Newer
Older
# ##### 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.
#
# 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.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software Foundation,
# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
#
# ##### END GPL LICENSE BLOCK #####
# <pep8 compliant>
# Script copyright (C) Campbell Barton
# Contributors: Campbell Barton, Jiri Hnidek, Paolo Ciccone
"""
This script imports a Wavefront OBJ files to Blender.
Usage:
Run this script from "File->Import" menu and then load the desired OBJ file.
Note, This loads mesh objects and materials only, nurbs and curves are not supported.
http://wiki.blender.org/index.php/Scripts/Manual/Import/wavefront_obj
"""
import os
import time
import bpy
import mathutils
from bpy_extras.image_utils import load_image
from progress_report import ProgressReport, ProgressReportSubstep
def line_value(line_split):
Returns 1 string representing the value for this line
None will be returned if theres only 1 word
if length == 1:
return None
elif length == 2:
return line_split[1]
elif length > 2:
return b' '.join(line_split[1:])
def obj_image_load(context_imagepath_map, line, DIR, recursive, relpath):
But we try all space-separated items from current line when file is not found with last one
(users keep generating/using image files with spaces in a format that does not support them, sigh...)
Also tries to replace '_' with ' ' for Max's exporter replaces spaces with underscores.
filepath_parts = line.split(b' ')
image = None
for i in range(-1, -len(filepath_parts), -1):
imagepath = os.fsdecode(b" ".join(filepath_parts[i:]))
image = context_imagepath_map.get(imagepath, ...)
if image is ...:
image = load_image(imagepath, DIR, recursive=recursive, relpath=relpath)
if image is None and "_" in imagepath:
image = load_image(imagepath.replace("_", " "), DIR, recursive=recursive, relpath=relpath)
if image is not None:
context_imagepath_map[imagepath] = image
break;
if image is None:
imagepath = os.fsdecode(filepath_parts[-1])
image = load_image(imagepath, DIR, recursive=recursive, place_holder=True, relpath=relpath)
context_imagepath_map[imagepath] = image
return image
Campbell Barton
committed
def create_materials(filepath, relpath,
material_libs, unique_materials, unique_material_images,
use_image_search, use_cycles, float_func):
Create all the used materials in this obj,
assign colors and images to the materials from all referenced material libs
Campbell Barton
committed
context_material_vars = set()
# Don't load the same image multiple times
context_imagepath_map = {}
cycles_material_wrap_map = {}
def load_material_image(blender_material, mat_wrap, use_cycles, context_material_name, img_data, line, type):
"""
Set textures defined in .mtl file.
"""
map_options = {}
curr_token = []
for token in img_data[:-1]:
if token.startswith(b'-'):
if curr_token:
map_options[curr_token[0]] = curr_token[1:]
curr_token[:] = []
curr_token.append(token)
Bastien Montagne
committed
if curr_token:
map_options[curr_token[0]] = curr_token[1:]
# Absolute path - c:\.. etc would work here
image = obj_image_load(context_imagepath_map, line, DIR, use_image_search, relpath)
texture = bpy.data.textures.new(name=type, type='IMAGE')
texture.image = image
map_offset = map_options.get(b'-o')
map_scale = map_options.get(b'-s')
# Adds textures for materials (rendering)
if type == 'Kd':
if use_cycles:
mat_wrap.diffuse_image_set(image)
mat_wrap.diffuse_mapping_set(coords='UV', translation=map_offset, scale=map_scale)
Campbell Barton
committed
mtex = blender_material.texture_slots.add()
mtex.texture = texture
mtex.texture_coords = 'UV'
mtex.use_map_color_diffuse = True
# adds textures to faces (Textured/Alt-Z mode)
# Only apply the diffuse texture to the face if the image has not been set with the inline usemat func.
Campbell Barton
committed
unique_material_images[context_material_name] = image # set the texface image
elif type == 'Ka':
if use_cycles:
# XXX Not supported?
print("WARNING, currently unsupported ambient texture, skipped.")
mtex = blender_material.texture_slots.add()
mtex.use_map_color_diffuse = False
mtex.texture = texture
mtex.texture_coords = 'UV'
mtex.use_map_ambient = True
elif type == 'Ks':
if use_cycles:
mat_wrap.specular_image_set(image)
mat_wrap.specular_mapping_set(coords='UV', translation=map_offset, scale=map_scale)
mtex = blender_material.texture_slots.add()
mtex.use_map_color_diffuse = False
mtex.texture = texture
mtex.texture_coords = 'UV'
mtex.use_map_color_spec = True
Bastien Montagne
committed
elif type == 'Ke':
if use_cycles:
# XXX Not supported?
print("WARNING, currently unsupported emit texture, skipped.")
