +  ['-O2','-msse2','-mfpmath=sse', '-pthread']
     incs += ' ../../../extern/binreloc/include'
 
 if env['OURPLATFORM'] in ('win32-vc', 'win32-mingw', 'linuxcross', 'win64-vc'):
     incs += ' ' + env['BF_PTHREADS_INC']
 
+if env['WITH_BF_RAYOPTIMIZATION']:
+    defs.append('__SSE__')
+
 env.BlenderLib ( libname = 'bf_render', sources = sources, includes = Split(incs), defines=defs, libtype='core', priority=145, compileflags=cflags )
 env.BlenderLib ( libname = 'bf_render_raytrace', sources = raysources, includes = Split(incs), defines=defs, libtype='core', priority=145, compileflags=cflags, cxx_compileflags=cxxflags )
diff --git a/source/blender/render/extern/include/RE_shader_ext.h b/source/blender/render/extern/include/RE_shader_ext.h
index b36163f57c0..2615be1440a 100644
--- a/source/blender/render/extern/include/RE_shader_ext.h
+++ b/source/blender/render/extern/include/RE_shader_ext.h
@@ -153,7 +153,7 @@ typedef struct ShadeInput
 	float dxstrand, dystrand;
 	
 	/* AO is a pre-process now */
-	float ao[3];
+	float ao[3], indirect[3];
 	
 	int xs, ys;				/* pixel to be rendered */
 	int mask;				/* subsample mask */
diff --git a/source/blender/render/intern/include/render_types.h b/source/blender/render/intern/include/render_types.h
index 48bf34d0696..d41851db5ff 100644
--- a/source/blender/render/intern/include/render_types.h
+++ b/source/blender/render/intern/include/render_types.h
@@ -398,7 +398,8 @@ typedef struct StrandSurface {
 	int (*face)[4];
 	float (*co)[3];
 	/* for occlusion caching */
-	float (*col)[3];
+	float (*ao)[3];
+	float (*indirect)[3];
 	/* for speedvectors */
 	float (*prevco)[3], (*nextco)[3];
 	int totvert, totface;
diff --git a/source/blender/render/intern/raytrace/reorganize.h b/source/blender/render/intern/raytrace/reorganize.h
index bcf7f6f2078..2d4a4d15fe2 100644
--- a/source/blender/render/intern/raytrace/reorganize.h
+++ b/source/blender/render/intern/raytrace/reorganize.h
@@ -26,13 +26,16 @@
  *
  * ***** END GPL LICENSE BLOCK *****
  */
-
 #include <stdio.h>
 #include <algorithm>
 #include <math.h>
 #include <vector>
 #include <queue>
 
+#ifdef _WIN32
+#define INFINITY FLT_MAX // in mingw math.h: (1.0F/0.0F). This generates compile error, though.
+#endif
+
 extern int tot_pushup;
 extern int tot_pushdown;
 
diff --git a/source/blender/render/intern/source/convertblender.c b/source/blender/render/intern/source/convertblender.c
index d09c6a48ef3..d0f7b2d60a5 100644
--- a/source/blender/render/intern/source/convertblender.c
+++ b/source/blender/render/intern/source/convertblender.c
@@ -4496,7 +4496,7 @@ void RE_Database_Free(Render *re)
 	}
 }
 
-static int allow_render_object(Object *ob, int nolamps, int onlyselected, Object *actob)
+static int allow_render_object(Render *re, Object *ob, int nolamps, int onlyselected, Object *actob)
 {
 	/* override not showing object when duplis are used with particles */
 	if(ob->transflag & OB_DUPLIPARTS)
@@ -4504,6 +4504,10 @@ static int allow_render_object(Object *ob, int nolamps, int onlyselected, Object
 	else if((ob->transflag & OB_DUPLI) && !(ob->transflag & OB_DUPLIFRAMES))
 		return 0;
 	
+	/* don't add non-basic meta objects, ends up having renderobjects with no geometry */
+	if (ob->type == OB_MBALL && ob!=find_basis_mball(re->scene, ob))
+		return 0;
+	
 	if(nolamps && (ob->type==OB_LAMP))
 		return 0;
 	
@@ -4609,7 +4613,7 @@ static void add_group_render_dupli_obs(Render *re, Group *group, int nolamps, in
 
 		if(ob->flag & OB_DONE) {
 			if(ob->transflag & OB_RENDER_DUPLI) {
-				if(allow_render_object(ob, nolamps, onlyselected, actob)) {
+				if(allow_render_object(re, ob, nolamps, onlyselected, actob)) {
 					init_render_object(re, ob, NULL, 0, timeoffset, vectorlay);
 					ob->transflag &= ~OB_RENDER_DUPLI;
 
@@ -4663,7 +4667,7 @@ static void database_init_objects(Render *re, unsigned int renderlay, int nolamp
 			/* OB_RENDER_DUPLI means instances for it were already created, now
 			 * it still needs to create the ObjectRen containing the data */
 			if(ob->transflag & OB_RENDER_DUPLI) {
-				if(allow_render_object(ob, nolamps, onlyselected, actob)) {
+				if(allow_render_object(re, ob, nolamps, onlyselected, actob)) {
 					init_render_object(re, ob, NULL, 0, timeoffset, vectorlay);
 					ob->transflag &= ~OB_RENDER_DUPLI;
 				}
@@ -4697,7 +4701,7 @@ static void database_init_objects(Render *re, unsigned int renderlay, int nolamp
 					if(obd->type==OB_MBALL)
 						continue;
 
-					if(!allow_render_object(obd, nolamps, onlyselected, actob))
+					if(!allow_render_object(re, obd, nolamps, onlyselected, actob))
 						continue;
 
 					if(allow_render_dupli_instance(re, dob, obd)) {
@@ -4767,10 +4771,10 @@ static void database_init_objects(Render *re, unsigned int renderlay, int nolamp
 				}
 				free_object_duplilist(lb);
 
-				if(allow_render_object(ob, nolamps, onlyselected, actob))
+				if(allow_render_object(re, ob, nolamps, onlyselected, actob))
 					init_render_object(re, ob, NULL, 0, timeoffset, vectorlay);
 			}
-			else if(allow_render_object(ob, nolamps, onlyselected, actob))
+			else if(allow_render_object(re, ob, nolamps, onlyselected, actob))
 				init_render_object(re, ob, NULL, 0, timeoffset, vectorlay);
 		}
 
@@ -4848,8 +4852,7 @@ void RE_Database_FromScene(Render *re, Scene *scene, int use_camera_view)
 	
 	/* still bad... doing all */
 	init_render_textures(re);
-	if (re->r.color_mgt_flag & R_COLOR_MANAGEMENT) color_manage_linearize(amb, &re->wrld.ambr);
-	else VECCOPY(amb, &re->wrld.ambr);
+	VECCOPY(amb, &re->wrld.ambr);
 	init_render_materials(re->r.mode, amb);
 	set_node_shader_lamp_loop(shade_material_loop);
 
