blender/release/scripts/op/io_scene_obj/export_obj.py

883 lines
33 KiB
Python
Raw Normal View History

# ##### BEGIN GPL LICENSE BLOCK #####
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
2009-12-13 14:00:39 +00:00
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
2009-12-13 14:00:39 +00:00
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software Foundation,
2010-02-12 13:34:04 +00:00
# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
#
# ##### END GPL LICENSE BLOCK #####
# <pep8 compliant>
import os
import time
import shutil
import bpy
import mathutils
def fixName(name):
2009-12-13 14:00:39 +00:00
if name == None:
return 'None'
else:
return name.replace(' ', '_')
def write_mtl(scene, filepath, copy_images, mtl_dict):
2009-12-13 14:00:39 +00:00
world = scene.world
worldAmb = world.ambient_color
dest_dir = os.path.dirname(filepath)
2009-12-13 14:00:39 +00:00
def copy_image(image):
fn = bpy.path.abspath(image.filepath)
fn_strip = os.path.basename(fn)
2009-12-13 14:00:39 +00:00
if copy_images:
rel = fn_strip
fn_abs_dest = os.path.join(dest_dir, fn_strip)
if not os.path.exists(fn_abs_dest):
shutil.copy(fn, fn_abs_dest)
else:
rel = fn
2009-12-13 14:00:39 +00:00
return rel
file = open(filepath, "w")
2009-12-13 14:00:39 +00:00
# XXX
# file.write('# Blender MTL File: %s\n' % Blender.Get('filepath').split('\\')[-1].split('/')[-1])
file.write('# Material Count: %i\n' % len(mtl_dict))
2009-12-13 14:00:39 +00:00
# Write material/image combinations we have used.
for key, (mtl_mat_name, mat, img) in mtl_dict.items():
2009-12-13 14:00:39 +00:00
# Get the Blender data for the material and the image.
# Having an image named None will make a bug, dont do it :)
file.write('newmtl %s\n' % mtl_mat_name) # Define a new material: matname_imgname
if mat:
file.write('Ns %.6f\n' % ((mat.specular_hardness-1) * 1.9607843137254901) ) # Hardness, convert blenders 1-511 to MTL's
file.write('Ka %.6f %.6f %.6f\n' % tuple([c*mat.ambient for c in worldAmb]) ) # Ambient, uses mirror colour,
2009-12-13 14:00:39 +00:00
file.write('Kd %.6f %.6f %.6f\n' % tuple([c*mat.diffuse_intensity for c in mat.diffuse_color]) ) # Diffuse
file.write('Ks %.6f %.6f %.6f\n' % tuple([c*mat.specular_intensity for c in mat.specular_color]) ) # Specular
if hasattr(mat, "ior"):
file.write('Ni %.6f\n' % mat.ior) # Refraction index
else:
file.write('Ni %.6f\n' % 1.0)
file.write('d %.6f\n' % mat.alpha) # Alpha (obj uses 'd' for dissolve)
# 0 to disable lighting, 1 for ambient & diffuse only (specular color set to black), 2 for full lighting.
if mat.use_shadeless:
2009-12-13 14:00:39 +00:00
file.write('illum 0\n') # ignore lighting
elif mat.specular_intensity == 0:
file.write('illum 1\n') # no specular.
else:
file.write('illum 2\n') # light normaly
else:
#write a dummy material here?
file.write('Ns 0\n')
file.write('Ka %.6f %.6f %.6f\n' % tuple([c for c in worldAmb]) ) # Ambient, uses mirror colour,
2009-12-13 14:00:39 +00:00
file.write('Kd 0.8 0.8 0.8\n')
file.write('Ks 0.8 0.8 0.8\n')
file.write('d 1\n') # No alpha
file.write('illum 2\n') # light normaly
# Write images!
if img: # We have an image on the face!
2009-12-13 14:00:39 +00:00
# write relative image path
rel = copy_image(img)
file.write('map_Kd %s\n' % rel) # Diffuse mapping image
# file.write('map_Kd %s\n' % img.filepath.split('\\')[-1].split('/')[-1]) # Diffuse mapping image
2009-12-13 14:00:39 +00:00
elif mat: # No face image. if we havea material search for MTex image.
for mtex in mat.texture_slots:
2009-12-13 14:00:39 +00:00
if mtex and mtex.texture.type == 'IMAGE':
try:
filepath = copy_image(mtex.texture.image)
# filepath = mtex.texture.image.filepath.split('\\')[-1].split('/')[-1]
file.write('map_Kd %s\n' % repr(filepath)[1:-1]) # Diffuse mapping image
2009-12-13 14:00:39 +00:00
break
except:
# Texture has no image though its an image type, best ignore.
pass
file.write('\n\n')
file.close()
# XXX not used
def copy_file(source, dest):
2009-12-13 14:00:39 +00:00
file = open(source, 'rb')
data = file.read()
file.close()
file = open(dest, 'wb')
file.write(data)
file.close()
# XXX not used
def copy_images(dest_dir):
2009-12-13 14:00:39 +00:00
if dest_dir[-1] != os.sep:
dest_dir += os.sep
# if dest_dir[-1] != sys.sep:
# dest_dir += sys.sep
2009-12-13 14:00:39 +00:00
# Get unique image names
uniqueImages = {}
for matname, mat, image in mtl_dict.values(): # Only use image name
2009-12-13 14:00:39 +00:00
# Get Texface images
if image:
uniqueImages[image] = image # Should use sets here. wait until Python 2.4 is default.
