安装Blender并使用

前言

该系列记录了如何用Blenderpro来构建自己的场景数据集，从环境搭建到后期构建数据集的整个流程

本文章是第一部分，BlenderPrc2的安装以及环境配置

部分参考https://blog.csdn.net/weixin_49521551/article/details/121573334
官方文档https://dlr-rm.github.io/BlenderProc/

一、Blenderpro是什么？

一个用于真实感渲染的程序性 Blender pipeline。

二、使用步骤

1.创建虚拟环境

为了方便日后的环境管理，我们在anaconda新建一个虚拟环境进行Blenderproc的安装，anaconda安装过程略，新环境命令如下：

conda create -n blender python=3.7  # blender处为你的虚拟环境名称
conda activate blender

2.安装Blenderproc（这一步直接按github官方教程走）

blenderproc的安装有两种方式，第一种是通过github安装，第二种是使用使用更简便的pip方式

via pip

在虚拟环境下直接执行
代码如下：

pip install Blenderproc

via github

但是不建议这样安装，还是使用作者在github上的Readme.md安装步骤更为合适

具体步骤为

git clone https://github.com/DLR-RM/BlenderProc
cd BlenderProc
pip install -e .

之后终端执行命令：

blenderproc run quickstart.py

如果环境没有问题，则会直接生成output文件夹，里面会生成输出的0.hdf5

但是一般环境都是不全的，所以下面介绍环境安装细节

三.配置环境

1.bug1

当运行代码时，发现会出现问题，如：

第一次运行blenderproc run quickstart.py时，会自动在blender官网下载blender软件的安装包，这个过程报错：urllib.error.HTTPError：HTTP Error 403：Forbidden

意味着需要在官网下载blender并安装，但是由于网路问题没有下载成功，给出解决办法：

手动下载安装包，并安装在路径/home/usename/下即可

2.bug2

当运行代码时，发现会出现问题，如：

[BUG]: Error: ModuleNotFoundError: No module named 'skimage'

意味着当前环境代码找不到scikit-image包，当然也能是找不到其他的包，比如imageio、opencv等都有可能。那么就需要安装这个包，最直接的方法当然是

pip install scikit-image

但是这个是错误的，因为blenderproc是一个单独的环境，自己是看不到的，实在想找其实是在blener的安装路径下的python/site-package下，那么解决方法就是使用指令

blenderproc pip install scikit-image

如果没有解决，可以尝试先卸载，再安装，并加上参数

blenderproc pip uninstall scikit-image
blenderproc pip install scikit-image --not-use-custom-package-path

这会将 scikit-image安装到 blenders 站点包文件夹中，而不是我们的自定义文件夹。
现在，脚本应该可以正常运行。

参考链接：https://github.com/DLR-RM/BlenderProc/issues/855

3.bug3

当运行代码时，发现会出现问题，如：

【错误记录/Blender】python中使用bpy模块

那还不简单，直接

pip install bpy

但是不行，会出现不知名问题，解决办法是下载离线包，再手动安装

地址如下：Releases · TylerGubala/blenderpy

然后再运行指令，即可安装成功

pip install bpy-2.91a0-cp37-cp37m-manylinux2014_x86_64.whl && bpy_post_install

这样在pip list里就可以找到bpy了

参考链接：https://blog.csdn.net/qq_33446100/article/details/117591371

4.bug4

当运行代码时，发现会出现问题，如：

【错误记录/mathutils】python中使用mathutils模块

使用blenderproc的时候，需要安装mathutils，但是直接pip会出现问题

官方安装方法

这里是mathutils的库，里面写了安装方式：

mathutils 3.3.0 on PyPI - Libraries.io

git clone https://gitlab.com/ideasman42/blender-mathutils.git
cd blender-mathutils# To build you can choose between pythons distutils or CMake.
# distutils:
python setup.py build
sudo python setup.py install# CMake:
cmake .
make
sudo make install# 非root模式可以去掉sudo

但是如果python版本在3.8及其以下，会在install时报错，原因在于PyModule_AddType无法解析，这里参考了其他帖子中评论的方法已解决该问题。