@@ -5538,8 +5541,7 @@ void RE_Database_Baking(Render *re, Scene *scene, int type, Object *actob)
 	/* still bad... doing all */
 	init_render_textures(re);
 	
-	if (re->r.color_mgt_flag & R_COLOR_MANAGEMENT) color_manage_linearize(amb, &re->wrld.ambr);
-	else VECCOPY(amb, &re->wrld.ambr);
+	VECCOPY(amb, &re->wrld.ambr);
 	init_render_materials(re->r.mode, amb);
 	
 	set_node_shader_lamp_loop(shade_material_loop);
diff --git a/source/blender/render/intern/source/occlusion.c b/source/blender/render/intern/source/occlusion.c
index f62668b84c0..3a5680eaf70 100644
--- a/source/blender/render/intern/source/occlusion.c
+++ b/source/blender/render/intern/source/occlusion.c
@@ -64,7 +64,7 @@
 #define CACHE_STEP 3
 
 typedef struct OcclusionCacheSample {
-	float co[3], n[3], col[3], intensity, dist2;
+	float co[3], n[3], ao[3], indirect[3], intensity, dist2;
 	int x, y, filled;
 } OcclusionCacheSample;
 
@@ -81,7 +81,7 @@ typedef struct OccFace {
 typedef struct OccNode {
 	float co[3], area;
 	float sh[9], dco;
-	float occlusion;
+	float occlusion, rad[3];
 	int childflag;
 	union {
 		//OccFace face;
@@ -97,6 +97,7 @@ typedef struct OcclusionTree {
 
 	OccFace *face;		/* instance and face indices */
 	float *occlusion;	/* occlusion for faces */
+	float (*rad)[3];	/* radiance for faces */
 	
 	OccNode *root;
 
@@ -110,6 +111,7 @@ typedef struct OcclusionTree {
 
 	int dothreadedbuild;
 	int totbuildthread;
+	int doindirect;
 
 	OcclusionCache *cache;
 } OcclusionTree;
@@ -117,7 +119,8 @@ typedef struct OcclusionTree {
 typedef struct OcclusionThread {
 	Render *re;
 	StrandSurface *mesh;
-	float (*facecol)[3];
+	float (*faceao)[3];
+	float (*faceindirect)[3];
 	int begin, end;
 	int thread;
 } OcclusionThread;
@@ -132,7 +135,6 @@ typedef struct OcclusionBuildThread {
 
 extern Render R; // meh
 
-#if 0
 static void occ_shade(ShadeSample *ssamp, ObjectInstanceRen *obi, VlakRen *vlr, float *rad)
 {
 	ShadeInput *shi= ssamp->shi;
@@ -160,7 +162,7 @@ static void occ_shade(ShadeSample *ssamp, ObjectInstanceRen *obi, VlakRen *vlr,
 	shi->co[0]= l*v3[0]+u*v1[0]+v*v2[0];
 	shi->co[1]= l*v3[1]+u*v1[1]+v*v2[1];
 	shi->co[2]= l*v3[2]+u*v1[2]+v*v2[2];
-	
+
 	shade_input_set_triangle_i(shi, obi, vlr, 0, 1, 2);
 
 	/* set up view vector */
@@ -179,6 +181,8 @@ static void occ_shade(ShadeSample *ssamp, ObjectInstanceRen *obi, VlakRen *vlr,
 	if(shi->flippednor)
 		shade_input_flip_normals(shi);
 
+	madd_v3_v3fl(shi->co, shi->vn, 0.0001f); /* ugly.. */
+
 	/* not a pretty solution, but fixes common cases */
 	if(shi->obr->ob && shi->obr->ob->transflag & OB_NEG_SCALE) {
 		negate_v3(shi->vn);
@@ -215,12 +219,11 @@ static void occ_build_shade(Render *re, OcclusionTree *tree)
 
 	for(a=0; a<tree->totface; a++) {
 		obi= &R.objectinstance[tree->face[a].obi];
-		vlr= RE_findOrAddVlak(obi->obr, tree->face[a].vlr);
+		vlr= RE_findOrAddVlak(obi->obr, tree->face[a].facenr);
 
 		occ_shade(&ssamp, obi, vlr, tree->rad[a]);
 	}
 }
-#endif
 
 /* ------------------------- Spherical Harmonics --------------------------- */
 
@@ -352,17 +355,19 @@ static void occ_face(const OccFace *face, float *co, float *normal, float *area)
 static void occ_sum_occlusion(OcclusionTree *tree, OccNode *node)
 {
 	OccNode *child;
-	float occ, area, totarea;
+	float occ, area, totarea, rad[3];
 	int a, b;
 
 	occ= 0.0f;
 	totarea= 0.0f;
+	zero_v3(rad);
 
 	for(b=0; b<TOTCHILD; b++) {
 		if(node->childflag & (1<<b)) {
 			a= node->child[b].face;
 			occ_face(&tree->face[a], 0, 0, &area);
 			occ += area*tree->occlusion[a];
+			madd_v3_v3fl(rad, tree->rad[a], area);
 			totarea += area;
 		}
 		else if(node->child[b].node) {
@@ -370,14 +375,18 @@ static void occ_sum_occlusion(OcclusionTree *tree, OccNode *node)
 			occ_sum_occlusion(tree, child);
 
 			occ += child->area*child->occlusion;
+			madd_v3_v3fl(rad, child->rad, child->area);
 			totarea += child->area;
 		}
 	}
 
-	if(totarea != 0.0f)
+	if(totarea != 0.0f) {
 		occ /= totarea;
+		mul_v3_fl(rad, 1.0f/totarea);
+	}
 	
 	node->occlusion= occ;
+	copy_v3_v3(node->rad, rad);
 }
 
 static int occ_find_bbox_axis(OcclusionTree *tree, int begin, int end, float *min, float *max)
@@ -644,6 +653,7 @@ static OcclusionTree *occ_tree_build(Render *re)
 	/* parameters */
 	tree->error= get_render_aosss_error(&re->r, re->wrld.ao_approx_error);
 	tree->distfac= (re->wrld.aomode & WO_AODIST)? re->wrld.aodistfac: 0.0f;
+	tree->doindirect= (re->wrld.ao_indirect_energy > 0.0f && re->wrld.ao_indirect_bounces > 0);
 
 	/* allocation */
 	tree->arena= BLI_memarena_new(0x8000 * sizeof(OccNode));
@@ -656,6 +666,9 @@ static OcclusionTree *occ_tree_build(Render *re)
 	tree->co= MEM_callocN(sizeof(float)*3*totface, "OcclusionCo");
 	tree->occlusion= MEM_callocN(sizeof(float)*totface, "OcclusionOcclusion");
 