# Get MTex images
if mat:
for mtex in mat.texture_slots:
2009-12-13 14:00:39 +00:00
if mtex and mtex.texture.type == 'IMAGE':
image_tex = mtex.texture.image
if image_tex:
try:
uniqueImages[image_tex] = image_tex
except:
pass
# Now copy images
copyCount = 0
# for bImage in uniqueImages.values():
# image_path = bpy.path.abspath(bImage.filepath)
# if bpy.sys.exists(image_path):
# # Make a name for the target path.
# dest_image_path = dest_dir + image_path.split('\\')[-1].split('/')[-1]
# if not bpy.utils.exists(dest_image_path): # Image isnt already there
# print('\tCopying "%s" > "%s"' % (image_path, dest_image_path))
# copy_file(image_path, dest_image_path)
# copyCount+=1
# paths= bpy.util.copy_images(uniqueImages.values(), dest_dir)
2009-12-13 14:00:39 +00:00
print('\tCopied %d images' % copyCount)
def test_nurbs_compat(ob):
if ob.type != 'CURVE':
2009-12-13 14:00:39 +00:00
return False
for nu in ob.data.splines:
if nu.point_count_v == 1 and nu.type != 'BEZIER': # not a surface and not bezier
2009-12-13 14:00:39 +00:00
return True
return False
def write_nurb(file, ob, ob_mat):
2009-12-13 14:00:39 +00:00
tot_verts = 0
cu = ob.data
# use negative indices
for nu in cu.splines:
if nu.type == 'POLY':
DEG_ORDER_U = 1
else:
DEG_ORDER_U = nu.order_u - 1 # odd but tested to be correct
2009-12-13 14:00:39 +00:00
if nu.type == 'BEZIER':
2009-12-13 14:00:39 +00:00
print("\tWarning, bezier curve:", ob.name, "only poly and nurbs curves supported")
continue
if nu.point_count_v > 1:
2009-12-13 14:00:39 +00:00
print("\tWarning, surface:", ob.name, "only poly and nurbs curves supported")
continue
if len(nu.points) <= DEG_ORDER_U:
print("\tWarning, order_u is lower then vert count, skipping:", ob.name)
2009-12-13 14:00:39 +00:00
continue
pt_num = 0
do_closed = nu.use_cyclic_u
do_endpoints = (do_closed == 0) and nu.use_endpoint_u
2009-12-13 14:00:39 +00:00
for pt in nu.points:
pt = ob_mat * pt.co.copy().resize3D()
2009-12-13 14:00:39 +00:00
file.write('v %.6f %.6f %.6f\n' % (pt[0], pt[1], pt[2]))
pt_num += 1
tot_verts += pt_num
file.write('g %s\n' % (fixName(ob.name))) # fixName(ob.getData(1)) could use the data name too
file.write('cstype bspline\n') # not ideal, hard coded
file.write('deg %d\n' % DEG_ORDER_U) # not used for curves but most files have it still
curve_ls = [-(i+1) for i in range(pt_num)]
# 'curv' keyword
if do_closed:
if DEG_ORDER_U == 1:
pt_num += 1
curve_ls.append(-1)
else:
pt_num += DEG_ORDER_U
curve_ls = curve_ls + curve_ls[0:DEG_ORDER_U]
file.write('curv 0.0 1.0 %s\n' % (' '.join([str(i) for i in curve_ls]))) # Blender has no U and V values for the curve
2009-12-13 14:00:39 +00:00
# 'parm' keyword
tot_parm = (DEG_ORDER_U + 1) + pt_num
tot_parm_div = float(tot_parm-1)
parm_ls = [(i/tot_parm_div) for i in range(tot_parm)]
if do_endpoints: # end points, force param
for i in range(DEG_ORDER_U+1):
parm_ls[i] = 0.0
parm_ls[-(1+i)] = 1.0
file.write('parm u %s\n' % ' '.join( [str(i) for i in parm_ls] ))
file.write('end\n')
return tot_verts
def write_file(filepath, objects, scene,
2009-12-13 14:00:39 +00:00
EXPORT_TRI=False,
EXPORT_EDGES=False,
EXPORT_NORMALS=False,
EXPORT_NORMALS_HQ=False,
EXPORT_UV=True,
EXPORT_MTL=True,
EXPORT_COPY_IMAGES=False,
EXPORT_APPLY_MODIFIERS=True,
EXPORT_ROTX90=True,
EXPORT_BLEN_OBS=True,
EXPORT_GROUP_BY_OB=False,
EXPORT_GROUP_BY_MAT=False,
EXPORT_KEEP_VERT_ORDER=False,
EXPORT_POLYGROUPS=False,
EXPORT_CURVE_AS_NURBS=True):
'''
Basic write function. The context and options must be already set
2009-12-13 14:00:39 +00:00
This can be accessed externaly
eg.
write( 'c:\\test\\foobar.obj', Blender.Object.GetSelected() ) # Using default options.
'''
# XXX
import math
def veckey3d(v):
return round(v.x, 6), round(v.y, 6), round(v.z, 6)
def veckey2d(v):
return round(v[0], 6), round(v[1], 6)
# return round(v.x, 6), round(v.y, 6)
def findVertexGroupName(face, vWeightMap):
"""
Searches the vertexDict to see what groups is assigned to a given face.