Bastien Montagne
committed
mtex = blender_material.texture_slots.add()
mtex.use_map_color_diffuse = False
mtex.texture = texture
mtex.texture_coords = 'UV'
mtex.use_map_emit = True
elif type == 'Bump':
bump_mult = map_options.get(b'-bm')
Bastien Montagne
committed
bump_mult = float(bump_mult[0]) if (bump_mult is not None and len(bump_mult) > 1) else 1.0
if use_cycles:
mat_wrap.normal_image_set(image)
mat_wrap.normal_mapping_set(coords='UV', translation=map_offset, scale=map_scale)
if bump_mult:
Bastien Montagne
committed
mat_wrap.normal_factor_set(bump_mult)
mtex = blender_material.texture_slots.add()
mtex.use_map_color_diffuse = False
mtex.texture = texture
mtex.texture_coords = 'UV'
mtex.use_map_normal = True
if bump_mult:
Bastien Montagne
committed
mtex.normal_factor = bump_mult
elif type == 'D':
if use_cycles:
mat_wrap.alpha_image_set(image)
mat_wrap.alpha_mapping_set(coords='UV', translation=map_offset, scale=map_scale)
mtex = blender_material.texture_slots.add()
mtex.use_map_color_diffuse = False
mtex.texture = texture
mtex.texture_coords = 'UV'
mtex.use_map_alpha = True
blender_material.use_transparency = True
blender_material.transparency_method = 'Z_TRANSPARENCY'
Campbell Barton
committed
if "alpha" not in context_material_vars:
blender_material.alpha = 0.0
# Todo, unset diffuse material alpha if it has an alpha channel
Campbell Barton
committed
elif type == 'disp':
if use_cycles:
mat_wrap.bump_image_set(image)
mat_wrap.bump_mapping_set(coords='UV', translation=map_offset, scale=map_scale)
Campbell Barton
committed
mtex = blender_material.texture_slots.add()
mtex.use_map_color_diffuse = False
mtex.texture = texture
mtex.texture_coords = 'UV'
mtex.use_map_displacement = True
elif type == 'refl':
map_type = map_options.get(b'-type')
if map_type and map_type != [b'sphere']:
print("WARNING, unsupported reflection type '%s', defaulting to 'sphere'"
"" % ' '.join(i.decode() for i in map_type))
if use_cycles:
mat_wrap.diffuse_image_set(image, projection='SPHERE')
mat_wrap.diffuse_mapping_set(coords='Reflection', translation=map_offset, scale=map_scale)
mtex = blender_material.texture_slots.add()
mtex.use_map_color_diffuse = False
mtex.texture = texture
mtex.texture_coords = 'REFLECTION'
mtex.use_map_color_diffuse = True
mtex.mapping = 'SPHERE'
raise Exception("invalid type %r" % type)
if map_offset:
mtex.offset.x = float(map_offset[0])
if len(map_offset) >= 2:
mtex.offset.y = float(map_offset[1])
if len(map_offset) >= 3:
mtex.offset.z = float(map_offset[2])
if map_scale:
mtex.scale.x = float(map_scale[0])
if len(map_scale) >= 2:
mtex.scale.y = float(map_scale[1])
if len(map_scale) >= 3:
mtex.scale.z = float(map_scale[2])
# Add an MTL with the same name as the obj if no MTLs are spesified.
temp_mtl = os.path.splitext((os.path.basename(filepath)))[0] + ".mtl"
if os.path.exists(os.path.join(DIR, temp_mtl)):
material_libs.add(temp_mtl)
del temp_mtl
ma = unique_materials[name] = bpy.data.materials.new(name.decode('utf-8', "replace"))
unique_material_images[name] = None # assign None to all material images to start with, add to later.
if use_cycles:
from modules import cycles_shader_compat
ma_wrap = cycles_shader_compat.CyclesShaderWrapper(ma)
cycles_material_wrap_map[ma] = ma_wrap
# XXX Why was this needed? Cannot find any good reason, and adds stupid empty matslot in case we do not separate
# mesh (see T44947).
#~ unique_materials[None] = None
#~ unique_material_images[None] = None
for libname in sorted(material_libs):
if not os.path.exists(mtlpath):
print("\tMaterial not found MTL: %r" % mtlpath)
else:
do_ambient = True
do_highlight = False
do_reflection = False
do_transparency = False
do_glass = False
do_fresnel = False
do_raytrace = False
Bastien Montagne
committed
emit_colors = [0.0, 0.0, 0.0]
# print('\t\tloading mtl: %e' % mtlpath)
context_mat_wrap = None
mtl = open(mtlpath, 'rb')
Campbell Barton
committed
line = line.strip()
if not line or line.startswith(b'#'):
continue
line_split = line.split()
line_id = line_split[0].lower()
if line_id == b'newmtl':
Bastien Montagne
committed
# Finalize previous mat, if any.
Bastien Montagne
committed
emit_value = sum(emit_colors) / 3.0
if emit_value > 1e-6:
if use_cycles:
print("WARNING, currently unsupported emit value, skipped.")
Bastien Montagne
committed
# We have to adapt it to diffuse color too...
emit_value /= sum(context_material.diffuse_color) / 3.0
context_material.emit = emit_value
if not do_ambient:
context_material.ambient = 0.0
if do_highlight:
context_mat_wrap.hardness_value_set(1.0)
# FIXME, how else to use this?
context_material.specular_intensity = 1.0
else:
if use_cycles:
context_mat_wrap.hardness_value_set(0.0)
context_mat_wrap.reflect_factor_set(1.0)
context_material.raytrace_mirror.use = True
context_material.raytrace_mirror.reflect_factor = 1.0
if do_transparency:
context_material.use_transparency = True
context_material.transparency_method = 'RAYTRACE' if do_raytrace else 'Z_TRANSPARENCY'
if "alpha" not in context_material_vars:
if use_cycles:
context_mat_wrap.alpha_value_set(0.0)
context_material.alpha = 0.0
if do_glass:
if use_cycles:
print("WARNING, currently unsupported glass material, skipped.")