+	if(tree->doindirect)
+		tree->rad= MEM_callocN(sizeof(float)*3*totface, "OcclusionRad");
+
 	/* make array of face pointers */
 	for(b=0, c=0, obi=re->instancetable.first; obi; obi=obi->next, c++) {
 		obr= obi->obr;
@@ -682,12 +695,10 @@ static OcclusionTree *occ_tree_build(Render *re)
 	tree->maxdepth= 1;
 	occ_build_recursive(tree, tree->root, 0, totface, 1);
 
-#if 0
 	if(tree->doindirect) {
 		occ_build_shade(re, tree);
 		occ_sum_occlusion(tree, tree->root);
 	}
-#endif
 	
 	MEM_freeN(tree->co);
 	tree->co= NULL;
@@ -710,8 +721,9 @@ static void occ_free_tree(OcclusionTree *tree)
 			if(tree->stack[a])
 				MEM_freeN(tree->stack[a]);
 		if(tree->occlusion) MEM_freeN(tree->occlusion);
-		if(tree->face) MEM_freeN(tree->face);
 		if(tree->cache) MEM_freeN(tree->cache);
+		if(tree->face) MEM_freeN(tree->face);
+		if(tree->rad) MEM_freeN(tree->rad);
 		MEM_freeN(tree);
 	}
 }
@@ -1171,13 +1183,13 @@ static float occ_form_factor(OccFace *face, float *p, float *n)
 	return contrib;
 }
 
-static void occ_lookup(OcclusionTree *tree, int thread, OccFace *exclude, float *pp, float *pn, float *occ, float *bentn)
+static void occ_lookup(OcclusionTree *tree, int thread, OccFace *exclude, float *pp, float *pn, float *occ, float rad[3], float bentn[3])
 {
 	OccNode *node, **stack;
 	OccFace *face;
-	float resultocc, v[3], p[3], n[3], co[3], invd2;
+	float resultocc, resultrad[3], v[3], p[3], n[3], co[3], invd2;
 	float distfac, fac, error, d2, weight, emitarea;
-	int b, totstack;
+	int b, f, totstack;
 
 	/* init variables */
 	VECCOPY(p, pp);
@@ -1185,12 +1197,13 @@ static void occ_lookup(OcclusionTree *tree, int thread, OccFace *exclude, float
 	VECADDFAC(p, p, n, 1e-4f);
 
 	if(bentn)
-		VECCOPY(bentn, n);
+		copy_v3_v3(bentn, n);
 	
 	error= tree->error;
 	distfac= tree->distfac;
 
 	resultocc= 0.0f;
+	zero_v3(resultrad);
 
 	/* init stack */
 	stack= tree->stack[thread];
@@ -1217,6 +1230,10 @@ static void occ_lookup(OcclusionTree *tree, int thread, OccFace *exclude, float
 			/* accumulate occlusion from spherical harmonics */
 			invd2 = 1.0f/sqrtf(d2);
 			weight= occ_solid_angle(node, v, d2, invd2, n);
+
+			if(rad)
+				madd_v3_v3fl(resultrad, node->rad, weight*fac);
+
 			weight *= node->occlusion;
 
 			if(bentn) {
@@ -1231,7 +1248,8 @@ static void occ_lookup(OcclusionTree *tree, int thread, OccFace *exclude, float
 			/* traverse into children */
 			for(b=0; b<TOTCHILD; b++) {
 				if(node->childflag & (1<<b)) {
-					face= tree->face+node->child[b].face;
+					f= node->child[b].face;
+					face= &tree->face[f];
 
 					/* accumulate occlusion with face form factor */
 					if(!exclude || !(face->obi == exclude->obi && face->facenr == exclude->facenr)) {
@@ -1248,7 +1266,11 @@ static void occ_lookup(OcclusionTree *tree, int thread, OccFace *exclude, float
 							fac= 1.0f;
 
 						weight= occ_form_factor(face, p, n);
-						weight *= tree->occlusion[node->child[b].face];
+
+						if(rad)
+							madd_v3_v3fl(resultrad, tree->rad[f], weight*fac);
+
+						weight *= tree->occlusion[f];
 
 						if(bentn) {
 							invd2= 1.0f/sqrtf(d2);
@@ -1269,9 +1291,57 @@ static void occ_lookup(OcclusionTree *tree, int thread, OccFace *exclude, float
 	}
 
 	if(occ) *occ= resultocc;
+	if(rad) copy_v3_v3(rad, resultrad);
+	/*if(rad && exclude) {
+		int a;
+		for(a=0; a<tree->totface; a++)
+			if((tree->face[a].obi == exclude->obi && tree->face[a].facenr == exclude->facenr))
+				copy_v3_v3(rad, tree->rad[a]);
+	}*/
 	if(bentn) normalize_v3(bentn);
 }
 
+static void occ_compute_bounces(Render *re, OcclusionTree *tree, int totbounce)
+{
+	float (*rad)[3], (*sum)[3], (*tmp)[3], co[3], n[3], occ;
+	int bounce, i;
+
+	rad= MEM_callocN(sizeof(float)*3*tree->totface, "OcclusionBounceRad");
+	sum= MEM_dupallocN(tree->rad);
+
+	for(bounce=1; bounce<totbounce; bounce++) {
+		for(i=0; i<tree->totface; i++) {
+			occ_face(&tree->face[i], co, n, NULL);
+			madd_v3_v3fl(co, n, 1e-8f);
+
+			occ_lookup(tree, 0, &tree->face[i], co, n, &occ, rad[i], NULL);
+			rad[i][0]= MAX2(rad[i][0], 0.0f);
+			rad[i][1]= MAX2(rad[i][1], 0.0f);
+			rad[i][2]= MAX2(rad[i][2], 0.0f);
+			add_v3_v3(sum[i], rad[i]);
+
+			if(re->test_break(re->tbh))
+				break;
+		}
+
+		if(re->test_break(re->tbh))
+			break;
+
+		tmp= tree->rad;
+		tree->rad= rad;
+		rad= tmp;
+
+		occ_sum_occlusion(tree, tree->root);
+	}
+
+	MEM_freeN(rad);
+	MEM_freeN(tree->rad);
+	tree->rad= sum;
+
+	if(!re->test_break(re->tbh))
+		occ_sum_occlusion(tree, tree->root);
+}
+
 static void occ_compute_passes(Render *re, OcclusionTree *tree, int totpass)
 {
 	float *occ, co[3], n[3];
@@ -1285,7 +1355,7 @@ static void occ_compute_passes(Render *re, OcclusionTree *tree, int totpass)
 			negate_v3(n);
 			VECADDFAC(co, co, n, 1e-8f);
 