We use a frequency system in order to sort out the name because a given vetex can
belong to two or more groups at the same time. To find the right name for the face
we list all the possible vertex group names with their frequency and then sort by
frequency in descend order. The top element is the one shared by the highest number
of vertices is the face's group
"""
weightDict = {}
2010-08-18 03:42:26 +00:00
for vert_index in face.vertices:
# for vert in face:
2009-12-13 14:00:39 +00:00
vWeights = vWeightMap[vert_index]
# vWeights = vWeightMap[vert]
2009-12-13 14:00:39 +00:00
for vGroupName, weight in vWeights:
weightDict[vGroupName] = weightDict.get(vGroupName, 0) + weight
[#18694] Added support for Wavefront OBJ groups for both the import and export scripts from Paolo Ciccone (pciccone) Im not that keen on this functionality since vertex groups in blender are for weight painting and OBJ's groups are more like sub-objects, The conversion between these is a poor approximation (one vertex group per face). But this is still useful for using blender with poser so accepting these 2 patches. Made some changes from the patch *Export* - vgroup off by default, since this is mainly useful for corner cases - findVertexGroupName() - changed how exporting selects a vertex group for a face, use the highest combind weight for each group instead of the number of checking how many times the verts are in a group because they could have a zero weight. - findVertexGroupName() - would fail if a faces verts were not in any groups, now return a (null) group - Allow modifiers to export with vgroups *Import* - disable by default - ignore (null) groups - fixed a problem where the patch broke SPLIT_GROUPS - move the group definition out of global namespace, breaks importing as a module. --- Patch submission info The import/export scripts included with Blender 2.48 don't save/preserve groups defined in OBJ file. This presents a couple of issues: users loading an obj file with groups are not able to use a rather important set of data. When the file is then saved the groups are lost bringing up both the issue of data loss and loss of functionality as groups are used for rigging and morph target creation for Poser/DAZ Studio and all compatible applications. Of course other animation/modeling packages that use OBJ files might be affected. The solution that I present here simply patches the excellent python scripts written by Campbell Barton and Jiri Hnidek. The new code implements the preservation of groups by creating Blender vertex groups during import and by re-writing those groups in the OBJ file during export. Of course group ownership is idetermined at the face/polygon level and not at the vertex level. This rather simple solution makes Blender instantly one of the better tools to create Poser Morph targets (Full Body Morphs) and conforming figures and clothing. Of course it's also adding a more complete support for the whole OBJ spec. The patches are fitting well inside the established logic, have a minimal impact and are totally transparent to the user. See the attached video (about 7 minutes long) for more details. I noticed that the original scripts are formatted with tabs. I think it would be a good idea to follow the standard Python recommendation and convert the scripts to use spaces. I didn't want to do it on my side as it would generate a patch that replaces the whole file, not something that I wanted to do with my first submission :) But it would help if people with access to SVN could do it. Here is a short video that demonstrates the updates: http://www.paolociccone.com/blender_obj_scripts.html Cheers! ----
2009-05-13 00:20:14 +00:00
2009-12-13 14:00:39 +00:00
if weightDict:
alist = [(weight,vGroupName) for vGroupName, weight in weightDict.items()] # sort least to greatest amount of weight
alist.sort()
return(alist[-1][1]) # highest value last
else:
return '(null)'
2009-12-13 14:00:39 +00:00
# TODO: implement this in C? dunno how it should be called...
def getVertsFromGroup(me, group_index):
ret = []
2010-08-18 03:42:26 +00:00
for i, v in enumerate(me.vertices):
2009-12-13 14:00:39 +00:00
for g in v.groups:
if g.group == group_index:
ret.append((i, g.weight))
[#18694] Added support for Wavefront OBJ groups for both the import and export scripts from Paolo Ciccone (pciccone) Im not that keen on this functionality since vertex groups in blender are for weight painting and OBJ's groups are more like sub-objects, The conversion between these is a poor approximation (one vertex group per face). But this is still useful for using blender with poser so accepting these 2 patches. Made some changes from the patch *Export* - vgroup off by default, since this is mainly useful for corner cases - findVertexGroupName() - changed how exporting selects a vertex group for a face, use the highest combind weight for each group instead of the number of checking how many times the verts are in a group because they could have a zero weight. - findVertexGroupName() - would fail if a faces verts were not in any groups, now return a (null) group - Allow modifiers to export with vgroups *Import* - disable by default - ignore (null) groups - fixed a problem where the patch broke SPLIT_GROUPS - move the group definition out of global namespace, breaks importing as a module. --- Patch submission info The import/export scripts included with Blender 2.48 don't save/preserve groups defined in OBJ file. This presents a couple of issues: users loading an obj file with groups are not able to use a rather important set of data. When the file is then saved the groups are lost bringing up both the issue of data loss and loss of functionality as groups are used for rigging and morph target creation for Poser/DAZ Studio and all compatible applications. Of course other animation/modeling packages that use OBJ files might be affected. The solution that I present here simply patches the excellent python scripts written by Campbell Barton and Jiri Hnidek. The new code implements the preservation of groups by creating Blender vertex groups during import and by re-writing those groups in the OBJ file during export. Of course group ownership is idetermined at the face/polygon level and not at the vertex level. This rather simple solution makes Blender instantly one of the better tools to create Poser Morph targets (Full Body Morphs) and conforming figures and clothing. Of course it's also adding a more complete support for the whole OBJ spec. The patches are fitting well inside the established logic, have a minimal impact and are totally transparent to the user. See the attached video (about 7 minutes long) for more details. I noticed that the original scripts are formatted with tabs. I think it would be a good idea to follow the standard Python recommendation and convert the scripts to use spaces. I didn't want to do it on my side as it would generate a patch that replaces the whole file, not something that I wanted to do with my first submission :) But it would help if people with access to SVN could do it. Here is a short video that demonstrates the updates: http://www.paolociccone.com/blender_obj_scripts.html Cheers! ----
2009-05-13 00:20:14 +00:00
2009-12-13 14:00:39 +00:00
return ret
print('OBJ Export path: %r' % filepath)
2009-12-13 14:00:39 +00:00
temp_mesh_name = '~tmp-mesh'
time1 = time.clock()
# time1 = sys.time()
# scn = Scene.GetCurrent()
file = open(filepath, "w")
2009-12-13 14:00:39 +00:00
# Write Header
file.write('# Blender v%s OBJ File: %r\n' % (bpy.app.version_string, os.path.basename(bpy.data.filepath)))
file.write('# www.blender.org\n')