if "ior" not in context_material_vars:
context_material.raytrace_transparency.ior = 1.5
if do_fresnel:
if use_cycles:
print("WARNING, currently unsupported fresnel option, skipped.")
context_material.raytrace_mirror.fresnel = 1.0 # could be any value for 'ON'
"""
if do_raytrace:
context_material.use_raytrace = True
else:
context_material.use_raytrace = False
"""
# XXX, this is not following the OBJ spec, but this was
# written when raytracing wasnt default, annoying to disable for blender users.
context_material.use_raytrace = True
context_material_name = line_value(line_split)
context_material = unique_materials.get(context_material_name)
if use_cycles and context_material is not None:
context_mat_wrap = cycles_material_wrap_map[context_material]
Campbell Barton
committed
context_material_vars.clear()
Bastien Montagne
committed
emit_colors[:] = [0.0, 0.0, 0.0]
do_ambient = True
do_highlight = False
do_reflection = False
do_transparency = False
do_glass = False
do_fresnel = False
do_raytrace = False
elif context_material:
# we need to make a material to assign properties to it.
if line_id == b'ka':
col = (float_func(line_split[1]), float_func(line_split[2]), float_func(line_split[3]))
if use_cycles:
context_mat_wrap.reflect_color_set(col)
context_material.mirror_color = col
# This is highly approximated, but let's try to stick as close from exporter as possible... :/
context_material.ambient = sum(context_material.mirror_color) / 3
elif line_id == b'kd':
col = (float_func(line_split[1]), float_func(line_split[2]), float_func(line_split[3]))
if use_cycles:
context_mat_wrap.diffuse_color_set(col)
context_material.diffuse_color = col
context_material.diffuse_intensity = 1.0
elif line_id == b'ks':
col = (float_func(line_split[1]), float_func(line_split[2]), float_func(line_split[3]))
if use_cycles:
context_mat_wrap.specular_color_set(col)
context_mat_wrap.hardness_value_set(1.0)
context_material.specular_color = col
context_material.specular_intensity = 1.0
Bastien Montagne
committed
elif line_id == b'ke':
# We cannot set context_material.emit right now, we need final diffuse color as well for this.
emit_colors[:] = [
float_func(line_split[1]), float_func(line_split[2]), float_func(line_split[3])]
elif line_id == b'ns':
if use_cycles:
context_mat_wrap.hardness_value_set(((float_func(line_split[1]) + 3.0) / 50.0) - 0.65)
context_material.specular_hardness = int((float_func(line_split[1]) * 0.51) + 1)
elif line_id == b'ni': # Refraction index (between 1 and 3).
if use_cycles:
print("WARNING, currently unsupported glass material, skipped.")
context_material.raytrace_transparency.ior = max(1, min(float_func(line_split[1]), 3))
Campbell Barton
committed
context_material_vars.add("ior")
elif line_id == b'd': # dissolve (transparency)
if use_cycles:
context_mat_wrap.alpha_value_set(float_func(line_split[1]))
Campbell Barton
committed
context_material.alpha = float_func(line_split[1])
context_material.use_transparency = True
context_material.transparency_method = 'Z_TRANSPARENCY'
Campbell Barton
committed
context_material_vars.add("alpha")
elif line_id == b'tr': # translucency
if use_cycles:
print("WARNING, currently unsupported translucency option, skipped.")
context_material.translucency = float_func(line_split[1])
elif line_id == b'tf':
Campbell Barton
committed
# rgb, filter color, blender has no support for this.
pass
elif line_id == b'illum':
Campbell Barton
committed
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
illum = int(line_split[1])
# inline comments are from the spec, v4.2
if illum == 0:
# Color on and Ambient off
do_ambient = False
elif illum == 1:
# Color on and Ambient on
pass
elif illum == 2:
# Highlight on
do_highlight = True
elif illum == 3:
# Reflection on and Ray trace on
do_reflection = True
do_raytrace = True
elif illum == 4:
# Transparency: Glass on
# Reflection: Ray trace on
do_transparency = True
do_reflection = True
do_glass = True
do_raytrace = True
elif illum == 5:
# Reflection: Fresnel on and Ray trace on
do_reflection = True
do_fresnel = True
do_raytrace = True
elif illum == 6:
# Transparency: Refraction on
# Reflection: Fresnel off and Ray trace on
do_transparency = True
do_reflection = True
do_raytrace = True
elif illum == 7:
# Transparency: Refraction on
# Reflection: Fresnel on and Ray trace on
do_transparency = True
do_reflection = True
do_fresnel = True
do_raytrace = True
elif illum == 8:
# Reflection on and Ray trace off
do_reflection = True
elif illum == 9:
# Transparency: Glass on
# Reflection: Ray trace off
do_transparency = True
do_reflection = True
do_glass = True
elif illum == 10:
# Casts shadows onto invisible surfaces
Campbell Barton
committed
pass
elif line_id == b'map_ka':
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'Ka')
elif line_id == b'map_ks':
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'Ks')
elif line_id == b'map_kd':
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'Kd')
Bastien Montagne
committed
elif line_id == b'map_ke':
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'Ke')
elif line_id in {b'map_bump', b'bump'}: # 'bump' is incorrect but some files use it.