解决方式：

①把src/mathutils/mathutils.c中的PyModule_AddType改成PyModule_AddObject，

②把参数改成三个，中间设置一个参数接收moudle名称，要记得在前面先初始化

  char *mod_name;#### 略/* each type has its own new() function */PyModule_AddObject(mod, mod_name, &vector_Type);PyModule_AddObject(mod, mod_name, &matrix_Type);PyModule_AddObject(mod, mod_name, &euler_Type);PyModule_AddObject(mod, mod_name, &quaternion_Type);PyModule_AddObject(mod, mod_name, &color_Type);

③把 y = _Py_HashDouble(NULL, (double)(array[i++]))中的NULL删掉

以上步骤执行完成后，按照官方安装方法继续执行即可。

src/mathutils/mathutils.c修改后的完整代码：

/* SPDX-License-Identifier: GPL-2.0-or-later *//** \file* \ingroup pymathutils*/#include <Python.h>#include "mathutils.h"#include "BLI_math.h"
#include "BLI_utildefines.h"#include "../generic/py_capi_utils.h"
#include "../generic/python_utildefines.h"#ifndef MATH_STANDALONE
#  include "BLI_dynstr.h"
#endifPyDoc_STRVAR(M_Mathutils_doc,"This module provides access to math operations.\n""\n"".. note::\n""\n""   Classes, methods and attributes that accept vectors also accept other numeric sequences,\n""   such as tuples, lists.\n""\n""The :mod:`mathutils` module provides the following classes:\n""\n""- :class:`Color`,\n""- :class:`Euler`,\n""- :class:`Matrix`,\n""- :class:`Quaternion`,\n""- :class:`Vector`,\n");
static int mathutils_array_parse_fast(float *array,int size,PyObject *value_fast,const char *error_prefix)
{PyObject *item;PyObject **value_fast_items = PySequence_Fast_ITEMS(value_fast);int i;i = size;do {i--;if (((array[i] = PyFloat_AsDouble((item = value_fast_items[i]))) == -1.0f) &&PyErr_Occurred()) {PyErr_Format(PyExc_TypeError,"%.200s: sequence index %d expected a number, ""found '%.200s' type, ",error_prefix,i,Py_TYPE(item)->tp_name);size = -1;break;}} while (i);return size;
}Py_hash_t mathutils_array_hash(const float *array, size_t array_len)
{int i;Py_uhash_t x; /* Unsigned for defined overflow behavior. */Py_hash_t y;Py_uhash_t mult;Py_ssize_t len;mult = _PyHASH_MULTIPLIER;len = array_len;x = 0x345678UL;i = 0;while (--len >= 0) {y = _Py_HashDouble((double)(array[i++]));if (y == -1) {return -1;}x = (x ^ y) * mult;/* the cast might truncate len; that doesn't change hash stability */mult += (Py_hash_t)(82520UL + len + len);}x += 97531UL;if (x == (Py_uhash_t)-1) {x = -2;}return x;
}int mathutils_array_parse(float *array, int array_num_min, int array_num_max, PyObject *value, const char *error_prefix)
{const uint flag = array_num_max;int num;array_num_max &= ~MU_ARRAY_FLAGS;#if 1 /* approx 6x speedup for mathutils types */if ((num = VectorObject_Check(value) ? ((VectorObject *)value)->vec_num : 0) ||(num = EulerObject_Check(value) ? 3 : 0) || (num = QuaternionObject_Check(value) ? 4 : 0) ||(num = ColorObject_Check(value) ? 3 : 0)) {if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {return -1;}if (flag & MU_ARRAY_SPILL) {CLAMP_MAX(num, array_num_max);}if (num > array_num_max || num < array_num_min) {if (array_num_max == array_num_min) {PyErr_Format(PyExc_ValueError,"%.200s: sequence length is %d, expected %d",error_prefix,num,array_num_max);}else {PyErr_Format(PyExc_ValueError,"%.200s: sequence length is %d, expected [%d - %d]",error_prefix,num,array_num_min,array_num_max);}return -1;}memcpy(array, ((const BaseMathObject *)value)->data, num * sizeof(float));}else