2009-12-13 14:00:39 +00:00
# Tell the obj file what material file to use.
if EXPORT_MTL:
mtlfilepath = os.path.splitext(filepath)[0] + ".mtl"
file.write('mtllib %s\n' % repr(os.path.basename(mtlfilepath))[1:-1]) # filepath can contain non utf8 chars, use repr
2009-12-13 14:00:39 +00:00
if EXPORT_ROTX90:
mat_xrot90= mathutils.Matrix.Rotation(-math.pi/2, 4, 'X')
2009-12-13 14:00:39 +00:00
# Initialize totals, these are updated each object
totverts = totuvco = totno = 1
face_vert_index = 1
globalNormals = {}
# A Dict of Materials
# (material.name, image.name):matname_imagename # matname_imagename has gaps removed.
mtl_dict = {}
2009-12-13 14:00:39 +00:00
# Get all meshes
for ob_main in objects:
# ignore dupli children
if ob_main.parent and ob_main.parent.dupli_type != 'NONE':
# XXX
print(ob_main.name, 'is a dupli child - ignoring')
continue
obs = []
if ob_main.dupli_type != 'NONE':
# XXX
print('creating dupli_list on', ob_main.name)
ob_main.create_dupli_list(scene)
2009-12-13 14:00:39 +00:00
obs = [(dob.object, dob.matrix) for dob in ob_main.dupli_list]
# XXX debug print
print(ob_main.name, 'has', len(obs), 'dupli children')
else:
obs = [(ob_main, ob_main.matrix_world)]
2009-12-13 14:00:39 +00:00
for ob, ob_mat in obs:
# Nurbs curve support
if EXPORT_CURVE_AS_NURBS and test_nurbs_compat(ob):
if EXPORT_ROTX90:
ob_mat = ob_mat * mat_xrot90
totverts += write_nurb(file, ob, ob_mat)
continue
# END NURBS
2009-12-13 14:00:39 +00:00
if ob.type != 'MESH':
continue
me = ob.create_mesh(scene, EXPORT_APPLY_MODIFIERS, 'PREVIEW')
2009-12-13 14:00:39 +00:00
if EXPORT_ROTX90:
me.transform(mat_xrot90 * ob_mat)
2009-12-13 14:00:39 +00:00
else:
me.transform(ob_mat)
# # Will work for non meshes now! :)
# me= BPyMesh.getMeshFromObject(ob, containerMesh, EXPORT_APPLY_MODIFIERS, EXPORT_POLYGROUPS, scn)
# if not me:
# continue
2009-12-13 14:00:39 +00:00
if EXPORT_UV:
faceuv = len(me.uv_textures) > 0
if faceuv:
uv_layer = me.uv_textures.active.data[:]
2009-12-13 14:00:39 +00:00
else:
faceuv = False
2010-08-18 03:42:26 +00:00
me_verts = me.vertices[:]
2010-08-15 15:04:51 +00:00
2009-12-13 14:00:39 +00:00
# XXX - todo, find a better way to do triangulation
# ...removed convert_to_triface because it relies on editmesh
'''
# We have a valid mesh
if EXPORT_TRI and me.faces:
# Add a dummy object to it.
has_quads = False
for f in me.faces:
2010-08-18 03:42:26 +00:00
if f.vertices[3] != 0:
2009-12-13 14:00:39 +00:00
has_quads = True
break
if has_quads:
newob = bpy.data.objects.new('temp_object', me)
2009-12-13 14:00:39 +00:00
# if we forget to set Object.data - crash
scene.objects.link(newob)
newob.convert_to_triface(scene)
# mesh will still be there
scene.objects.unlink(newob)
'''
# Make our own list so it can be sorted to reduce context switching
face_index_pairs = [ (face, index) for index, face in enumerate(me.faces)]
# faces = [ f for f in me.faces ]
if EXPORT_EDGES:
edges = me.edges
else:
edges = []
2010-08-18 03:42:26 +00:00
if not (len(face_index_pairs)+len(edges)+len(me.vertices)): # Make sure there is somthing to write
2009-12-13 14:00:39 +00:00
# clean up
bpy.data.meshes.remove(me)
2009-12-13 14:00:39 +00:00
continue # dont bother with this mesh.
# XXX
# High Quality Normals
if EXPORT_NORMALS and face_index_pairs:
me.calc_normals()
# if EXPORT_NORMALS_HQ:
# BPyMesh.meshCalcNormals(me)
# else:
# # transforming normals is incorrect
# # when the matrix is scaled,
# # better to recalculate them
# me.calcNormals()
2009-12-13 14:00:39 +00:00
materials = me.materials
materialNames = []
materialItems = [m for m in materials]
if materials:
for mat in materials:
if mat: # !=None
materialNames.append(mat.name)
else:
materialNames.append(None)
# Cant use LC because some materials are None.
# materialNames = map(lambda mat: mat.name, materials) # Bug Blender, dosent account for null materials, still broken.
# Possible there null materials, will mess up indicies
# but at least it will export, wait until Blender gets fixed.
materialNames.extend((16-len(materialNames)) * [None])
materialItems.extend((16-len(materialItems)) * [None])
# Sort by Material, then images
# so we dont over context switch in the obj file.
if EXPORT_KEEP_VERT_ORDER:
pass
elif faceuv:
face_index_pairs.sort(key=lambda a: (a[0].material_index, hash(uv_layer[a[1]].image), a[0].use_smooth))
2009-12-13 14:00:39 +00:00
elif len(materials) > 1:
face_index_pairs.sort(key = lambda a: (a[0].material_index, a[0].use_smooth))
2009-12-13 14:00:39 +00:00
else:
# no materials
face_index_pairs.sort(key = lambda a: a[0].use_smooth)
# if EXPORT_KEEP_VERT_ORDER:
# pass
# elif faceuv:
# try: faces.sort(key = lambda a: (a.mat, a.image, a.use_smooth))
# except: faces.sort(lambda a,b: cmp((a.mat, a.image, a.use_smooth), (b.mat, b.image, b.use_smooth)))
# elif len(materials) > 1:
# try: faces.sort(key = lambda a: (a.mat, a.use_smooth))
# except: faces.sort(lambda a,b: cmp((a.mat, a.use_smooth), (b.mat, b.use_smooth)))
# else:
# # no materials
# try: faces.sort(key = lambda a: a.use_smooth)
# except: faces.sort(lambda a,b: cmp(a.use_smooth, b.use_smooth))
2009-12-13 14:00:39 +00:00
# Set the default mat to no material and no image.
contextMat = (0, 0) # Can never be this, so we will label a new material teh first chance we get.