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'Bump')
elif line_id in {b'map_d', b'map_tr'}: # Alpha map - Dissolve
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'D')
elif line_id in {b'map_disp', b'disp'}: # displacementmap
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'disp')
Campbell Barton
committed
elif line_id in {b'map_refl', b'refl'}: # reflectionmap
img_data = line.split()[1:]
if img_data:
load_material_image(context_material, context_mat_wrap, use_cycles,
context_material_name, img_data, line, 'refl')
Campbell Barton
committed
else:
print("\t%r:%r (ignored)" % (filepath, line))
mtl.close()
def split_mesh(verts_loc, faces, unique_materials, filepath, SPLIT_OB_OR_GROUP):
Takes vert_loc and faces, and separates into multiple sets of
(verts_loc, faces, unique_materials, dataname)
filename = os.path.splitext((os.path.basename(filepath)))[0]
Bastien Montagne
committed
if not SPLIT_OB_OR_GROUP or not faces:
use_verts_nor = any((False if f[1] is ... else True) for f in faces)
use_verts_tex = any((False if f[2] is ... else True) for f in faces)
Bastien Montagne
committed
# use the filename for the object name since we aren't chopping up the mesh.
return [(verts_loc, faces, unique_materials, filename, use_verts_nor, use_verts_tex)]
def key_to_name(key):
# if the key is a tuple, join it to make a string
if not key:
return filename # assume its a string. make sure this is true if the splitting code is changed
else:
return key.decode('utf-8', 'replace')
# Return a key that makes the faces unique.
oldkey = -1 # initialize to a value that will never match the key
for face in faces:
if oldkey != key:
# Check the key has changed.
Bastien Montagne
committed
(verts_split, faces_split, unique_materials_split, vert_remap,
use_verts_nor, use_verts_tex) = face_split_dict.setdefault(key, ([], [], {}, {}, [], []))
if not use_verts_nor and face[1] is not ...:
Bastien Montagne
committed
use_verts_nor.append(True)
if not use_verts_tex and face[2] is not ...:
Bastien Montagne
committed
use_verts_tex.append(True)
# Remap verts to new vert list and add where needed
for enum, i in enumerate(face_vert_loc_indices):
map_index = vert_remap.get(i)
if map_index is None:
map_index = len(verts_split)
vert_remap[i] = map_index # set the new remapped index so we only add once and can reference next time.
verts_split.append(verts_loc[i]) # add the vert to the local verts
face_vert_loc_indices[enum] = map_index # remap to the local index
if matname and matname not in unique_materials_split:
unique_materials_split[matname] = unique_materials[matname]
faces_split.append(face)
# remove one of the items and reorder
Bastien Montagne
committed
return [(verts_split, faces_split, unique_materials_split, key_to_name(key), bool(use_vnor), bool(use_vtex))
for key, (verts_split, faces_split, unique_materials_split, _, use_vnor, use_vtex)
in face_split_dict.items()]
def create_mesh(new_objects,
verts_tex,
faces,
unique_materials,
unique_material_images,
unique_smooth_groups,
vertex_groups,
dataname,
):
Takes all the data gathered and generates a mesh, adding the new object to new_objects
deals with ngons, sharp edges and assigning materials
if unique_smooth_groups:
sharp_edges = set()
smooth_group_users = {context_smooth_group: {} for context_smooth_group in unique_smooth_groups.keys()}
fgon_edges = set() # Used for storing fgon keys when we need to tesselate/untesselate them (ngons with hole).
edges = []
tot_loops = 0
# reverse loop through face indices
face_vert_nor_indices,
face_vert_tex_indices,
context_material,
context_smooth_group,
context_object,
Bastien Montagne
committed
face_invalid_blenpoly,
len_face_vert_loc_indices = len(face_vert_loc_indices)
if len_face_vert_loc_indices == 1:
faces.pop(f_idx) # cant add single vert faces
Bastien Montagne
committed
# Face with a single item in face_vert_nor_indices is actually a polyline!
elif len(face_vert_nor_indices) == 1 or len_face_vert_loc_indices == 2:
edges.extend((face_vert_loc_indices[i], face_vert_loc_indices[i + 1])
for i in range(len_face_vert_loc_indices - 1))
faces.pop(f_idx)
# Smooth Group
if unique_smooth_groups and context_smooth_group:
# Is a part of of a smooth group and is a face
if context_smooth_group_old is not context_smooth_group:
edge_dict = smooth_group_users[context_smooth_group]
context_smooth_group_old = context_smooth_group
Bastien Montagne
committed
prev_vidx = face_vert_loc_indices[-1]
for vidx in face_vert_loc_indices:
edge_key = (prev_vidx, vidx) if (prev_vidx < vidx) else (vidx, prev_vidx)
prev_vidx = vidx
edge_dict[edge_key] = edge_dict.get(edge_key, 0) + 1
Bastien Montagne
committed
# NGons into triangles
if face_invalid_blenpoly:
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
# ignore triangles with invalid indices
if len(face_vert_loc_indices) > 3:
from bpy_extras.mesh_utils import ngon_tessellate
ngon_face_indices = ngon_tessellate(verts_loc, face_vert_loc_indices)
faces.extend([([face_vert_loc_indices[ngon[0]],
face_vert_loc_indices[ngon[1]],
face_vert_loc_indices[ngon[2]],
],
[face_vert_nor_indices[ngon[0]],
face_vert_nor_indices[ngon[1]],
face_vert_nor_indices[ngon[2]],
] if face_vert_nor_indices else [],
[face_vert_tex_indices[ngon[0]],
face_vert_tex_indices[ngon[1]],
face_vert_tex_indices[ngon[2]],
] if face_vert_tex_indices else [],
context_material,
context_smooth_group,
context_object,
[],
)
for ngon in ngon_face_indices]
)
tot_loops += 3 * len(ngon_face_indices)
# edges to make ngons
if len(ngon_face_indices) > 1:
edge_users = set()
for ngon in ngon_face_indices:
prev_vidx = face_vert_loc_indices[ngon[-1]]
for ngidx in ngon:
vidx = face_vert_loc_indices[ngidx]
if vidx == prev_vidx:
continue # broken OBJ... Just skip.