#endif{PyObject *value_fast = NULL;/* non list/tuple cases */if (!(value_fast = PySequence_Fast(value, error_prefix))) {/* PySequence_Fast sets the error */return -1;}num = PySequence_Fast_GET_SIZE(value_fast);if (flag & MU_ARRAY_SPILL) {CLAMP_MAX(num, array_num_max);}if (num > array_num_max || num < array_num_min) {if (array_num_max == array_num_min) {PyErr_Format(PyExc_ValueError,"%.200s: sequence length is %d, expected %d",error_prefix,num,array_num_max);}else {PyErr_Format(PyExc_ValueError,"%.200s: sequence length is %d, expected [%d - %d]",error_prefix,num,array_num_min,array_num_max);}Py_DECREF(value_fast);return -1;}num = mathutils_array_parse_fast(array, num, value_fast, error_prefix);Py_DECREF(value_fast);}if (num != -1) {if (flag & MU_ARRAY_ZERO) {const int array_num_left = array_num_max - num;if (array_num_left) {memset(&array[num], 0, sizeof(float) * array_num_left);}}}return num;
}int mathutils_array_parse_alloc(float **array,int array_num,PyObject *value,const char *error_prefix)
{int num;#if 1 /* approx 6x speedup for mathutils types */if ((num = VectorObject_Check(value) ? ((VectorObject *)value)->vec_num : 0) ||(num = EulerObject_Check(value) ? 3 : 0) || (num = QuaternionObject_Check(value) ? 4 : 0) ||(num = ColorObject_Check(value) ? 3 : 0)) {if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {return -1;}if (num < array_num) {PyErr_Format(PyExc_ValueError,"%.200s: sequence size is %d, expected > %d",error_prefix,num,array_num);return -1;}*array = PyMem_Malloc(num * sizeof(float));memcpy(*array, ((const BaseMathObject *)value)->data, num * sizeof(float));return num;}#endifPyObject *value_fast = NULL;// *array = NULL;int ret;/* non list/tuple cases */if (!(value_fast = PySequence_Fast(value, error_prefix))) {/* PySequence_Fast sets the error */return -1;}num = PySequence_Fast_GET_SIZE(value_fast);if (num < array_num) {Py_DECREF(value_fast);PyErr_Format(PyExc_ValueError,"%.200s: sequence size is %d, expected > %d",error_prefix,num,array_num);return -1;}*array = PyMem_Malloc(num * sizeof(float));ret = mathutils_array_parse_fast(*array, num, value_fast, error_prefix);Py_DECREF(value_fast);if (ret == -1) {PyMem_Free(*array);}return ret;
}int mathutils_array_parse_alloc_v(float **array,int array_dim,PyObject *value,const char *error_prefix)
{PyObject *value_fast;const int array_dim_flag = array_dim;int i, num;/* non list/tuple cases */if (!(value_fast = PySequence_Fast(value, error_prefix))) {/* PySequence_Fast sets the error */return -1;}num = PySequence_Fast_GET_SIZE(value_fast);if (num != 0) {PyObject **value_fast_items = PySequence_Fast_ITEMS(value_fast);float *fp;array_dim &= ~MU_ARRAY_FLAGS;fp = *array = PyMem_Malloc(num * array_dim * sizeof(float));for (i = 0; i < num; i++, fp += array_dim) {PyObject *item = value_fast_items[i];if (mathutils_array_parse(fp, array_dim, array_dim_flag, item, error_prefix) == -1) {PyMem_Free(*array);*array = NULL;num = -1;break;}}}Py_DECREF(value_fast);return num;
}int mathutils_int_array_parse(int *array, int array_dim, PyObject *value, const char *error_prefix)