-			occ_lookup(tree, 0, &tree->face[i], co, n, &occ[i], NULL);
+			occ_lookup(tree, 0, &tree->face[i], co, n, &occ[i], NULL, NULL);
 			if(re->test_break(re->tbh))
 				break;
 		}
@@ -1305,9 +1375,9 @@ static void occ_compute_passes(Render *re, OcclusionTree *tree, int totpass)
 	MEM_freeN(occ);
 }
 
-static void sample_occ_tree(Render *re, OcclusionTree *tree, OccFace *exclude, float *co, float *n, int thread, int onlyshadow, float *skycol)
+static void sample_occ_tree(Render *re, OcclusionTree *tree, OccFace *exclude, float *co, float *n, int thread, int onlyshadow, float *ao, float *indirect)
 {
-	float nn[3], bn[3], fac, occ, occlusion, correction;
+	float nn[3], bn[3], fac, occ, occlusion, correction, rad[3];
 	int aocolor;
 
 	aocolor= re->wrld.aocolor;
@@ -1317,7 +1387,7 @@ static void sample_occ_tree(Render *re, OcclusionTree *tree, OccFace *exclude, f
 	VECCOPY(nn, n);
 	negate_v3(nn);
 
-	occ_lookup(tree, thread, exclude, co, nn, &occ, (aocolor)? bn: NULL);
+	occ_lookup(tree, thread, exclude, co, nn, &occ, (tree->doindirect)? rad: NULL, (aocolor)? bn: NULL);
 
 	correction= re->wrld.ao_approx_correction;
 
@@ -1330,9 +1400,9 @@ static void sample_occ_tree(Render *re, OcclusionTree *tree, OccFace *exclude, f
 		/* sky shading using bent normal */
 		if(ELEM(aocolor, WO_AOSKYCOL, WO_AOSKYTEX)) {
 			fac= 0.5*(1.0f+bn[0]*re->grvec[0]+ bn[1]*re->grvec[1]+ bn[2]*re->grvec[2]);
-			skycol[0]= (1.0f-fac)*re->wrld.horr + fac*re->wrld.zenr;
-			skycol[1]= (1.0f-fac)*re->wrld.horg + fac*re->wrld.zeng;
-			skycol[2]= (1.0f-fac)*re->wrld.horb + fac*re->wrld.zenb;
+			ao[0]= (1.0f-fac)*re->wrld.horr + fac*re->wrld.zenr;
+			ao[1]= (1.0f-fac)*re->wrld.horg + fac*re->wrld.zeng;
+			ao[2]= (1.0f-fac)*re->wrld.horb + fac*re->wrld.zenb;
 		}
 #if 0
 		else {	/* WO_AOSKYTEX */
@@ -1343,17 +1413,20 @@ static void sample_occ_tree(Render *re, OcclusionTree *tree, OccFace *exclude, f
 			dxyview[0]= 1.0f;
 			dxyview[1]= 1.0f;
 			dxyview[2]= 0.0f;
-			shadeSkyView(skycol, co, bn, dxyview);
+			shadeSkyView(ao, co, bn, dxyview);
 		}
 #endif
 
-		mul_v3_fl(skycol, occlusion);
+		mul_v3_fl(ao, occlusion);
 	}
 	else {
-		skycol[0]= occlusion;
-		skycol[1]= occlusion;
-		skycol[2]= occlusion;
+		ao[0]= occlusion;
+		ao[1]= occlusion;
+		ao[2]= occlusion;
 	}
+
+	if(tree->doindirect) copy_v3_v3(indirect, rad);
+	else zero_v3(indirect);
 }
 
 /* ---------------------------- Caching ------------------------------- */
@@ -1374,7 +1447,7 @@ static OcclusionCacheSample *find_occ_sample(OcclusionCache *cache, int x, int y
 		return &cache->sample[y*cache->w + x];
 }
 
-static int sample_occ_cache(OcclusionTree *tree, float *co, float *n, int x, int y, int thread, float *col)
+static int sample_occ_cache(OcclusionTree *tree, float *co, float *n, int x, int y, int thread, float *ao, float *indirect)
 {
 	OcclusionCache *cache;
 	OcclusionCacheSample *samples[4], *sample;
@@ -1394,7 +1467,8 @@ static int sample_occ_cache(OcclusionTree *tree, float *co, float *n, int x, int
 			VECSUB(d, sample->co, co);
 			dist2= INPR(d, d);
 			if(dist2 < 0.5f*sample->dist2 && INPR(sample->n, n) > 0.98f) {
-				VECCOPY(col, sample->col);
+				VECCOPY(ao, sample->ao);
+				VECCOPY(indirect, sample->indirect);
 				return 1;
 			}
 		}
@@ -1420,7 +1494,8 @@ static int sample_occ_cache(OcclusionTree *tree, float *co, float *n, int x, int
 		return 0;
 
 	/* compute weighted interpolation between samples */
-	col[0]= col[1]= col[2]= 0.0f;
+	zero_v3(ao);
+	zero_v3(indirect);
 	totw= 0.0f;
 
 	x1= samples[0]->x;
@@ -1446,16 +1521,14 @@ static int sample_occ_cache(OcclusionTree *tree, float *co, float *n, int x, int
 		w= wb[i]*wn[i]*wz[i];
 
 		totw += w;
-		col[0] += w*samples[i]->col[0];
-		col[1] += w*samples[i]->col[1];
-		col[2] += w*samples[i]->col[2];
+		madd_v3_v3fl(ao, samples[i]->ao, w);
+		madd_v3_v3fl(indirect, samples[i]->indirect, w);
 	}
 
 	if(totw >= 0.9f) {
 		totw= 1.0f/totw;
-		col[0] *= totw;
-		col[1] *= totw;
-		col[2] *= totw;
+		mul_v3_fl(ao, totw);
+		mul_v3_fl(indirect, totw);
 		return 1;
 	}
 
@@ -1469,7 +1542,7 @@ static void sample_occ_surface(ShadeInput *shi)
 	int *face, *index = RE_strandren_get_face(shi->obr, strand, 0);
 	float w[4], *co1, *co2, *co3, *co4;
 
-	if(mesh && mesh->face && mesh->co && mesh->col && index) {
+	if(mesh && mesh->face && mesh->co && mesh->ao && index) {
 		face= mesh->face[*index];
 
 		co1= mesh->co[face[0]];
@@ -1477,19 +1550,27 @@ static void sample_occ_surface(ShadeInput *shi)
 		co3= mesh->co[face[2]];
 		co4= (face[3])? mesh->co[face[3]]: NULL;
 
-		interp_weights_face_v3( w,co1, co2, co3, co4, strand->vert->co);
+		interp_weights_face_v3(w, co1, co2, co3, co4, strand->vert->co);
+
+		zero_v3(shi->ao);
+		zero_v3(shi->indirect);
 