edge_key = (prev_vidx, vidx) if (prev_vidx < vidx) else (vidx, prev_vidx)
prev_vidx = vidx
if edge_key in edge_users:
fgon_edges.add(edge_key)
else:
edge_users.add(edge_key)
Bastien Montagne
committed
faces.pop(f_idx)
else:
tot_loops += len_face_vert_loc_indices
# Build sharp edges
if unique_smooth_groups:
for edge_dict in smooth_group_users.values():
for key, users in edge_dict.items():
if users == 1: # This edge is on the boundry of a group
sharp_edges.add(key)
# map the material names to an index
material_mapping = {name: i for i, name in enumerate(unique_materials)} # enumerate over unique_materials keys()
for name, index in material_mapping.items():
me = bpy.data.meshes.new(dataname)
# make sure the list isnt too big
for material in materials:
me.materials.append(material)
me.vertices.add(len(verts_loc))
me.loops.add(tot_loops)
me.polygons.add(len(faces))
# verts_loc is a list of (x, y, z) tuples
me.vertices.foreach_set("co", unpack_list(verts_loc))
loops_vert_idx = []
faces_loop_start = []
faces_loop_total = []
lidx = 0
for f in faces:
vidx = f[0]
nbr_vidx = len(vidx)
loops_vert_idx.extend(vidx)
faces_loop_start.append(lidx)
faces_loop_total.append(nbr_vidx)
lidx += nbr_vidx
me.loops.foreach_set("vertex_index", loops_vert_idx)
me.polygons.foreach_set("loop_start", faces_loop_start)
me.polygons.foreach_set("loop_total", faces_loop_total)
# Note: we store 'temp' normals in loops, since validate() may alter final mesh,
# we can only set custom lnors *after* calling it.
me.create_normals_split()
if verts_tex and me.polygons:
me.uv_textures.new()
context_material_old = -1 # avoid a dict lookup
mat = 0 # rare case it may be un-initialized.
for i, (face, blen_poly) in enumerate(zip(faces, me.polygons)):
if len(face[0]) < 3:
raise Exception("bad face") # Shall not happen, we got rid of those earlier!
(face_vert_loc_indices,
face_vert_nor_indices,
face_vert_tex_indices,
context_material,
context_smooth_group,
context_object,
Bastien Montagne
committed
face_invalid_blenpoly,
if context_smooth_group:
blen_poly.use_smooth = True
if context_material:
if context_material_old is not context_material:
mat = material_mapping[context_material]
context_material_old = context_material
blen_poly.material_index = mat
Campbell Barton
committed
Bastien Montagne
committed
if verts_nor and face_vert_nor_indices:
for face_noidx, lidx in zip(face_vert_nor_indices, blen_poly.loop_indices):
me.loops[lidx].normal[:] = verts_nor[0 if (face_noidx is ...) else face_noidx]
Bastien Montagne
committed
if verts_tex and face_vert_tex_indices:
Campbell Barton
committed
if context_material:
image = unique_material_images[context_material]
if image: # Can be none if the material dosnt have an image.
me.uv_textures[0].data[i].image = image
Campbell Barton
committed
blen_uvs = me.uv_layers[0]
for face_uvidx, lidx in zip(face_vert_tex_indices, blen_poly.loop_indices):
blen_uvs.data[lidx].uv = verts_tex[0 if (face_uvidx is ...) else face_uvidx]
use_edges = use_edges and bool(edges)
me.edges.add(len(edges))
# edges should be a list of (a, b) tuples
me.edges.foreach_set("vertices", unpack_list(edges))
me.validate(clean_customdata=False) # *Very* important to not remove lnors here!
Philipp Oeser
committed
me.update(calc_edges=use_edges)
Bastien Montagne
committed
# Un-tessellate as much as possible, in case we had to triangulate some ngons...
if fgon_edges:
import bmesh
bm = bmesh.new()
bm.from_mesh(me)
verts = bm.verts[:]
get = bm.edges.get
edges = [get((verts[vidx1], verts[vidx2])) for vidx1, vidx2 in fgon_edges]
try:
bmesh.ops.dissolve_edges(bm, edges=edges, use_verts=False)
except:
# Possible dissolve fails for some edges, but don't fail silently in case this is a real bug.
import traceback
traceback.print_exc()
bm.to_mesh(me)
bm.free()
