Différences

Ci-dessous, les différences entre deux révisions de la page.

--- blender [2017/12/02 17:15]
bvandepo
+++ blender [2017/12/15 15:50] (Version actuelle)
jcombier
@@ Ligne 578: / Ligne 578: @@
   >>> import numpy
   >>> numpy.version.version
+=====Animation modèle de visage=====
+charger la scène Swirski-EyeModel.blend
+Mode d'affichage default
+Développer l'arbre Armature Head->the rig->Pose_head.002->eyetargetparent
+à droite, choisir mode Bone (OS)
+en bas, chosir Pose Mode
+à droite dans  transform->Location, changer x,y,z pour spécifier la cible visée par l'oeil
+passer en mode scripting
+recliquer sur Bone et changer  transform->Location, la commande python correspondante apparait dans la console:
+  bpy.context.object.pose.bones["eyetargetparent"].location[0] = -0.113456
+bpy.context dépend de l'objet actuellement selectionné dans la GUI, pour modifier dans le script, il faut retrouver dans l'arborescence de la scene (s'aider de l'auto complétion avec CTRL+SPACE)
+  scene = bpy.context.scene
+  armature = bpy.data.objects['Armature Head']
+  armature.pose.bones['eyetargetparent'].location[0] = -0.113456
+  ./blender-2.74-linux-glibc211-x86_64/blender --python ./Swirski-EyeModel.py -b
+<file python Swirski-EyeModel.py>
+import bpy
+import math
+import mathutils
+import os
+import sys
+import numpy as np
+#lancer avec  ./blender-2.74-linux-glibc211-x86_64/blender --python ./Swirski-EyeModel.py -b
+#****** CREATE DIRECTORY WHERE TO SAVE IMAGES ***********
+img_dir = '/tmp/images/'
+if not os.path.exists(img_dir):
+    os.makedirs(img_dir)
+bpy.ops.wm.open_mainfile(filepath="/media/HD500GO/blender/Swirski-EyeModel.blend")
+scene = bpy.context.scene
+armature = bpy.data.objects['Armature Head']
+camera_obj = bpy.data.objects['Camera']
+camera = bpy.data.cameras['Camera']
+eyeL = bpy.data.objects['eye.L']
+pupilGroup = eyeL.vertex_groups['eyepulpex.L']
+pupilVertices = [v for v in eyeL.data.vertices if pupilGroup.index in [g.group for g in v.groups]]
+pupil_base_radius = max((v1.co - v2.co).length for v1 in pupilVertices for v2 in pupilVertices) * eyeL.scale[1]
+eyeLbone = armature.pose.bones['def_eye.L']
+pupilLbone = armature.pose.bones['eyepulpex.L']
+def strVec(vec):
+    return "({},{},{})".format(vec[0], vec[1], vec[2])
+datafilepath = os.path.join(os.path.dirname(bpy.data.filepath), "render_eye_data.txt")
+# switch on nodes
+scene.use_nodes = True
+tree = scene.node_tree
+links = tree.links
+# clear default nodes
+for n in tree.nodes:
+    tree.nodes.remove(n)
+# create input render layer node
+rl = tree.nodes.new('CompositorNodeRLayers')
+rl.location = 185,285
+# create output node
+v = tree.nodes.new('CompositorNodeViewer')
+v.location = 750,210
+v.use_alpha = False
+# create output node
+of_c_node = tree.nodes.new('CompositorNodeOutputFile')
+of_c_node.location = 600, 200
+#of_node.base_path =
+of_c_node.format.file_format = 'PNG'
+# Links
+links.new(rl.outputs[0], v.inputs[0])  # link Image output to Viewer input
+links.new(rl.outputs[0], of_c_node.inputs[0])
+# Define path where to save image
+base_path = "/tmp/images" #"./images_test_png"
+with open(datafilepath, "w") as datafile:
+    #for frame in range(scene.frame_end+1):
+    for frame in range(2):
+        scene.frame_set(frame)
+        #modification position de visée de l'oeil
+        print('Jessica et Bertrand sont passés par la');
+        armature.pose.bones['eyetargetparent'].location[0] = -0.113456
+        camera_mat = camera_obj.matrix_world * mathutils.Matrix([[1,0,0,0],[0,-1,0,0],[0,0,-1,0],[0,0,0,1]])
+        camera_invmat = camera_mat.inverted()
+        armature_mat = armature.matrix_world
+        head_world = armature_mat*eyeLbone.head
+        tail_world = armature_mat*eyeLbone.tail
+        head_cam = camera_invmat*head_world
+        tail_cam = camera_invmat*tail_world
+        eye_centre = head_cam
+        eye_radius = eyeLbone.bone.length
+        pupil_gaze = (tail_cam - head_cam).normalized()
+        pupil_radius = pupilLbone.scale[0] * pupil_base_radius
+        pupil_centre = eye_centre + eye_radius*pupil_gaze
+        #framestr = "{} | head_world: {} | tail_world: {} | vec_world: {} | head_cam: {} | tail_cam: {} | vec_cam {}".format(
+        #       frame, *[strVec(x) for x in [head_world,tail_world,vec_world,head_cam,tail_cam,vec_cam]])
+        framestr = "{} {} {} {} {} {}".format(
+            frame,
+            strVec(eye_centre),
+            eye_radius,
+            strVec(pupil_centre),
+            strVec(pupil_gaze),
+            pupil_radius
+            )
+        print(framestr)
+        print(framestr, file=datafile)
+        #************* RENDER AND SAVE IMAGES ***********
+# Define image resolution for rendering
+        scene.render.resolution_x = 250
+        scene.render.resolution_y = 250
+# resolution percentage : have to be 100% to have the whole image resolution defined earlier
+        scene.render.resolution_percentage = 100
+# Activate which camera will be used for rendering
+# (if more than one camera are defined, the rendering have to be repeated for each camera)
+        scene.camera = bpy.data.objects["Camera"]
+# get viewer pixels directly
+        #pixels = bpy.data.images['Viewer Node'].pixels
+        #print(len(pixels)) # size is always width * height * 4 (rgba)
+# copy buffer to numpy array for faster manipulation
+#       arr = np.array(pixels[:])
+#        print('one pixel \n',arr[100:104])
+# Define path where to save images
+        str4 = base_path + "/image"
+        of_c_node.base_path = str4 + "000"
+        #out_z_node.base_path = str4 +"_Z_"+ "000"
+# Render
+        bpy.ops.render.render()#write_still=True)
+#************* EXIT BLENDER  ***********
+        sys.exit(0)
+</file>
+{{http://homepages.laas.fr/bvandepo/files/oeil.png}}
+==== Accélérer le rendu ====
+Plusieurs astuces existent pour accelérer le rendu. Ces astuces diffèrent suivant si on est en "Blender render" ou en "Cycle render".
+Lien pour le "Cycle render":
+  * [[https://www.blenderguru.com/articles/4-easy-ways-to-speed-up-cycles]]
+  * [[http://boundlessblending.com/blender-fast-rendering/]]
+Lien pour "Blender render":
+  * [[https://www.blenderguru.com/articles/13-ways-to-reduce-render-times]]
+Pour le calcul de Blender occupe 100% du CPU il faut indiquer le bon nombre de threads et la bonne taille de tuile ("Tile size") : [[https://blenderartists.org/forum/showthread.php?410469-Why-Blender-Cycles-Rendering-Not-Using-100-CPU]]

WIKI de Bertrand VANDEPORTAELE

Outils pour utilisateurs

Outils du site

Différences

Outils de la page