-		shi->ao[0]= shi->ao[1]= shi->ao[2]= 0.0f;
-		VECADDFAC(shi->ao, shi->ao, mesh->col[face[0]], w[0]);
-		VECADDFAC(shi->ao, shi->ao, mesh->col[face[1]], w[1]);
-		VECADDFAC(shi->ao, shi->ao, mesh->col[face[2]], w[2]);
-		if(face[3])
-			VECADDFAC(shi->ao, shi->ao, mesh->col[face[3]], w[3]);
+		madd_v3_v3fl(shi->ao, mesh->ao[face[0]], w[0]);
+		madd_v3_v3fl(shi->indirect, mesh->indirect[face[0]], w[0]);
+		madd_v3_v3fl(shi->ao, mesh->ao[face[1]], w[1]);
+		madd_v3_v3fl(shi->indirect, mesh->indirect[face[1]], w[1]);
+		madd_v3_v3fl(shi->ao, mesh->ao[face[2]], w[2]);
+		madd_v3_v3fl(shi->indirect, mesh->indirect[face[2]], w[2]);
+		if(face[3]) {
+			madd_v3_v3fl(shi->ao, mesh->ao[face[3]], w[3]);
+			madd_v3_v3fl(shi->indirect, mesh->indirect[face[3]], w[3]);
+		}
 	}
 	else {
 		shi->ao[0]= 1.0f;
 		shi->ao[1]= 1.0f;
 		shi->ao[2]= 1.0f;
+		zero_v3(shi->indirect);
 	}
 }
 
@@ -1500,7 +1581,7 @@ static void *exec_strandsurface_sample(void *data)
 	OcclusionThread *othread= (OcclusionThread*)data;
 	Render *re= othread->re;
 	StrandSurface *mesh= othread->mesh;
-	float col[3], co[3], n[3], *co1, *co2, *co3, *co4;
+	float ao[3], indirect[3], co[3], n[3], *co1, *co2, *co3, *co4;
 	int a, *face;
 
 	for(a=othread->begin; a<othread->end; a++) {
@@ -1521,8 +1602,9 @@ static void *exec_strandsurface_sample(void *data)
 		}
 		negate_v3(n);
 
-		sample_occ_tree(re, re->occlusiontree, NULL, co, n, othread->thread, 0, col);
-		VECCOPY(othread->facecol[a], col);
+		sample_occ_tree(re, re->occlusiontree, NULL, co, n, othread->thread, 0, ao, indirect);
+		VECCOPY(othread->faceao[a], ao);
+		VECCOPY(othread->faceindirect[a], indirect);
 	}
 
 	return 0;
@@ -1533,7 +1615,7 @@ void make_occ_tree(Render *re)
 	OcclusionThread othreads[BLENDER_MAX_THREADS];
 	StrandSurface *mesh;
 	ListBase threads;
-	float col[3], (*facecol)[3];
+	float ao[3], indirect[3], (*faceao)[3], (*faceindirect)[3];
 	int a, totface, totthread, *face, *count;
 
 	/* ugly, needed for occ_face */
@@ -1545,21 +1627,25 @@ void make_occ_tree(Render *re)
 	re->occlusiontree= occ_tree_build(re);
 	
 	if(re->occlusiontree) {
-		if(re->wrld.ao_approx_passes)
+		if(re->wrld.ao_approx_passes > 0)
 			occ_compute_passes(re, re->occlusiontree, re->wrld.ao_approx_passes);
+		if(re->wrld.ao_indirect_bounces > 1)
+			occ_compute_bounces(re, re->occlusiontree, re->wrld.ao_indirect_bounces);
 
 		for(mesh=re->strandsurface.first; mesh; mesh=mesh->next) {
-			if(!mesh->face || !mesh->co || !mesh->col)
+			if(!mesh->face || !mesh->co || !mesh->ao)
 				continue;
 
 			count= MEM_callocN(sizeof(int)*mesh->totvert, "OcclusionCount");
-			facecol= MEM_callocN(sizeof(float)*3*mesh->totface, "StrandSurfFaceCol");
+			faceao= MEM_callocN(sizeof(float)*3*mesh->totface, "StrandSurfFaceAO");
+			faceindirect= MEM_callocN(sizeof(float)*3*mesh->totface, "StrandSurfFaceIndirect");
 
 			totthread= (mesh->totface > 10000)? re->r.threads: 1;
 			totface= mesh->totface/totthread;
 			for(a=0; a<totthread; a++) {
 				othreads[a].re= re;
-				othreads[a].facecol= facecol;
+				othreads[a].faceao= faceao;
+				othreads[a].faceindirect= faceindirect;
 				othreads[a].thread= a;
 				othreads[a].mesh= mesh;
 				othreads[a].begin= a*totface;
@@ -1581,26 +1667,36 @@ void make_occ_tree(Render *re)
 			for(a=0; a<mesh->totface; a++) {
 				face= mesh->face[a];
 
-				VECCOPY(col, facecol[a]);
-				VECADD(mesh->col[face[0]], mesh->col[face[0]], col);
+				VECCOPY(ao, faceao[a]);
+				VECCOPY(indirect, faceindirect[a]);
+
+				VECADD(mesh->ao[face[0]], mesh->ao[face[0]], ao);
+				VECADD(mesh->indirect[face[0]], mesh->indirect[face[0]], indirect);
 				count[face[0]]++;
-				VECADD(mesh->col[face[1]], mesh->col[face[1]], col);
+				VECADD(mesh->ao[face[1]], mesh->ao[face[1]], ao);
+				VECADD(mesh->indirect[face[1]], mesh->indirect[face[1]], indirect);
 				count[face[1]]++;
-				VECADD(mesh->col[face[2]], mesh->col[face[2]], col);
+				VECADD(mesh->ao[face[2]], mesh->ao[face[2]], ao);
+				VECADD(mesh->indirect[face[2]], mesh->indirect[face[2]], indirect);
 				count[face[2]]++;
 
 				if(face[3]) {
-					VECADD(mesh->col[face[3]], mesh->col[face[3]], col);
+					VECADD(mesh->ao[face[3]], mesh->ao[face[3]], ao);
+					VECADD(mesh->indirect[face[3]], mesh->indirect[face[3]], indirect);
 					count[face[3]]++;
 				}
 			}
 
-			for(a=0; a<mesh->totvert; a++)
-				if(count[a])
-					mul_v3_fl(mesh->col[a], 1.0f/count[a]);
+			for(a=0; a<mesh->totvert; a++) {
+				if(count[a]) {
+					mul_v3_fl(mesh->ao[a], 1.0f/count[a]);
+					mul_v3_fl(mesh->indirect[a], 1.0f/count[a]);
+				}
+			}
 