{int size, i;PyObject *value_fast, **value_fast_items, *item;if (!(value_fast = PySequence_Fast(value, error_prefix))) {/* PySequence_Fast sets the error */return -1;}if ((size = PySequence_Fast_GET_SIZE(value_fast)) != array_dim) {PyErr_Format(PyExc_ValueError,"%.200s: sequence size is %d, expected %d",error_prefix,size,array_dim);Py_DECREF(value_fast);return -1;}value_fast_items = PySequence_Fast_ITEMS(value_fast);i = size;while (i > 0) {i--;if (((array[i] = PyC_Long_AsI32((item = value_fast_items[i]))) == -1) && PyErr_Occurred()) {PyErr_Format(PyExc_TypeError, "%.200s: sequence index %d expected an int", error_prefix, i);size = -1;break;}}Py_DECREF(value_fast);return size;
}int mathutils_array_parse_alloc_vi(int **array,int array_dim,PyObject *value,const char *error_prefix)
{PyObject *value_fast;int i, size;if (!(value_fast = PySequence_Fast(value, error_prefix))) {/* PySequence_Fast sets the error */return -1;}size = PySequence_Fast_GET_SIZE(value_fast);if (size != 0) {PyObject **value_fast_items = PySequence_Fast_ITEMS(value_fast);int *ip;ip = *array = PyMem_Malloc(size * array_dim * sizeof(int));for (i = 0; i < size; i++, ip += array_dim) {PyObject *item = value_fast_items[i];if (mathutils_int_array_parse(ip, array_dim, item, error_prefix) == -1) {PyMem_Free(*array);*array = NULL;size = -1;break;}}}Py_DECREF(value_fast);return size;
}int mathutils_array_parse_alloc_viseq(int **array, int **start_table, int **len_table, PyObject *value, const char *error_prefix)
{PyObject *value_fast, *subseq;int i, size, start, subseq_len;int *ip;*array = NULL;*start_table = NULL;*len_table = NULL;if (!(value_fast = PySequence_Fast(value, error_prefix))) {/* PySequence_Fast sets the error */return -1;}size = PySequence_Fast_GET_SIZE(value_fast);if (size != 0) {PyObject **value_fast_items = PySequence_Fast_ITEMS(value_fast);*start_table = PyMem_Malloc(size * sizeof(int));*len_table = PyMem_Malloc(size * sizeof(int));/* First pass to set starts and len, and calculate size of array needed */start = 0;for (i = 0; i < size; i++) {subseq = value_fast_items[i];if ((subseq_len = (int)PySequence_Size(subseq)) == -1) {PyErr_Format(PyExc_ValueError, "%.200s: sequence expected to have subsequences", error_prefix);PyMem_Free(*start_table);PyMem_Free(*len_table);Py_DECREF(value_fast);*start_table = NULL;*len_table = NULL;return -1;}(*start_table)[i] = start;(*len_table)[i] = subseq_len;start += subseq_len;}ip = *array = PyMem_Malloc(start * sizeof(int));/* Second pass to parse the subsequences into array */for (i = 0; i < size; i++) {subseq = value_fast_items[i];subseq_len = (*len_table)[i];if (mathutils_int_array_parse(ip, subseq_len, subseq, error_prefix) == -1) {PyMem_Free(*array);PyMem_Free(*start_table);PyMem_Free(*len_table);*array = NULL;*len_table = NULL;*start_table = NULL;size = -1;break;}ip += subseq_len;}}Py_DECREF(value_fast);return size;
}int mathutils_any_to_rotmat(float rmat[3][3], PyObject *value, const char *error_prefix)
{if (EulerObject_Check(value)) {if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {return -1;}eulO_to_mat3(rmat, ((const EulerObject *)value)->eul, ((const EulerObject *)value)->order);return 0;}if (QuaternionObject_Check(value)) {if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {return -1;}float tquat[4];normalize_qt_qt(tquat, ((const QuaternionObject *)value)->quat);quat_to_mat3(rmat, tquat);return 0;}if (MatrixObject_Check(value)) {if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {return -1;}if (((MatrixObject *)value)->row_num < 3 || ((MatrixObject *)value)->col_num < 3) {PyErr_Format(PyExc_ValueError, "%.200s: matrix must have minimum 3x3 dimensions", error_prefix);return -1;}matrix_as_3x3(rmat, (MatrixObject *)value);normalize_m3(rmat);return 0;}PyErr_Format(PyExc_TypeError,"%.200s: expected a Euler, Quaternion or Matrix type, ""found %.200s",error_prefix,Py_TYPE(value)->tp_name);return -1;