contextSmooth = None # Will either be true or false, set bad to force initialization switch.
if EXPORT_BLEN_OBS or EXPORT_GROUP_BY_OB:
name1 = ob.name
name2 = ob.data.name
if name1 == name2:
obnamestring = fixName(name1)
else:
obnamestring = '%s_%s' % (fixName(name1), fixName(name2))
if EXPORT_BLEN_OBS:
file.write('o %s\n' % obnamestring) # Write Object name
else: # if EXPORT_GROUP_BY_OB:
file.write('g %s\n' % obnamestring)
# Vert
2010-08-15 15:04:51 +00:00
for v in me_verts:
2009-12-13 14:00:39 +00:00
file.write('v %.6f %.6f %.6f\n' % tuple(v.co))
# UV
if faceuv:
2010-08-15 15:04:51 +00:00
uv_face_mapping = [[0,0,0,0] for i in range(len(face_index_pairs))] # a bit of a waste for tri's :/
2009-12-13 14:00:39 +00:00
uv_dict = {} # could use a set() here
uv_layer = me.uv_textures.active.data
2009-12-13 14:00:39 +00:00
for f, f_index in face_index_pairs:
2010-08-15 15:04:51 +00:00
for uv_index, uv in enumerate(uv_layer[f_index].uv):
2009-12-13 14:00:39 +00:00
uvkey = veckey2d(uv)
try:
uv_face_mapping[f_index][uv_index] = uv_dict[uvkey]
except:
uv_face_mapping[f_index][uv_index] = uv_dict[uvkey] = len(uv_dict)
file.write('vt %.6f %.6f\n' % tuple(uv))
2009-12-13 14:00:39 +00:00
uv_unique_count = len(uv_dict)
# del uv, uvkey, uv_dict, f_index, uv_index
2009-12-13 14:00:39 +00:00
# Only need uv_unique_count and uv_face_mapping
# NORMAL, Smooth/Non smoothed.
if EXPORT_NORMALS:
2010-08-15 15:04:51 +00:00
for f, f_index in face_index_pairs:
if f.use_smooth:
2010-08-18 03:42:26 +00:00
for v_idx in f.vertices:
2010-08-15 15:04:51 +00:00
v = me_verts[v_idx]
2009-12-13 14:00:39 +00:00
noKey = veckey3d(v.normal)
if noKey not in globalNormals:
globalNormals[noKey] = totno
totno +=1
file.write('vn %.6f %.6f %.6f\n' % noKey)
else:
# Hard, 1 normal from the face.
noKey = veckey3d(f.normal)
if noKey not in globalNormals:
globalNormals[noKey] = totno
totno +=1
file.write('vn %.6f %.6f %.6f\n' % noKey)
if not faceuv:
f_image = None
# XXX
if EXPORT_POLYGROUPS:
# Retrieve the list of vertex groups
# vertGroupNames = me.getVertGroupNames()
[#18694] Added support for Wavefront OBJ groups for both the import and export scripts from Paolo Ciccone (pciccone) Im not that keen on this functionality since vertex groups in blender are for weight painting and OBJ's groups are more like sub-objects, The conversion between these is a poor approximation (one vertex group per face). But this is still useful for using blender with poser so accepting these 2 patches. Made some changes from the patch *Export* - vgroup off by default, since this is mainly useful for corner cases - findVertexGroupName() - changed how exporting selects a vertex group for a face, use the highest combind weight for each group instead of the number of checking how many times the verts are in a group because they could have a zero weight. - findVertexGroupName() - would fail if a faces verts were not in any groups, now return a (null) group - Allow modifiers to export with vgroups *Import* - disable by default - ignore (null) groups - fixed a problem where the patch broke SPLIT_GROUPS - move the group definition out of global namespace, breaks importing as a module. --- Patch submission info The import/export scripts included with Blender 2.48 don't save/preserve groups defined in OBJ file. This presents a couple of issues: users loading an obj file with groups are not able to use a rather important set of data. When the file is then saved the groups are lost bringing up both the issue of data loss and loss of functionality as groups are used for rigging and morph target creation for Poser/DAZ Studio and all compatible applications. Of course other animation/modeling packages that use OBJ files might be affected. The solution that I present here simply patches the excellent python scripts written by Campbell Barton and Jiri Hnidek. The new code implements the preservation of groups by creating Blender vertex groups during import and by re-writing those groups in the OBJ file during export. Of course group ownership is idetermined at the face/polygon level and not at the vertex level. This rather simple solution makes Blender instantly one of the better tools to create Poser Morph targets (Full Body Morphs) and conforming figures and clothing. Of course it's also adding a more complete support for the whole OBJ spec. The patches are fitting well inside the established logic, have a minimal impact and are totally transparent to the user. See the attached video (about 7 minutes long) for more details. I noticed that the original scripts are formatted with tabs. I think it would be a good idea to follow the standard Python recommendation and convert the scripts to use spaces. I didn't want to do it on my side as it would generate a patch that replaces the whole file, not something that I wanted to do with my first submission :) But it would help if people with access to SVN could do it. Here is a short video that demonstrates the updates: http://www.paolociccone.com/blender_obj_scripts.html Cheers! ----
2009-05-13 00:20:14 +00:00
2009-12-13 14:00:39 +00:00
currentVGroup = ''
# Create a dictionary keyed by face id and listing, for each vertex, the vertex groups it belongs to
2010-08-15 15:04:51 +00:00
vgroupsMap = [[] for _i in range(len(me_verts))]
# vgroupsMap = [[] for _i in xrange(len(me_verts))]
2009-12-13 14:00:39 +00:00
for g in ob.vertex_groups:
# for vertexGroupName in vertGroupNames:
2010-08-15 15:04:51 +00:00
for v_idx, vWeight in getVertsFromGroup(me, g.index):
# for v_idx, vWeight in me.getVertsFromGroup(vertexGroupName, 1):
vgroupsMap[v_idx].append((g.name, vWeight))
[#18694] Added support for Wavefront OBJ groups for both the import and export scripts from Paolo Ciccone (pciccone) Im not that keen on this functionality since vertex groups in blender are for weight painting and OBJ's groups are more like sub-objects, The conversion between these is a poor approximation (one vertex group per face). But this is still useful for using blender with poser so accepting these 2 patches. Made some changes from the patch *Export* - vgroup off by default, since this is mainly useful for corner cases - findVertexGroupName() - changed how exporting selects a vertex group for a face, use the highest combind weight for each group instead of the number of checking how many times the verts are in a group because they could have a zero weight. - findVertexGroupName() - would fail if a faces verts were not in any groups, now return a (null) group - Allow modifiers to export with vgroups *Import* - disable by default - ignore (null) groups - fixed a problem where the patch broke