# XXX If validate changes the geometry, this is likely to be broken...
Philipp Oeser
committed
if unique_smooth_groups and sharp_edges:
for e in me.edges:
if e.key in sharp_edges:
e.use_edge_sharp = True
me.show_edge_sharp = True
if verts_nor:
clnors = array.array('f', [0.0] * (len(me.loops) * 3))
me.loops.foreach_get("normal", clnors)
if not unique_smooth_groups:
me.polygons.foreach_set("use_smooth", [True] * len(me.polygons))
me.normals_split_custom_set(tuple(zip(*(iter(clnors),) * 3)))
me.use_auto_smooth = True
me.show_edge_sharp = True
ob = bpy.data.objects.new(me.name, me)
new_objects.append(ob)
# Create the vertex groups. No need to have the flag passed here since we test for the
# content of the vertex_groups. If the user selects to NOT have vertex groups saved then
# the following test will never run
for group_name, group_indices in vertex_groups.items():
Campbell Barton
committed
group = ob.vertex_groups.new(group_name.decode('utf-8', "replace"))
group.add(group_indices, 1.0, 'REPLACE')
def create_nurbs(context_nurbs, vert_loc, new_objects):
Add nurbs object to blender, only support one type at the moment
deg = context_nurbs.get(b'deg', (3,))
curv_range = context_nurbs.get(b'curv_range')
curv_idx = context_nurbs.get(b'curv_idx', [])
parm_u = context_nurbs.get(b'parm_u', [])
parm_v = context_nurbs.get(b'parm_v', [])
name = context_nurbs.get(b'name', b'ObjNurb')
cstype = context_nurbs.get(b'cstype')
if cstype is None:
print('\tWarning, cstype not found')
return
if cstype != b'bspline':
print('\tWarning, cstype is not supported (only bspline)')
return
if not curv_idx:
print('\tWarning, curv argument empty or not set')
return
if len(deg) > 1 or parm_v:
print('\tWarning, surfaces not supported')
return
Campbell Barton
committed
cu = bpy.data.curves.new(name.decode('utf-8', "replace"), 'CURVE')
cu.dimensions = '3D'
nu = cu.splines.new('NURBS')
nu.points.add(len(curv_idx) - 1) # a point is added to start with
nu.points.foreach_set("co", [co_axis for vt_idx in curv_idx for co_axis in (vert_loc[vt_idx] + (1.0,))])
nu.order_u = deg[0] + 1
# get for endpoint flag from the weighting
do_endpoints = True
do_endpoints = False
break
if abs(parm_u[-(i + 1)] - curv_range[1]) > 0.0001:
do_endpoints = False
break
else:
do_endpoints = False
if do_endpoints:
nu.use_endpoint_u = True
# close
'''
do_closed = False
if len(parm_u) > deg[0]+1:
for i in xrange(deg[0]+1):
#print curv_idx[i], curv_idx[-(i+1)]
if curv_idx[i]==curv_idx[-(i+1)]:
do_closed = True
break
if do_closed:
nu.use_cyclic_u = True
'''
Campbell Barton
committed
ob = bpy.data.objects.new(name.decode('utf-8', "replace"), cu)
new_objects.append(ob)
def strip_slash(line_split):
if line_split[-1][-1] == 92: # '\' char
if len(line_split[-1]) == 1:
line_split.pop() # remove the \ item
else:
line_split[-1] = line_split[-1][:-1] # remove the \ from the end last number
return True
return False
def get_float_func(filepath):
find the float function for this obj file
- whether to replace commas or not
file = open(filepath, 'rb')
line = line.lstrip()
if line.startswith(b'v'): # vn vt v
if b',' in line:
return lambda f: float(f.replace(b',', b'.'))
elif b'.' in line:
return float
return float
def load(context,
filepath,
*,
global_clamp_size=0.0,
use_smooth_groups=True,
use_edges=True,
use_split_objects=True,
use_split_groups=True,
use_image_search=True,
use_groups_as_vgroups=False,
Campbell Barton
committed
relpath=None,
Called by the user interface or another script.
load_obj(path) - should give acceptable results.
This function passes the file and sends the data off
to be split into objects and then converted into mesh objects
def handle_vec(line_start, context_multi_line, line_split, tag, data, vec, vec_len):
ret_context_multi_line = tag if strip_slash(line_split) else b''
if line_start == tag:
vec[:] = [float_func(v) for v in line_split[1:]]
elif context_multi_line == tag:
vec += [float_func(v) for v in line_split]
if not ret_context_multi_line:
data.append(tuple(vec[:vec_len]))
return ret_context_multi_line
def create_face(context_material, context_smooth_group, context_object):
face_vert_loc_indices = []
face_vert_nor_indices = []
face_vert_tex_indices = []
return (
face_vert_loc_indices,
face_vert_nor_indices,
face_vert_tex_indices,
context_material,
context_smooth_group,
context_object,
Bastien Montagne
committed
[], # If non-empty, that face is a Blender-invalid ngon (holes...), need a mutable object for that...