}/* ----------------------------------MATRIX FUNCTIONS-------------------- *//* Utility functions *//* LomontRRDCompare4, Ever Faster Float Comparisons by Randy Dillon */
/* XXX We may want to use 'safer' BLI's compare_ff_relative ultimately?* LomontRRDCompare4() is an optimized version of Dawson's AlmostEqual2sComplement()* (see [1] and [2]).* Dawson himself now claims this is not a 'safe' thing to do* (pushing ULP method beyond its limits),* an recommends using work from [3] instead, which is done in BLI func...** [1] http://www.randydillon.org/Papers/2007/everfast.htm* [2] http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm* [3] https://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/* instead.*/
#define SIGNMASK(i) (-(int)(((uint)(i)) >> 31))int EXPP_FloatsAreEqual(float af, float bf, int maxDiff)
{/* solid, fast routine across all platforms* with constant time behavior */const int ai = *(const int *)(&af);const int bi = *(const int *)(&bf);const int test = SIGNMASK(ai ^ bi);int diff, v1, v2;BLI_assert((0 == test) || (0xFFFFFFFF == test));diff = (ai ^ (test & 0x7fffffff)) - bi;v1 = maxDiff + diff;v2 = maxDiff - diff;return (v1 | v2) >= 0;
}/*---------------------- EXPP_VectorsAreEqual -------------------------* Builds on EXPP_FloatsAreEqual to test vectors */
int EXPP_VectorsAreEqual(const float *vecA, const float *vecB, int size, int floatSteps)
{int x;for (x = 0; x < size; x++) {if (EXPP_FloatsAreEqual(vecA[x], vecB[x], floatSteps) == 0) {return 0;}}return 1;
}#ifndef MATH_STANDALONE
PyObject *mathutils_dynstr_to_py(struct DynStr *ds)
{const int ds_len = BLI_dynstr_get_len(ds); /* space for \0 */char *ds_buf = PyMem_Malloc(ds_len + 1);PyObject *ret;BLI_dynstr_get_cstring_ex(ds, ds_buf);BLI_dynstr_free(ds);ret = PyUnicode_FromStringAndSize(ds_buf, ds_len);PyMem_Free(ds_buf);return ret;
}
#endif/* Mathutils Callbacks *//* For mathutils internal use only,* eventually should re-alloc but to start with we only have a few users. */
#define MATHUTILS_TOT_CB 17
static Mathutils_Callback *mathutils_callbacks[MATHUTILS_TOT_CB] = {NULL};uchar Mathutils_RegisterCallback(Mathutils_Callback *cb)
{uchar i;/* find the first free slot */for (i = 0; mathutils_callbacks[i]; i++) {if (mathutils_callbacks[i] == cb) {/* already registered? */return i;}}BLI_assert(i + 1 < MATHUTILS_TOT_CB);mathutils_callbacks[i] = cb;return i;
}int _BaseMathObject_CheckCallback(BaseMathObject *self)
{Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];if (LIKELY(cb->check(self) != -1)) {return 0;}return -1;
}/* use macros to check for NULL */
int _BaseMathObject_ReadCallback(BaseMathObject *self)
{Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];if (LIKELY(cb->get(self, self->cb_subtype) != -1)) {return 0;}if (!PyErr_Occurred()) {PyErr_Format(PyExc_RuntimeError, "%s read, user has become invalid", Py_TYPE(self)->tp_name);}return -1;
}int _BaseMathObject_WriteCallback(BaseMathObject *self)
{Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];if (LIKELY(cb->set(self, self->cb_subtype) != -1)) {return 0;}if (!PyErr_Occurred()) {PyErr_Format(PyExc_RuntimeError, "%s write, user has become invalid", Py_TYPE(self)->tp_name);}return -1;