SPLIT_GROUPS - move the group definition out of global namespace, breaks importing as a module. --- Patch submission info The import/export scripts included with Blender 2.48 don't save/preserve groups defined in OBJ file. This presents a couple of issues: users loading an obj file with groups are not able to use a rather important set of data. When the file is then saved the groups are lost bringing up both the issue of data loss and loss of functionality as groups are used for rigging and morph target creation for Poser/DAZ Studio and all compatible applications. Of course other animation/modeling packages that use OBJ files might be affected. The solution that I present here simply patches the excellent python scripts written by Campbell Barton and Jiri Hnidek. The new code implements the preservation of groups by creating Blender vertex groups during import and by re-writing those groups in the OBJ file during export. Of course group ownership is idetermined at the face/polygon level and not at the vertex level. This rather simple solution makes Blender instantly one of the better tools to create Poser Morph targets (Full Body Morphs) and conforming figures and clothing. Of course it's also adding a more complete support for the whole OBJ spec. The patches are fitting well inside the established logic, have a minimal impact and are totally transparent to the user. See the attached video (about 7 minutes long) for more details. I noticed that the original scripts are formatted with tabs. I think it would be a good idea to follow the standard Python recommendation and convert the scripts to use spaces. I didn't want to do it on my side as it would generate a patch that replaces the whole file, not something that I wanted to do with my first submission :) But it would help if people with access to SVN could do it. Here is a short video that demonstrates the updates: http://www.paolociccone.com/blender_obj_scripts.html Cheers! ----
2009-05-13 00:20:14 +00:00
for f, f_index in face_index_pairs:
2010-08-18 03:42:26 +00:00
f_v = [me_verts[v_idx] for v_idx in f.vertices]
2010-08-18 03:42:26 +00:00
# if f.vertices[3] == 0:
# f_v.pop()
# f_v= f.v
f_smooth= f.use_smooth
2009-12-13 14:00:39 +00:00
f_mat = min(f.material_index, len(materialNames)-1)
# f_mat = min(f.mat, len(materialNames)-1)
2009-12-13 14:00:39 +00:00
if faceuv:
2010-08-15 15:04:51 +00:00
tface = uv_layer[f_index]
2009-12-13 14:00:39 +00:00
f_image = tface.image
f_uv = tface.uv
# f_uv= [tface.uv1, tface.uv2, tface.uv3]
2010-08-18 03:42:26 +00:00
# if len(f.vertices) == 4:
# f_uv.append(tface.uv4)
# f_image = f.image
# f_uv= f.uv
2009-12-13 14:00:39 +00:00
# MAKE KEY
if faceuv and f_image: # Object is always true.
key = materialNames[f_mat], f_image.name
2009-12-13 14:00:39 +00:00
else:
key = materialNames[f_mat], None # No image, use None instead.
2009-12-13 14:00:39 +00:00
# Write the vertex group
if EXPORT_POLYGROUPS:
if len(ob.vertex_groups):
# find what vertext group the face belongs to
theVGroup = findVertexGroupName(f,vgroupsMap)
if theVGroup != currentVGroup:
2009-12-13 14:00:39 +00:00
currentVGroup = theVGroup
file.write('g %s\n' % theVGroup)
# # Write the vertex group
# if EXPORT_POLYGROUPS:
# if vertGroupNames:
# # find what vertext group the face belongs to
# theVGroup = findVertexGroupName(f,vgroupsMap)
# if theVGroup != currentVGroup:
# currentVGroup = theVGroup
# file.write('g %s\n' % theVGroup)
[#18694] Added support for Wavefront OBJ groups for both the import and export scripts from Paolo Ciccone (pciccone) Im not that keen on this functionality since vertex groups in blender are for weight painting and OBJ's groups are more like sub-objects, The conversion between these is a poor approximation (one vertex group per face). But this is still useful for using blender with poser so accepting these 2 patches. Made some changes from the patch *Export* - vgroup off by default, since this is mainly useful for corner cases - findVertexGroupName() - changed how exporting selects a vertex group for a face, use the highest combind weight for each group instead of the number of checking how many times the verts are in a group because they could have a zero weight. - findVertexGroupName() - would fail if a faces verts were not in any groups, now return a (null) group - Allow modifiers to export with vgroups *Import* - disable by default - ignore (null) groups - fixed a problem where the patch broke SPLIT_GROUPS - move the group definition out of global namespace, breaks importing as a module. --- Patch submission info The import/export scripts included with Blender 2.48 don't save/preserve groups defined in OBJ file. This presents a couple of issues: users loading an obj file with groups are not able to use a rather important set of data. When the file is then saved the groups are lost bringing up both the issue of data loss and loss of functionality as groups are used for rigging and morph target creation for Poser/DAZ Studio and all compatible applications. Of course other animation/modeling packages that use OBJ files might be affected. The solution that I present here simply patches the excellent python scripts written by Campbell Barton and Jiri Hnidek. The new code implements the preservation of groups by creating Blender vertex groups during import and by re-writing those groups in the OBJ file during export. Of course group ownership is idetermined at the face/polygon level and not at the vertex level. This rather simple solution makes Blender instantly one of the better tools to create Poser Morph targets (Full Body Morphs) and conforming figures and clothing. Of course it's also adding a more complete support for the whole OBJ spec. The patches are fitting well inside the established logic, have a minimal impact and are totally transparent to the user. See the attached video (about 7 minutes long) for more details. I noticed that the original scripts are formatted with tabs. I think it would be a good idea to follow the standard Python recommendation and convert the scripts to use spaces. I didn't want to do it on my side as it would generate a patch that replaces the whole file, not something that I wanted to do with my first submission :) But it would help if people with access to SVN could do it. Here is a short video that demonstrates the updates: http://www.paolociccone.com/blender_obj_scripts.html Cheers! ----
2009-05-13 00:20:14 +00:00
2009-12-13 14:00:39 +00:00
# CHECK FOR CONTEXT SWITCH
if key == contextMat:
pass # Context already switched, dont do anything
2009-12-13 14:00:39 +00:00
else:
if key[0] == None and key[1] == None:
# Write a null material, since we know the context has changed.
if EXPORT_GROUP_BY_MAT:
# can be mat_image or (null)
file.write('g %s_%s\n' % (fixName(ob.name), fixName(ob.data.name)) ) # can be mat_image or (null)
file.write('usemtl (null)\n') # mat, image
else:
mat_data= mtl_dict.get(key)
2009-12-13 14:00:39 +00:00
if not mat_data:
# First add to global dict so we can export to mtl
# Then write mtl
# Make a new names from the mat and image name,
# converting any spaces to underscores with fixName.
# If none image dont bother adding it to the name
if key[1] == None:
mat_data = mtl_dict[key] = ('%s'%fixName(key[0])), materialItems[f_mat], f_image
2009-12-13 14:00:39 +00:00
else:
mat_data = mtl_dict[key] = ('%s_%s' % (fixName(key[0]), fixName(key[1]))), materialItems[f_mat], f_image
2009-12-13 14:00:39 +00:00
if EXPORT_GROUP_BY_MAT:
file.write('g %s_%s_%s\n' % (fixName(ob.name), fixName(ob.data.name), mat_data[0]) ) # can be mat_image or (null)
file.write('usemtl %s\n' % mat_data[0]) # can be mat_image or (null)
contextMat = key
if f_smooth != contextSmooth:
if f_smooth: # on now off
file.write('s 1\n')
contextSmooth = f_smooth
else: # was off now on
file.write('s off\n')
contextSmooth = f_smooth
file.write('f')
if faceuv:
if EXPORT_NORMALS:
if f_smooth: # Smoothed, use vertex normals
for vi, v in enumerate(f_v):
file.write( ' %d/%d/%d' % \
2010-08-15 15:04:51 +00:00
(v.index + totverts,
2009-12-13 14:00:39 +00:00
totuvco + uv_face_mapping[f_index][vi],
2010-08-15 15:04:51 +00:00
globalNormals[ veckey3d(v.normal) ]) ) # vert, uv, normal
2009-12-13 14:00:39 +00:00
else: # No smoothing, face normals
no = globalNormals[ veckey3d(f.normal) ]
for vi, v in enumerate(f_v):
file.write( ' %d/%d/%d' % \
2010-08-15 15:04:51 +00:00
(v.index + totverts,
2009-12-13 14:00:39 +00:00
totuvco + uv_face_mapping[f_index][vi],
no) ) # vert, uv, normal
else: # No Normals
for vi, v in enumerate(f_v):
file.write( ' %d/%d' % (\
2010-08-15 15:04:51 +00:00
v.index + totverts,\
2009-12-13 14:00:39 +00:00
totuvco + uv_face_mapping[f_index][vi])) # vert, uv
face_vert_index += len(f_v)
else: # No UV's
if EXPORT_NORMALS:
if f_smooth: # Smoothed, use vertex normals
for v in f_v:
file.write( ' %d//%d' %
2010-08-15 15:04:51 +00:00
(v.index + totverts, globalNormals[ veckey3d(v.normal) ]) )
2009-12-13 14:00:39 +00:00
else: # No smoothing, face normals
no = globalNormals[ veckey3d(f.normal) ]
for v in f_v:
2010-08-15 15:04:51 +00:00
file.write( ' %d//%d' % (v.index + totverts, no) )
2009-12-13 14:00:39 +00:00
else: # No Normals
for v in f_v:
2010-08-15 15:04:51 +00:00
file.write( ' %d' % (v.index + totverts) )
2009-12-13 14:00:39 +00:00
file.write('\n')
# Write edges.
if EXPORT_EDGES:
for ed in edges:
if ed.is_loose:
2010-08-18 03:42:26 +00:00
file.write('f %d %d\n' % (ed.vertices[0] + totverts, ed.vertices[1] + totverts))
2009-12-13 14:00:39 +00:00
# Make the indicies global rather then per mesh
2010-08-15 15:04:51 +00:00
totverts += len(me_verts)
2009-12-13 14:00:39 +00:00
if faceuv:
totuvco += uv_unique_count
# clean up
bpy.data.meshes.remove(me)
2009-12-13 14:00:39 +00:00
if ob_main.dupli_type != 'NONE':
ob_main.free_dupli_list()
file.close()
# Now we have all our materials, save them
if EXPORT_MTL:
write_mtl(scene, mtlfilepath, EXPORT_COPY_IMAGES, mtl_dict)