)
with ProgressReport(context.window_manager) as progress:
progress.enter_substeps(1, "Importing OBJ %r..." % filepath)
if global_matrix is None:
global_matrix = mathutils.Matrix()
if use_split_objects or use_split_groups:
use_groups_as_vgroups = False
time_main = time.time()
verts_loc = []
verts_nor = []
verts_tex = []
faces = [] # tuples of the faces
material_libs = set() # filenames to material libs this OBJ uses
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
vertex_groups = {} # when use_groups_as_vgroups is true
# Get the string to float conversion func for this file- is 'float' for almost all files.
float_func = get_float_func(filepath)
# Context variables
context_material = None
context_smooth_group = None
context_object = None
context_vgroup = None
# Nurbs
context_nurbs = {}
nurbs = []
context_parm = b'' # used by nurbs too but could be used elsewhere
# Until we can use sets
unique_materials = {}
unique_material_images = {}
unique_smooth_groups = {}
# unique_obects= {} - no use for this variable since the objects are stored in the face.
# when there are faces that end with \
# it means they are multiline-
# since we use xreadline we cant skip to the next line
# so we need to know whether
context_multi_line = b''
# Per-face handling data.
face_vert_loc_indices = None
face_vert_nor_indices = None
face_vert_tex_indices = None
face_vert_nor_valid = face_vert_tex_valid = False
face_items_usage = set()
face_invalid_blenpoly = None
prev_vidx = None
face = None
vec = []
progress.enter_substeps(3, "Parsing OBJ file...")
with open(filepath, 'rb') as f:
for line in f: # .readlines():
line_split = line.split()
if not line_split:
continue
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
line_start = line_split[0] # we compare with this a _lot_
if line_start == b'v' or context_multi_line == b'v':
context_multi_line = handle_vec(line_start, context_multi_line, line_split, b'v', verts_loc, vec, 3)
elif line_start == b'vn' or context_multi_line == b'vn':
context_multi_line = handle_vec(line_start, context_multi_line, line_split, b'vn', verts_nor, vec, 3)
elif line_start == b'vt' or context_multi_line == b'vt':
context_multi_line = handle_vec(line_start, context_multi_line, line_split, b'vt', verts_tex, vec, 2)
# Handle faces lines (as faces) and the second+ lines of fa multiline face here
# use 'f' not 'f ' because some objs (very rare have 'fo ' for faces)
elif line_start == b'f' or context_multi_line == b'f':
if not context_multi_line:
line_split = line_split[1:]
# Instantiate a face
face = create_face(context_material, context_smooth_group, context_object)
(face_vert_loc_indices, face_vert_nor_indices, face_vert_tex_indices,
_1, _2, _3, face_invalid_blenpoly) = face
faces.append(face)
face_items_usage.clear()
# Else, use face_vert_loc_indices and face_vert_tex_indices previously defined and used the obj_face
context_multi_line = b'f' if strip_slash(line_split) else b''
for v in line_split:
obj_vert = v.split(b'/')
idx = int(obj_vert[0]) - 1
vert_loc_index = (idx + len(verts_loc) + 1) if (idx < 0) else idx
# Add the vertex to the current group
# *warning*, this wont work for files that have groups defined around verts
if use_groups_as_vgroups and context_vgroup:
vertex_groups[context_vgroup].append(vert_loc_index)
# This a first round to quick-detect ngons that *may* use a same edge more than once.
# Potential candidate will be re-checked once we have done parsing the whole face.
if not face_invalid_blenpoly:
# If we use more than once a same vertex, invalid ngon is suspected.
if vert_loc_index in face_items_usage:
face_invalid_blenpoly.append(True)
else:
face_items_usage.add(vert_loc_index)
face_vert_loc_indices.append(vert_loc_index)
# formatting for faces with normals and textures is
# loc_index/tex_index/nor_index
if len(obj_vert) > 1 and obj_vert[1] and obj_vert[1] != b'0':
idx = int(obj_vert[1]) - 1
face_vert_tex_indices.append((idx + len(verts_tex) + 1) if (idx < 0) else idx)
face_vert_tex_valid = True
else:
face_vert_tex_indices.append(...)
if len(obj_vert) > 2 and obj_vert[2] and obj_vert[2] != b'0':
idx = int(obj_vert[2]) - 1
face_vert_nor_indices.append((idx + len(verts_nor) + 1) if (idx < 0) else idx)
face_vert_nor_valid = True
else:
face_vert_nor_indices.append(...)
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
if not context_multi_line:
# Clear nor/tex indices in case we had none defined for this face.
if not face_vert_nor_valid:
face_vert_nor_indices.clear()
if not face_vert_tex_valid:
face_vert_tex_indices.clear()
face_vert_nor_valid = face_vert_tex_valid = False
# Means we have finished a face, we have to do final check if ngon is suspected to be blender-invalid...
if face_invalid_blenpoly:
face_invalid_blenpoly.clear()
face_items_usage.clear()
prev_vidx = face_vert_loc_indices[-1]
for vidx in face_vert_loc_indices:
edge_key = (prev_vidx, vidx) if (prev_vidx < vidx) else (vidx, prev_vidx)
if edge_key in face_items_usage:
face_invalid_blenpoly.append(True)
break
face_items_usage.add(edge_key)
prev_vidx = vidx
elif use_edges and (line_start == b'l' or context_multi_line == b'l'):
# very similar to the face load function above with some parts removed
if not context_multi_line:
line_split = line_split[1:]
# Instantiate a face
face = create_face(context_material, context_smooth_group, context_object)
face_vert_loc_indices = face[0]