 			MEM_freeN(count);
-			MEM_freeN(facecol);
+			MEM_freeN(faceao);
+			MEM_freeN(faceindirect);
 		}
 	}
 }
@@ -1626,12 +1722,12 @@ void sample_occ(Render *re, ShadeInput *shi)
 			sample_occ_surface(shi);
 		}
 		/* try to get result from the cache if possible */
-		else if(shi->depth!=0 || !sample_occ_cache(tree, shi->co, shi->vno, shi->xs, shi->ys, shi->thread, shi->ao)) {
+		else if(shi->depth!=0 || !sample_occ_cache(tree, shi->co, shi->vno, shi->xs, shi->ys, shi->thread, shi->ao, shi->indirect)) {
 			/* no luck, let's sample the occlusion */
 			exclude.obi= shi->obi - re->objectinstance;
 			exclude.facenr= shi->vlr->index;
 			onlyshadow= (shi->mat->mode & MA_ONLYSHADOW);
-			sample_occ_tree(re, tree, &exclude, shi->co, shi->vno, shi->thread, onlyshadow, shi->ao);
+			sample_occ_tree(re, tree, &exclude, shi->co, shi->vno, shi->thread, onlyshadow, shi->ao, shi->indirect);
 
 			/* fill result into sample, each time */
 			if(tree->cache) {
@@ -1641,8 +1737,10 @@ void sample_occ(Render *re, ShadeInput *shi)
 					sample= &cache->sample[(shi->ys-cache->y)*cache->w + (shi->xs-cache->x)];
 					VECCOPY(sample->co, shi->co);
 					VECCOPY(sample->n, shi->vno);
-					VECCOPY(sample->col, shi->ao);
-					sample->intensity= MAX3(sample->col[0], sample->col[1], sample->col[2]);
+					VECCOPY(sample->ao, shi->ao);
+					VECCOPY(sample->indirect, shi->indirect);
+					sample->intensity= MAX3(sample->ao[0], sample->ao[1], sample->ao[2]);
+					sample->intensity= MAX2(sample->intensity, MAX3(sample->indirect[0], sample->indirect[1], sample->indirect[2]));
 					sample->dist2= INPR(shi->dxco, shi->dxco) + INPR(shi->dyco, shi->dyco);
 					sample->filled= 1;
 				}
@@ -1653,6 +1751,10 @@ void sample_occ(Render *re, ShadeInput *shi)
 		shi->ao[0]= 1.0f;
 		shi->ao[1]= 1.0f;
 		shi->ao[2]= 1.0f;
+
+		shi->indirect[0]= 0.0f;
+		shi->indirect[1]= 0.0f;
+		shi->indirect[2]= 0.0f;
 	}
 }
 
@@ -1720,12 +1822,14 @@ void cache_occ_samples(Render *re, RenderPart *pa, ShadeSample *ssamp)
 				onlyshadow= (shi->mat->mode & MA_ONLYSHADOW);
 				exclude.obi= shi->obi - re->objectinstance;
 				exclude.facenr= shi->vlr->index;
-				sample_occ_tree(re, tree, &exclude, shi->co, shi->vno, shi->thread, onlyshadow, shi->ao);
+				sample_occ_tree(re, tree, &exclude, shi->co, shi->vno, shi->thread, onlyshadow, shi->ao, shi->indirect);
 
 				VECCOPY(sample->co, shi->co);
 				VECCOPY(sample->n, shi->vno);
-				VECCOPY(sample->col, shi->ao);
-				sample->intensity= MAX3(sample->col[0], sample->col[1], sample->col[2]);
+				VECCOPY(sample->ao, shi->ao);
+				VECCOPY(sample->indirect, shi->indirect);
+				sample->intensity= MAX3(sample->ao[0], sample->ao[1], sample->ao[2]);
+				sample->intensity= MAX2(sample->intensity, MAX3(sample->indirect[0], sample->indirect[1], sample->indirect[2]));
 				sample->dist2= INPR(shi->dxco, shi->dxco) + INPR(shi->dyco, shi->dyco);
 				sample->x= shi->xs;
 				sample->y= shi->ys;
diff --git a/source/blender/render/intern/source/pipeline.c b/source/blender/render/intern/source/pipeline.c
index b357a17e244..08440798826 100644
--- a/source/blender/render/intern/source/pipeline.c
+++ b/source/blender/render/intern/source/pipeline.c
@@ -1065,12 +1065,25 @@ void RE_ResultGet32(Render *re, unsigned int *rect)
 		int tot= rres.rectx*rres.recty;
 		char *cp= (char *)rect;
 		
-		for(;tot>0; tot--, cp+=4, fp+=4) {
-			cp[0] = FTOCHAR(fp[0]);
-			cp[1] = FTOCHAR(fp[1]);
-			cp[2] = FTOCHAR(fp[2]);
-			cp[3] = FTOCHAR(fp[3]);
+		if (re->r.color_mgt_flag & R_COLOR_MANAGEMENT) {
+			/* Finally convert back to sRGB rendered image */ 
+			for(;tot>0; tot--, cp+=4, fp+=4) {
+				cp[0] = FTOCHAR(linearrgb_to_srgb(fp[0]));
+				cp[1] = FTOCHAR(linearrgb_to_srgb(fp[1]));
+				cp[2] = FTOCHAR(linearrgb_to_srgb(fp[2]));
+				cp[3] = FTOCHAR(fp[3]);
+			}
 		}
+		else {
+			/* Color management is off : no conversion necessary */
+			for(;tot>0; tot--, cp+=4, fp+=4) {
+				cp[0] = FTOCHAR(fp[0]);
+				cp[1] = FTOCHAR(fp[1]);
+				cp[2] = FTOCHAR(fp[2]);
+				cp[3] = FTOCHAR(fp[3]);
+			}
+		}
+
 	}
 	else
 		/* else fill with black */
diff --git a/source/blender/render/intern/source/pixelshading.c b/source/blender/render/intern/source/pixelshading.c
index d4b7c403f50..6fef8279c7e 100644
--- a/source/blender/render/intern/source/pixelshading.c
+++ b/source/blender/render/intern/source/pixelshading.c
@@ -537,15 +537,9 @@ void shadeSkyView(float *colf, float *rco, float *view, float *dxyview, short th
 		/* the fraction of how far we are above the bottom of the screen */
 		blend= fabs(0.5+ view[1]);
 	}
-	
-	if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
-		color_manage_linearize(hor, &R.wrld.horr);
-		color_manage_linearize(zen, &R.wrld.zenr);
-	}
-	else {
-		VECCOPY(hor, &R.wrld.horr);
-		VECCOPY(zen, &R.wrld.zenr);
-	}
+
+	VECCOPY(hor, &R.wrld.horr);
+	VECCOPY(zen, &R.wrld.zenr);
 