# if EXPORT_COPY_IMAGES:
# dest_dir = os.path.basename(filepath)
# # dest_dir = filepath
# # # Remove chars until we are just the path.
# # while dest_dir and dest_dir[-1] not in '\\/':
# # dest_dir = dest_dir[:-1]
# if dest_dir:
# copy_images(dest_dir, mtl_dict)
# else:
# print('\tError: "%s" could not be used as a base for an image path.' % filepath)
2009-12-13 14:00:39 +00:00
print("OBJ Export time: %.2f" % (time.clock() - time1))
#
def _write(context, filepath,
EXPORT_TRI, # ok
EXPORT_EDGES,
EXPORT_NORMALS, # not yet
EXPORT_NORMALS_HQ, # not yet
EXPORT_UV, # ok
EXPORT_MTL,
EXPORT_COPY_IMAGES,
EXPORT_APPLY_MODIFIERS, # ok
EXPORT_ROTX90, # wrong
EXPORT_BLEN_OBS,
EXPORT_GROUP_BY_OB,
EXPORT_GROUP_BY_MAT,
EXPORT_KEEP_VERT_ORDER,
EXPORT_POLYGROUPS,
EXPORT_CURVE_AS_NURBS,
EXPORT_SEL_ONLY, # ok
EXPORT_ALL_SCENES, # XXX not working atm
EXPORT_ANIMATION): # Not used
2010-08-15 15:04:51 +00:00
base_name, ext = os.path.splitext(filepath)
2009-12-13 14:00:39 +00:00
context_name = [base_name, '', '', ext] # Base name, scene name, frame number, extension
orig_scene = context.scene
# Exit edit mode before exporting, so current object states are exported properly.
if bpy.ops.object.mode_set.poll():
bpy.ops.object.mode_set(mode='OBJECT')
# if EXPORT_ALL_SCENES:
# export_scenes = bpy.data.scenes
# else:
# export_scenes = [orig_scene]
2009-12-13 14:00:39 +00:00
# XXX only exporting one scene atm since changing
# current scene is not possible.
# Brecht says that ideally in 2.5 we won't need such a function,
# allowing multiple scenes open at once.
export_scenes = [orig_scene]
# Export all scenes.
for scene in export_scenes:
# scene.makeCurrent() # If already current, this is not slow.
# context = scene.getRenderingContext()
orig_frame = scene.frame_current
2009-12-13 14:00:39 +00:00
if EXPORT_ALL_SCENES: # Add scene name into the context_name
context_name[1] = '_%s' % bpy.path.clean_name(scene.name) # WARNING, its possible that this could cause a collision. we could fix if were feeling parranoied.
2009-12-13 14:00:39 +00:00
# Export an animation?
if EXPORT_ANIMATION:
scene_frames = range(scene.frame_start, context.frame_end + 1) # Up to and including the end frame.
2009-12-13 14:00:39 +00:00
else:
scene_frames = [orig_frame] # Dont export an animation.
# Loop through all frames in the scene and export.
for frame in scene_frames:
if EXPORT_ANIMATION: # Add frame to the filepath.
2009-12-13 14:00:39 +00:00
context_name[2] = '_%.6d' % frame
scene.frame_current = frame
2009-12-13 14:00:39 +00:00
if EXPORT_SEL_ONLY:
objects = context.selected_objects
2009-12-13 14:00:39 +00:00
else:
objects = scene.objects
2009-12-13 14:00:39 +00:00
full_path= ''.join(context_name)
# erm... bit of a problem here, this can overwrite files when exporting frames. not too bad.
# EXPORT THE FILE.
write_file(full_path, objects, scene,
EXPORT_TRI,
EXPORT_EDGES,
EXPORT_NORMALS,
EXPORT_NORMALS_HQ,
EXPORT_UV,
EXPORT_MTL,
EXPORT_COPY_IMAGES,
EXPORT_APPLY_MODIFIERS,
EXPORT_ROTX90,
EXPORT_BLEN_OBS,
EXPORT_GROUP_BY_OB,
EXPORT_GROUP_BY_MAT,
EXPORT_KEEP_VERT_ORDER,
EXPORT_POLYGROUPS,
EXPORT_CURVE_AS_NURBS)
scene.frame_current = orig_frame
2009-12-13 14:00:39 +00:00
# Restore old active scene.
# orig_scene.makeCurrent()
# Window.WaitCursor(0)
2009-12-13 14:00:39 +00:00
'''
Currently the exporter lacks these features:
* multiple scene export (only active scene is written)
* particles
'''
def save(operator, context, filepath="",
use_triangles=False,
use_edges=False,
use_normals=False,
use_hq_normals=False,
use_uvs=True,
use_materials=True,
copy_images=False,
use_modifiers=True,
use_rotate_x90=True,
use_blen_objects=True,
group_by_object=False,
group_by_material=False,
keep_vertex_order=False,
use_vertex_groups=False,
use_nurbs=True,
use_selection=True,
use_all_scenes=False,
use_animation=False,
):
_write(context, filepath,
EXPORT_TRI=use_triangles,
EXPORT_EDGES=use_edges,
EXPORT_NORMALS=use_normals,
EXPORT_NORMALS_HQ=use_hq_normals,
EXPORT_UV=use_uvs,
EXPORT_MTL=use_materials,
EXPORT_COPY_IMAGES=copy_images,
EXPORT_APPLY_MODIFIERS=use_modifiers,
EXPORT_ROTX90=use_rotate_x90,
EXPORT_BLEN_OBS=use_blen_objects,
EXPORT_GROUP_BY_OB=group_by_object,
EXPORT_GROUP_BY_MAT=group_by_material,
EXPORT_KEEP_VERT_ORDER=keep_vertex_order,
EXPORT_POLYGROUPS=use_vertex_groups,
EXPORT_CURVE_AS_NURBS=use_nurbs,
EXPORT_SEL_ONLY=use_selection,
EXPORT_ALL_SCENES=use_all_scenes,
EXPORT_ANIMATION=use_animation,
)
return {'FINISHED'}