}int _BaseMathObject_ReadIndexCallback(BaseMathObject *self, int index)
{Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];if (LIKELY(cb->get_index(self, self->cb_subtype, index) != -1)) {return 0;}if (!PyErr_Occurred()) {PyErr_Format(PyExc_RuntimeError, "%s read index, user has become invalid", Py_TYPE(self)->tp_name);}return -1;
}int _BaseMathObject_WriteIndexCallback(BaseMathObject *self, int index)
{Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];if (LIKELY(cb->set_index(self, self->cb_subtype, index) != -1)) {return 0;}if (!PyErr_Occurred()) {PyErr_Format(PyExc_RuntimeError, "%s write index, user has become invalid", Py_TYPE(self)->tp_name);}return -1;
}void _BaseMathObject_RaiseFrozenExc(const BaseMathObject *self)
{PyErr_Format(PyExc_TypeError, "%s is frozen (immutable)", Py_TYPE(self)->tp_name);
}void _BaseMathObject_RaiseNotFrozenExc(const BaseMathObject *self)
{PyErr_Format(PyExc_TypeError, "%s is not frozen (mutable), call freeze first", Py_TYPE(self)->tp_name);
}/* BaseMathObject generic functions for all mathutils types */
char BaseMathObject_owner_doc[] = "The item this is wrapping or None  (read-only).";
PyObject *BaseMathObject_owner_get(BaseMathObject *self, void *UNUSED(closure))
{PyObject *ret = self->cb_user ? self->cb_user : Py_None;return Py_INCREF_RET(ret);
}char BaseMathObject_is_wrapped_doc[] ="True when this object wraps external data (read-only).\n\n:type: boolean";
PyObject *BaseMathObject_is_wrapped_get(BaseMathObject *self, void *UNUSED(closure))
{return PyBool_FromLong((self->flag & BASE_MATH_FLAG_IS_WRAP) != 0);
}char BaseMathObject_is_frozen_doc[] ="True when this object has been frozen (read-only).\n\n:type: boolean";
PyObject *BaseMathObject_is_frozen_get(BaseMathObject *self, void *UNUSED(closure))
{return PyBool_FromLong((self->flag & BASE_MATH_FLAG_IS_FROZEN) != 0);
}char BaseMathObject_is_valid_doc[] ="True when the owner of this data is valid.\n\n:type: boolean";
PyObject *BaseMathObject_is_valid_get(BaseMathObject *self, void *UNUSED(closure))
{return PyBool_FromLong(BaseMath_CheckCallback(self) == 0);
}char BaseMathObject_freeze_doc[] =".. function:: freeze()\n""\n""   Make this object immutable.\n""\n""   After this the object can be hashed, used in dictionaries & sets.\n""\n""   :return: An instance of this object.\n";
PyObject *BaseMathObject_freeze(BaseMathObject *self)
{if ((self->flag & BASE_MATH_FLAG_IS_WRAP) || (self->cb_user != NULL)) {PyErr_SetString(PyExc_TypeError, "Cannot freeze wrapped/owned data");return NULL;}self->flag |= BASE_MATH_FLAG_IS_FROZEN;return Py_INCREF_RET((PyObject *)self);
}int BaseMathObject_traverse(BaseMathObject *self, visitproc visit, void *arg)
{Py_VISIT(self->cb_user);return 0;
}int BaseMathObject_clear(BaseMathObject *self)
{Py_CLEAR(self->cb_user);return 0;
}void BaseMathObject_dealloc(BaseMathObject *self)
{/* only free non wrapped */if ((self->flag & BASE_MATH_FLAG_IS_WRAP) == 0) {PyMem_Free(self->data);}if (self->cb_user) {PyObject_GC_UnTrack(self);BaseMathObject_clear(self);}Py_TYPE(self)->tp_free(self);  // PyObject_DEL(self); /* breaks sub-types. */
}/*----------------------------MODULE INIT-------------------------*/
static struct PyMethodDef M_Mathutils_methods[] = {{NULL, NULL, 0, NULL},