 	/* Careful: SKYTEX and SKYBLEND are NOT mutually exclusive! If           */
 	/* SKYBLEND is active, the texture and color blend are added.           */
@@ -633,10 +627,7 @@ void shadeSkyPixel(float *collector, float fx, float fy, short thread)
 	} 
 	else if((R.wrld.skytype & (WO_SKYBLEND+WO_SKYTEX))==0) {
 		/* 2. solid color */
-		if(R.r.color_mgt_flag & R_COLOR_MANAGEMENT)
-			color_manage_linearize(collector, &R.wrld.horr);
-		else
-			VECCOPY(collector, &R.wrld.horr);
+		VECCOPY(collector, &R.wrld.horr);
 
 		collector[3] = 0.0f;
 	} 
diff --git a/source/blender/render/intern/source/shadeinput.c b/source/blender/render/intern/source/shadeinput.c
index 396c713cfb7..fd1e27ab28a 100644
--- a/source/blender/render/intern/source/shadeinput.c
+++ b/source/blender/render/intern/source/shadeinput.c
@@ -91,42 +91,16 @@ extern struct Render R;
  * doing inverse gamma correction where applicable */
 void shade_input_init_material(ShadeInput *shi)
 {
-	if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
-		color_manage_linearize(&shi->r, &shi->mat->r);
-		color_manage_linearize(&shi->specr, &shi->mat->specr);
-		color_manage_linearize(&shi->mirr, &shi->mat->mirr);
-		
-		/* material ambr / ambg / ambb is overwritten from world
-		color_manage_linearize(shi->ambr, shi->mat->ambr);
-		*/
-		
-		/* note, keep this synced with render_types.h */
-		memcpy(&shi->amb, &shi->mat->amb, 11*sizeof(float));
-		shi->har= shi->mat->har;
-	} else {
-		/* note, keep this synced with render_types.h */
-		memcpy(&shi->r, &shi->mat->r, 23*sizeof(float));
-		shi->har= shi->mat->har;
-	}
-
-}
-
-static void shadeinput_colors_linearize(ShadeInput *shi)
-{
-	color_manage_linearize(&shi->r, &shi->r);
-	color_manage_linearize(&shi->specr, &shi->specr);
-	color_manage_linearize(&shi->mirr, &shi->mirr);
+	/* note, keep this synced with render_types.h */
+	memcpy(&shi->r, &shi->mat->r, 23*sizeof(float));
+	shi->har= shi->mat->har;
 }
 
 /* also used as callback for nodes */
 /* delivers a fully filled in ShadeResult, for all passes */
 void shade_material_loop(ShadeInput *shi, ShadeResult *shr)
 {
-	/* because node materials don't have access to rendering context,
-	 * inverse gamma correction must happen here. evil. */
-	if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT && shi->nodes == 1)
-		shadeinput_colors_linearize(shi);
-	
+
 	shade_lamp_loop(shi, shr);	/* clears shr */
 	
 	if(shi->translucency!=0.0f) {
@@ -626,7 +600,7 @@ void shade_input_set_strand_texco(ShadeInput *shi, StrandRen *strand, StrandVert
 	
 	if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
 		if(mode & (MA_VERTEXCOL|MA_VERTEXCOLP|MA_FACETEXTURE)) {
-			color_manage_linearize(shi->vcol, shi->vcol);
+			srgb_to_linearrgb_v3_v3(shi->vcol, shi->vcol);
 		}
 	}
 	
@@ -1291,7 +1265,7 @@ void shade_input_set_shade_texco(ShadeInput *shi)
 	*/
 	if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
 		if(mode & (MA_VERTEXCOL|MA_VERTEXCOLP|MA_FACETEXTURE)) {
-			color_manage_linearize(shi->vcol, shi->vcol);
+			srgb_to_linearrgb_v3_v3(shi->vcol, shi->vcol);
 		}
 	}
 	
diff --git a/source/blender/render/intern/source/shadeoutput.c b/source/blender/render/intern/source/shadeoutput.c
index f167122f497..958a2e34215 100644
--- a/source/blender/render/intern/source/shadeoutput.c
+++ b/source/blender/render/intern/source/shadeoutput.c
@@ -1041,6 +1041,7 @@ void ambient_occlusion_to_diffuse(ShadeInput *shi, float *diff)
 			}
 			
 			VECMUL(diff, f);
+			madd_v3_v3fl(diff, shi->indirect, R.wrld.ao_indirect_energy*shi->amb);
 		}
 		else
 			diff[0]= diff[1]= diff[2]= 0.0f;
diff --git a/source/blender/render/intern/source/sss.c b/source/blender/render/intern/source/sss.c
index 25cfc0f1253..e551cf0f3d2 100644
--- a/source/blender/render/intern/source/sss.c
+++ b/source/blender/render/intern/source/sss.c
@@ -926,20 +926,17 @@ static void sss_create_tree_mat(Render *re, Material *mat)
 	if(!re->test_break(re->tbh)) {
 		SSSData *sss= MEM_callocN(sizeof(*sss), "SSSData");
 		float ior= mat->sss_ior, cfac= mat->sss_colfac;
-		float col[3], *radius= mat->sss_radius;
+		float *radius= mat->sss_radius;
 		float fw= mat->sss_front, bw= mat->sss_back;
 		float error = mat->sss_error;
 
 		error= get_render_aosss_error(&re->r, error);
 		if((re->r.scemode & R_PREVIEWBUTS) && error < 0.5f)
 			error= 0.5f;
-
-		if (re->r.color_mgt_flag & R_COLOR_MANAGEMENT) color_manage_linearize(col, mat->sss_col);
-		else VECCOPY(col, mat->sss_col);
 		
-		sss->ss[0]= scatter_settings_new(col[0], radius[0], ior, cfac, fw, bw);
-		sss->ss[1]= scatter_settings_new(col[1], radius[1], ior, cfac, fw, bw);
-		sss->ss[2]= scatter_settings_new(col[2], radius[2], ior, cfac, fw, bw);
+		sss->ss[0]= scatter_settings_new(mat->sss_col[0], radius[0], ior, cfac, fw, bw);
+		sss->ss[1]= scatter_settings_new(mat->sss_col[1], radius[1], ior, cfac, fw, bw);
+		sss->ss[2]= scatter_settings_new(mat->sss_col[2], radius[2], ior, cfac, fw, bw);
 		sss->tree= scatter_tree_new(sss->ss, mat->sss_scale, error,
 			co, color, area, totpoint);
 
diff --git a/source/blender/render/intern/source/strand.c b/source/blender/render/intern/source/strand.c
index 47a7c052b18..ecea7360974 100644
--- a/source/blender/render/intern/source/strand.c
+++ b/source/blender/render/intern/source/strand.c
@@ -959,7 +959,8 @@ StrandSurface *cache_strand_surface(Render *re, ObjectRen *obr, DerivedMesh *dm,
 		mesh->totvert= totvert;
 		mesh->totface= totface;
 		mesh->face= MEM_callocN(sizeof(int)*4*mesh->totface, "StrandSurfFaces");
-		mesh->col= MEM_callocN(sizeof(float)*3*mesh->totvert, "StrandSurfCol");
+		mesh->ao= MEM_callocN(sizeof(float)*3*mesh->totvert, "StrandSurfAO");
+		mesh->indirect= MEM_callocN(sizeof(float)*3*mesh->totvert, "StrandSurfIndirect");
 		BLI_addtail(&re->strandsurface, mesh);
 	}
 