# XXX A bit hackish, we use special 'value' of face_vert_nor_indices (a single True item) to tag this
# as a polyline, and not a regular face...
face[1][:] = [True]
faces.append(face)
# Else, use face_vert_loc_indices previously defined and used the obj_face
context_multi_line = b'l' if strip_slash(line_split) else b''
for v in line_split:
obj_vert = v.split(b'/')
idx = int(obj_vert[0]) - 1
face_vert_loc_indices.append((idx + len(verts_loc) + 1) if (idx < 0) else idx)
elif line_start == b's':
if use_smooth_groups:
context_smooth_group = line_value(line_split)
if context_smooth_group == b'off':
context_smooth_group = None
elif context_smooth_group: # is not None
unique_smooth_groups[context_smooth_group] = None
elif line_start == b'o':
if use_split_objects:
context_object = line_value(line_split)
# unique_obects[context_object]= None
elif line_start == b'g':
if use_split_groups:
context_object = line_value(line.split())
# print 'context_object', context_object
# unique_obects[context_object]= None
elif use_groups_as_vgroups:
context_vgroup = line_value(line.split())
if context_vgroup and context_vgroup != b'(null)':
vertex_groups.setdefault(context_vgroup, [])
else:
context_vgroup = None # dont assign a vgroup
elif line_start == b'usemtl':
context_material = line_value(line.split())
unique_materials[context_material] = None
elif line_start == b'mtllib': # usemap or usemat
# can have multiple mtllib filenames per line, mtllib can appear more than once,
# so make sure only occurrence of material exists
material_libs |= {os.fsdecode(f) for f in line.split()[1:]}
# Nurbs support
elif line_start == b'cstype':
context_nurbs[b'cstype'] = line_value(line.split()) # 'rat bspline' / 'bspline'
elif line_start == b'curv' or context_multi_line == b'curv':
curv_idx = context_nurbs[b'curv_idx'] = context_nurbs.get(b'curv_idx', []) # in case were multiline
if not context_multi_line:
context_nurbs[b'curv_range'] = float_func(line_split[1]), float_func(line_split[2])
line_split[0:3] = [] # remove first 3 items
if strip_slash(line_split):
context_multi_line = b'curv'
else:
context_multi_line = b''
for i in line_split:
vert_loc_index = int(i) - 1
if vert_loc_index < 0:
vert_loc_index = len(verts_loc) + vert_loc_index + 1
curv_idx.append(vert_loc_index)
elif line_start == b'parm' or context_multi_line == b'parm':
if context_multi_line:
context_multi_line = b''
Bastien Montagne
committed
else:
context_parm = line_split[1]
line_split[0:2] = [] # remove first 2
Bastien Montagne
committed
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
if strip_slash(line_split):
context_multi_line = b'parm'
else:
context_multi_line = b''
if context_parm.lower() == b'u':
context_nurbs.setdefault(b'parm_u', []).extend([float_func(f) for f in line_split])
elif context_parm.lower() == b'v': # surfaces not supported yet
context_nurbs.setdefault(b'parm_v', []).extend([float_func(f) for f in line_split])
# else: # may want to support other parm's ?
elif line_start == b'deg':
context_nurbs[b'deg'] = [int(i) for i in line.split()[1:]]
elif line_start == b'end':
# Add the nurbs curve
if context_object:
context_nurbs[b'name'] = context_object
nurbs.append(context_nurbs)
context_nurbs = {}
context_parm = b''
''' # How to use usemap? depricated?
elif line_start == b'usema': # usemap or usemat
context_image= line_value(line_split)
'''
progress.step("Done, loading materials and images...")
create_materials(filepath, relpath, material_libs, unique_materials,
unique_material_images, use_image_search, use_cycles, float_func)
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
progress.step("Done, building geometries (verts:%i faces:%i materials: %i smoothgroups:%i) ..." %
(len(verts_loc), len(faces), len(unique_materials), len(unique_smooth_groups)))
# deselect all
if bpy.ops.object.select_all.poll():
bpy.ops.object.select_all(action='DESELECT')
scene = context.scene
new_objects = [] # put new objects here
# Split the mesh by objects/materials, may
SPLIT_OB_OR_GROUP = bool(use_split_objects or use_split_groups)
for data in split_mesh(verts_loc, faces, unique_materials, filepath, SPLIT_OB_OR_GROUP):
verts_loc_split, faces_split, unique_materials_split, dataname, use_vnor, use_vtex = data
# Create meshes from the data, warning 'vertex_groups' wont support splitting
#~ print(dataname, use_vnor, use_vtex)
create_mesh(new_objects,
use_edges,
verts_loc_split,
verts_nor if use_vnor else [],
verts_tex if use_vtex else [],
faces_split,
unique_materials_split,
unique_material_images,
unique_smooth_groups,
vertex_groups,
dataname,
)
# nurbs support
for context_nurbs in nurbs:
create_nurbs(context_nurbs, verts_loc, new_objects)
# Create new obj
for obj in new_objects:
base = scene.objects.link(obj)
base.select = True
# we could apply this anywhere before scaling.
obj.matrix_world = global_matrix
axis_min = [1000000000] * 3
axis_max = [-1000000000] * 3
if global_clamp_size:
# Get all object bounds
for ob in new_objects:
for v in ob.bound_box:
for axis, value in enumerate(v):
if axis_min[axis] > value:
axis_min[axis] = value
if axis_max[axis] < value:
axis_max[axis] = value
# Scale objects
max_axis = max(axis_max[0] - axis_min[0], axis_max[1] - axis_min[1], axis_max[2] - axis_min[2])
scale = 1.0
while global_clamp_size < max_axis * scale:
scale = scale / 10.0
Campbell Barton
committed
for obj in new_objects:
obj.scale = scale, scale, scale
progress.leave_substeps("Done.")
progress.leave_substeps("Finished importing: %r" % filepath)
return {'FINISHED'}