};static struct PyModuleDef M_Mathutils_module_def = {PyModuleDef_HEAD_INIT,"mathutils",         /* m_name */M_Mathutils_doc,     /* m_doc */0,                   /* m_size */M_Mathutils_methods, /* m_methods */NULL,                /* m_slots */NULL,                /* m_traverse */NULL,                /* m_clear */NULL,                /* m_free */
};/* submodules only */
#include "mathutils_geometry.h"
#include "mathutils_interpolate.h"
#ifndef MATH_STANDALONE
#  include "mathutils_bvhtree.h"
#  include "mathutils_kdtree.h"
#  include "mathutils_noise.h"
#endifPyMODINIT_FUNC PyInit_mathutils(void)
{PyObject *mod;PyObject *submodule;PyObject *sys_modules = PyImport_GetModuleDict();char *mod_name;if (PyType_Ready(&vector_Type) < 0) {return NULL;}if (PyType_Ready(&matrix_Type) < 0) {return NULL;}if (PyType_Ready(&matrix_access_Type) < 0) {return NULL;}if (PyType_Ready(&euler_Type) < 0) {return NULL;}if (PyType_Ready(&quaternion_Type) < 0) {return NULL;}if (PyType_Ready(&color_Type) < 0) {return NULL;}mod = PyModule_Create(&M_Mathutils_module_def);/* each type has its own new() function */PyModule_AddObject(mod, mod_name, &vector_Type);PyModule_AddObject(mod, mod_name, &matrix_Type);PyModule_AddObject(mod, mod_name, &euler_Type);PyModule_AddObject(mod, mod_name, &quaternion_Type);PyModule_AddObject(mod, mod_name, &color_Type);/* submodule */PyModule_AddObject(mod, "geometry", (submodule = PyInit_mathutils_geometry()));/* XXX, python doesn't do imports with this usefully yet* 'from mathutils.geometry import PolyFill'* ...fails without this. */PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);PyModule_AddObject(mod, "interpolate", (submodule = PyInit_mathutils_interpolate()));/* XXX, python doesn't do imports with this usefully yet* 'from mathutils.geometry import PolyFill'* ...fails without this. */PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);#ifndef MATH_STANDALONE/* Noise submodule */PyModule_AddObject(mod, "noise", (submodule = PyInit_mathutils_noise()));PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);/* BVHTree submodule */PyModule_AddObject(mod, "bvhtree", (submodule = PyInit_mathutils_bvhtree()));PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);/* KDTree_3d submodule */PyModule_AddObject(mod, "kdtree", (submodule = PyInit_mathutils_kdtree()));PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);
#endifmathutils_matrix_row_cb_index = Mathutils_RegisterCallback(&mathutils_matrix_row_cb);mathutils_matrix_col_cb_index = Mathutils_RegisterCallback(&mathutils_matrix_col_cb);mathutils_matrix_translation_cb_index = Mathutils_RegisterCallback(&mathutils_matrix_translation_cb);return mod;
}

到这里，mathutils也安装完成

参考链接：https://blog.csdn.net/Q_pril/article/details/137771478

5.bug5

当运行代码时，发现会出现问题，如：

No module named 'bop_toolkit_lib'

意味着找不到模块bop_toolkit_lib，需要安装，首先想到的是使用指令

blenderproc pip install bop_toolkit_lib

但是在pip里是没有这个包的，因此不会安装成功，所以这里直接先在github上下载，即运行指令

git clone https://github.com/thodan/bop_toolkit
cd bop_toolkit
pip install -r requirements.txt -e .

安装完之后，还是不会找到模块，因此，解决办法是

blenderproc pip install "git+https://github.com/thodan/bop_toolkit"

参考链接：https://github.com/DLR-RM/BlenderProc/issues/559

四.运行

终端执行命令

blenderproc run quickstart.py

运行之后在目录下会有0.dhf5的文件生成，此时运行

blenderproc vis hdf5 output/0.hdf5

即可将结果进行可视化，结果如图：

到这一步，blenderproc的环境安装配置也就完成了，更多的基础示例可以下载github项目工程并参照示例进行。