@@ -997,7 +998,8 @@ void free_strand_surface(Render *re)
 		if(mesh->co) MEM_freeN(mesh->co);
 		if(mesh->prevco) MEM_freeN(mesh->prevco);
 		if(mesh->nextco) MEM_freeN(mesh->nextco);
-		if(mesh->col) MEM_freeN(mesh->col);
+		if(mesh->ao) MEM_freeN(mesh->ao);
+		if(mesh->indirect) MEM_freeN(mesh->indirect);
 		if(mesh->face) MEM_freeN(mesh->face);
 	}
 
diff --git a/source/blender/render/intern/source/texture.c b/source/blender/render/intern/source/texture.c
index a9f6e7a53fd..16372d7a15d 100644
--- a/source/blender/render/intern/source/texture.c
+++ b/source/blender/render/intern/source/texture.c
@@ -2051,8 +2051,13 @@ void do_material_tex(ShadeInput *shi)
 				else texres.tin= texres.ta;
 				
 				/* inverse gamma correction */
-				if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
-					color_manage_linearize(tcol, tcol);
+				if (tex->type==TEX_IMAGE) {
+					Image *ima = tex->ima;
+					ImBuf *ibuf = BKE_image_get_ibuf(ima, &tex->iuser);
+					
+					/* don't linearize float buffers, assumed to be linear */
+					if (ibuf && !(ibuf->rect_float) && R.r.color_mgt_flag & R_COLOR_MANAGEMENT)
+						srgb_to_linearrgb_v3_v3(tcol, tcol);
 				}
 				
 				if(mtex->mapto & MAP_COL) {
@@ -2404,11 +2409,6 @@ void do_volume_tex(ShadeInput *shi, float *xyz, int mapto_flag, float *col, floa
 						texres.tin= texres.ta;
 				}
 				
-				/* inverse gamma correction */
-				if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
-					color_manage_linearize(tcol, tcol);
-				}
-				
 				/* used for emit */
 				if((mapto_flag & MAP_EMISSION_COL) && (mtex->mapto & MAP_EMISSION_COL)) {
 					float colemitfac= mtex->colemitfac*stencilTin;
@@ -2556,8 +2556,13 @@ void do_halo_tex(HaloRen *har, float xn, float yn, float *colf)
 		else texres.tin= texres.ta;
 
 		/* inverse gamma correction */
-		if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
-			color_manage_linearize(&texres.tr, &texres.tr);
+		if (mtex->tex->type==TEX_IMAGE) {
+			Image *ima = mtex->tex->ima;
+			ImBuf *ibuf = BKE_image_get_ibuf(ima, &mtex->tex->iuser);
+			
+			/* don't linearize float buffers, assumed to be linear */
+			if (ibuf && !(ibuf->rect_float) && R.r.color_mgt_flag & R_COLOR_MANAGEMENT)
+				srgb_to_linearrgb_v3_v3(&texres.tr, &texres.tr);
 		}
 
 		fact= texres.tin*mtex->colfac;
@@ -2605,6 +2610,7 @@ void do_halo_tex(HaloRen *har, float xn, float yn, float *colf)
 void do_sky_tex(float *rco, float *lo, float *dxyview, float *hor, float *zen, float *blend, int skyflag, short thread)
 {
 	MTex *mtex;
+	Tex *tex;
 	TexResult texres= {0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0, NULL};
 	float *co, fact, stencilTin=1.0;
 	float tempvec[3], texvec[3], dxt[3], dyt[3];
@@ -2618,7 +2624,8 @@ void do_sky_tex(float *rco, float *lo, float *dxyview, float *hor, float *zen, f
 		if(R.wrld.mtex[tex_nr]) {
 			mtex= R.wrld.mtex[tex_nr];
 			
-			if(mtex->tex==0) continue;
+			tex= mtex->tex;
+			if(tex==0) continue;
 			/* if(mtex->mapto==0) continue; */
 			
 			/* which coords */
@@ -2700,7 +2707,7 @@ void do_sky_tex(float *rco, float *lo, float *dxyview, float *hor, float *zen, f
 			else texvec[2]= mtex->size[2]*(mtex->ofs[2]);
 			
 			/* texture */
-			if(mtex->tex->type==TEX_IMAGE) do_2d_mapping(mtex, texvec, NULL, NULL, dxt, dyt);
+			if(tex->type==TEX_IMAGE) do_2d_mapping(mtex, texvec, NULL, NULL, dxt, dyt);
 		
 			rgb= multitex(mtex->tex, texvec, dxt, dyt, R.osa, &texres, thread, mtex->which_output);
 			
@@ -2748,8 +2755,13 @@ void do_sky_tex(float *rco, float *lo, float *dxyview, float *hor, float *zen, f
 				tcol[0]= texres.tr; tcol[1]= texres.tg; tcol[2]= texres.tb;
 
 				/* inverse gamma correction */
-				if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
-					color_manage_linearize(tcol, tcol);
+				if (tex->type==TEX_IMAGE) {
+					Image *ima = tex->ima;
+					ImBuf *ibuf = BKE_image_get_ibuf(ima, &tex->iuser);
+					
+					/* don't linearize float buffers, assumed to be linear */
+					if (ibuf && !(ibuf->rect_float) && R.r.color_mgt_flag & R_COLOR_MANAGEMENT)
+						srgb_to_linearrgb_v3_v3(tcol, tcol);
 				}
 
 				if(mtex->mapto & WOMAP_HORIZ) {
@@ -2947,8 +2959,13 @@ void do_lamp_tex(LampRen *la, float *lavec, ShadeInput *shi, float *colf, int ef
 				else texres.tin= texres.ta;
 
 				/* inverse gamma correction */
-				if (R.r.color_mgt_flag & R_COLOR_MANAGEMENT) {
-					color_manage_linearize(&texres.tr, &texres.tr);
+				if (tex->type==TEX_IMAGE) {
+					Image *ima = tex->ima;
+					ImBuf *ibuf = BKE_image_get_ibuf(ima, &tex->iuser);
+					
+					/* don't linearize float buffers, assumed to be linear */
+					if (ibuf && !(ibuf->rect_float) && R.r.color_mgt_flag & R_COLOR_MANAGEMENT)
+						srgb_to_linearrgb_v3_v3(&texres.tr, &texres.tr);
 				}
 
 				/* lamp colors were premultiplied with this */