def calc_bone_by_complement(frames, bone_name, frameno): fillbf = VmdBoneFrame() # ボーン登録がなければ初期値 if bone_name not in frames: return fillbf for bidx, bf in enumerate(frames[bone_name]): if bf.frame >= frameno: # 前のキーIDXを0に見立てて、その間の補間曲線を埋める fillbf.name = bf.name fillbf.frame = frameno # 実際に登録はしない fillbf.key = False # 指定されたフレーム以降 prev_bf = frames[bone_name][bidx - 1] if prev_bf.rotation != bf.rotation: # 回転補間曲線 rn = calc_interpolate_bezier(bf.complement[48], bf.complement[52], bf.complement[56], bf.complement[60], prev_bf.frame, bf.frame, fillbf.frame) fillbf.rotation = QQuaternion.nlerp(prev_bf.rotation, bf.rotation, rn) # logger.debug("key: %s, n: %s, rn: %s, r: %s ", k, prev_frame + n, rn, QQuaternion.nlerp(prev_bf.rotation, bf.rotation, rn) ) # logger.debug("rotation: prev: %s, fill: %s ", prev_bf.rotation.toEulerAngles(), fillbf.rotation.toEulerAngles() ) else: fillbf.rotation = copy.deepcopy(prev_bf.rotation) # 補間曲線を元に間を埋める if prev_bf.position != bf.position: # http://rantyen.blog.fc2.com/blog-entry-65.html # X移動補間曲線 xn = calc_interpolate_bezier(bf.complement[0], bf.complement[4], bf.complement[8], bf.complement[12], prev_bf.frame, bf.frame, fillbf.frame) # Y移動補間曲線 yn = calc_interpolate_bezier(bf.complement[16], bf.complement[20], bf.complement[24], bf.complement[28], prev_bf.frame, bf.frame, fillbf.frame) # Z移動補間曲線 zn = calc_interpolate_bezier(bf.complement[32], bf.complement[36], bf.complement[40], bf.complement[44], prev_bf.frame, bf.frame, fillbf.frame) fillbf.position.setX(prev_bf.position.x() + ( (bf.position.x() - prev_bf.position.x()) * xn)) fillbf.position.setY(prev_bf.position.y() + ( (bf.position.y() - prev_bf.position.y()) * yn)) fillbf.position.setZ(prev_bf.position.z() + ( (bf.position.z() - prev_bf.position.z()) * zn)) # logger.debug("key: %s, n: %s, xn: %s, yn: %s, zn: %s, xa: %s", k, prev_frame + n, xn, yn, zn, ( bf.position.x() - prev_bf.position.x()) * xn ) # logger.debug("position: prev: %s, fill: %s ", prev_bf.position, fillbf.position ) else: fillbf.position = copy.deepcopy(prev_bf.position) # logger.debug("position stop: %s,%s prev: %s, fill: %s ", prev_frame + n, k, prev_bf.position, bf.position ) return fillbf # 最後まで行っても見つからなければ、最終項目を返す return copy.deepcopy(frames[bone_name][-1])
def read_vmd_file(self, filepath): # VMDファイルをバイナリ読み込み self.buffer = open(filepath, "rb").read() # vmdバージョン signature = self.unpack(30, "30s") logger.debug("signature %s", signature) motion = VmdMotion() # モーションパス motion.path = filepath # モデル名 model_bname, model_name = self.read_text(20) logger.debug("model_bname %s, model_name: %s", model_bname, model_name) motion.model_name = model_name # モーション数 motion.motion_cnt = self.read_uint(4) logger.debug("motion.motion_cnt %s", motion.motion_cnt) # モーションのあるキーのINDEX motion_indexes = {} # 1F分のモーション情報 for n in range(motion.motion_cnt): frame = VmdBoneFrame() frame.key = True frame.read = True # ボーン ---------------------- # ボーン名 bone_bname, bone_name = self.read_text(15) frame.name = bone_bname frame.format_name = bone_name logger.debug("name: %s, format_name %s", bone_bname, bone_name) # フレームIDX frame.frame = self.read_uint(4) logger.debug("frame.frame %s", frame.frame) # 位置X,Y,Z frame.position = self.read_Vector3D() logger.debug("frame.position %s", frame.position) # オリジナルを保持 frame.org_position = copy.deepcopy(frame.position) # 回転X,Y,Z,scalar frame.rotation = self.read_Quaternion() logger.debug("frame.rotation %s", frame.rotation) logger.debug("frame.rotation.euler %s", frame.rotation.toEulerAngles()) # オリジナルを保持 frame.org_rotation = copy.deepcopy(frame.rotation) # 補間曲線 frame.complement = list(self.unpack(64, "64B", True)) logger.debug("complement %s", frame.complement) # オリジナルの補間曲線を保持しておく frame.org_complement = copy.deepcopy(frame.complement) logger.debug("org_complement %s", frame.org_complement) if bone_name not in motion.frames: # まだ辞書にない場合、配列追加 motion.frames[bone_name] = [] motion_indexes[bone_name] = {} is_not_existed = True if frame.frame in motion_indexes[bone_name]: is_not_existed = False # 辞書の該当部分にボーンフレームを追加 if is_not_existed == True: motion.frames[bone_name].append(frame) motion_indexes[bone_name][frame.frame] = frame.frame if frame.frame > motion.last_motion_frame: # 最終フレームを記録 motion.last_motion_frame = frame.frame if n % 10000 == 0: print("VMDモーション読み込み キー: %s" % n) # ソート for k, v in motion.frames.items(): motion.frames[k] = sorted(v, key=lambda u: u.frame) # モーフ数 motion.morph_cnt = self.read_uint(4) logger.debug("motion.morph_cnt %s", motion.morph_cnt) # モーションのあるキーのINDEX morph_indexes = {} # 1F分のモーフ情報 for n in range(motion.morph_cnt): morph = VmdMorphFrame() # ボーン ---------------------- # ボーン名 morph_bname, morph_name = self.read_text(15) morph.name = morph_bname morph.format_name = morph_name logger.debug("name: %s, format_name %s", morph_bname, morph_name) # フレームIDX morph.frame = self.read_uint(4) logger.debug("morph.frame %s", morph.frame) # 度数 morph.ratio = self.read_float(4) logger.debug("morph.ratio %s", morph.ratio) if morph_name not in motion.morphs: # まだ辞書にない場合、配列追加 motion.morphs[morph_name] = [] morph_indexes[morph_name] = {} is_not_existed = True if morph.frame in morph_indexes[morph_name]: is_not_existed = False if is_not_existed == True: # まだなければ辞書の該当部分にモーフフレームを追加 motion.morphs[morph_name].append(morph) morph_indexes[morph_name][morph.frame] = morph.frame if n % 1000 == 0: print("VMDモーション読み込み モーフ: %s" % n) # ソート for k, v in motion.morphs.items(): motion.morphs[k] = sorted(v, key=lambda u: u.frame) # カメラ数 motion.camera_cnt = self.read_uint(4) logger.debug("motion.camera_cnt %s", motion.camera_cnt) # 1F分のカメラ情報 for _ in range(motion.camera_cnt): camera = VmdCameraFrame() # フレームIDX camera.frame = self.read_uint(4) logger.debug("camera.frame %s", camera.frame) # 距離 camera.length = self.read_float(4) logger.debug("camera.length %s", camera.length) # 位置X,Y,Z camera.position = self.read_Vector3D() logger.debug("camera.position %s", camera.position) # 角度(オイラー角) camera.euler = self.read_Vector3D() logger.debug("camera.euler %s", camera.euler) # 補間曲線 camera.complement = self.unpack(24, "24B", True) logger.debug("camera.complement %s", camera.complement) # 視野角 camera.angle = self.read_uint(4) logger.debug("camera.angle %s", camera.angle) # パース有無 camera.perspective = self.unpack(1, "B") logger.debug("camera.perspective %s", camera.perspective) # カメラを追加 motion.cameras.append(camera) # ソート motion.cameras = sorted(motion.cameras, key=lambda u: u.frame) # 照明数 try: motion.light_cnt = self.read_uint(4) logger.debug("motion.light_cnt %s", motion.light_cnt) except Exception as e: # 情報がない場合、catchして握りつぶす motion.light_cnt = 0 # 1F分の照明情報 for _ in range(motion.light_cnt): light = VmdLightFrame() # フレームIDX light.frame = self.read_uint(4) logger.debug("light.frame %s", light.frame) # 照明色(RGBだが、下手に数値が変わるのも怖いのでV3D) light.color = self.read_Vector3D() logger.debug("light.color %s", light.color) # 照明位置 light.position = self.read_Vector3D() logger.debug("light.position %s", light.position) # 追加 motion.lights.append(light) # セルフシャドウ数 try: motion.shadow_cnt = self.read_uint(4) logger.debug("motion.shadow_cnt %s", motion.shadow_cnt) # 1F分のシャドウ情報 for _ in range(motion.shadow_cnt): shadow = VmdShadowFrame() # フレームIDX shadow.frame = self.read_uint(4) logger.debug("shadow.frame %s", shadow.frame) # シャドウ種別 shadow.type = self.read_uint(1) logger.debug("shadow.type %s", shadow.type) # 距離 shadow.distance = self.read_float() logger.debug("shadow.distance %s", shadow.distance) # 追加 motion.shadows.append(shadow) except Exception as e: # 情報がない場合、catchして握りつぶす motion.shadow_cnt = 0 # IK数 try: motion.ik_cnt = self.read_uint(4) logger.debug("motion.ik_cnt %s", motion.ik_cnt) # 1F分のIK情報 for _ in range(motion.ik_cnt): ik = VmdShowIkFrame() # フレームIDX ik.frame = self.read_uint(4) logger.debug("ik.frame %s", ik.frame) # モデル表示, 0:OFF, 1:ON ik.show = self.read_uint(1) logger.debug("ik.show %s", ik.show) # 記録するIKの数 ik.ik_count = self.read_uint(4) logger.debug("ik.ik_count %s", ik.ik_count) for _ in range(ik.ik_count): ik_info = VmdInfoIk() # IK名 ik_bname, ik_name = self.read_text(20) ik_info.name = ik_bname logger.debug("ik_info.name %s", ik_name) # モデル表示, 0:OFF, 1:ON ik_info.onoff = self.read_uint(1) logger.debug("ik_info.onoff %s", ik_info.onoff) ik.ik.append(ik_info) # 追加 motion.showiks.append(ik) except Exception as e: # 昔のMMD(MMDv7.39.x64以前)はIK情報がないため、catchして握りつぶす motion.ik_cnt = 0 # ハッシュを設定 motion.digest = self.hexdigest(filepath) logger.debug("motion: %s, hash: %s", motion.path, motion.digest) return motion
def main(csv_bone_path, csv_morph_path, csv_camera_path): try: bone_frames = [] morph_frames = [] camera_frames = [] if csv_bone_path: output_vmd_path = re.sub( r'\.csv$', "_{0:%Y%m%d_%H%M%S}.vmd".format(datetime.now()), csv_bone_path) # ボーンCSV読み込み with open(csv_bone_path, encoding='cp932', mode='r') as f: reader = csv.reader(f) next(reader) # ヘッダーを読み飛ばす for row in reader: bf = VmdBoneFrame() # ボーン名 bf.format_name = row[0] # ボーン名(バイト) bf.name = bf.format_name.encode('cp932').decode( 'shift_jis').encode('shift_jis') # フレーム bf.frame = int(row[1]) # 位置 bf.position = QVector3D(float(row[2]), float(row[3]), float(row[4])) # 回転 bf.rotation = QQuaternion.fromEulerAngles( float(row[5]), float(row[6]) * -1, float(row[7]) * -1) # 補間曲線 bf.complement = [ int(row[8]), int(row[9]), int(row[10]), int(row[11]), int(row[12]), int(row[13]), int(row[14]), int(row[15]), int(row[16]), int(row[17]), int(row[18]), int(row[19]), int(row[20]), int(row[21]), int(row[22]), int(row[23]), int(row[24]), int(row[25]), int(row[26]), int(row[27]), int(row[28]), int(row[29]), int(row[30]), int(row[31]), int(row[32]), int(row[33]), int(row[34]), int(row[35]), int(row[36]), int(row[37]), int(row[38]), int(row[39]), int(row[40]), int(row[41]), int(row[42]), int(row[43]), int(row[44]), int(row[45]), int(row[46]), int(row[47]), int(row[48]), int(row[49]), int(row[50]), int(row[51]), int(row[52]), int(row[53]), int(row[54]), int(row[55]), int(row[56]), int(row[57]), int(row[58]), int(row[59]), int(row[60]), int(row[61]), int(row[62]), int(row[63]), int(row[64]), int(row[65]), int(row[66]), int(row[67]), int(row[68]), int(row[69]), int(row[70]), int(row[71]) ] bone_frames.append(bf) # logger.debug("bf: %s %s", bf.name, bf) if csv_morph_path: output_vmd_path = re.sub( r'\.csv$', "_{0:%Y%m%d_%H%M%S}.vmd".format(datetime.now()), csv_morph_path) # モーフCSV読み込み with open(csv_morph_path, encoding='cp932', mode='r') as f: reader = csv.reader(f) next(reader) # ヘッダーを読み飛ばす for row in reader: mf = VmdMorphFrame() # ボーン名 mf.format_name = row[0] # ボーン名(バイト) mf.name = mf.format_name.encode('cp932').decode( 'shift_jis').encode('shift_jis') # フレーム mf.frame = int(row[1]) # 位置 mf.ratio = float(row[2]) morph_frames.append(mf) writer = VmdWriter() if len(morph_frames) > 0 or len(bone_frames) > 0: # ボーン・モーフモーション生成 writer.write_vmd_file(output_vmd_path, "CSV Convert Model", bone_frames, morph_frames, [], [], [], []) print("ボーン・モーフVMD出力成功: %s" % output_vmd_path) if csv_camera_path: # カメラCSV読み込み with open(csv_camera_path, encoding='cp932', mode='r') as f: reader = csv.reader(f) next(reader) # ヘッダーを読み飛ばす for row in reader: cf = VmdCameraFrame() # フレーム cf.frame = int(row[0]) # 位置 cf.position = QVector3D(float(row[1]), float(row[2]), float(row[3])) # 回転(オイラー角) cf.euler = QVector3D(float(row[4]), float(row[5]), float(row[6])) # 距離 cf.length = -(float(row[7])) # 視野角 cf.angle = int(row[8]) # パース cf.perspective = int(row[9]) # 補間曲線 cf.complement = [ int(row[10]), int(row[11]), int(row[12]), int(row[13]), int(row[14]), int(row[15]), int(row[16]), int(row[17]), int(row[18]), int(row[19]), int(row[20]), int(row[21]), int(row[22]), int(row[23]), int(row[24]), int(row[25]), int(row[26]), int(row[27]), int(row[28]), int(row[29]), int(row[30]), int(row[31]), int(row[32]), int(row[33]) ] camera_frames.append(cf) # logger.debug("bf: %s %s", bf.name, bf) if len(camera_frames) > 0: # カメラモーション生成 output_vmd_path = re.sub( r'\.csv$', "_{0:%Y%m%d_%H%M%S}.vmd".format(datetime.now()), csv_camera_path) writer.write_vmd_file(output_vmd_path, "CSV Convert Model", [], [], camera_frames, [], [], []) print("カメラVMD出力成功: %s" % output_vmd_path) except Exception: print("■■■■■■■■■■■■■■■■■") print("■ **ERROR** ") print("■ CSV解析処理が意図せぬエラーで終了しました。") print("■■■■■■■■■■■■■■■■■") print(traceback.format_exc())
def positions_to_frames(pos, vis, frame_num=0, center_enabled=False, head_rotation=None): """convert positions to bone frames""" frames = [] if len(pos) < 17: return frames # センター if center_enabled: bf = VmdBoneFrame() bf.name = b'\x83\x5a\x83\x93\x83\x5e\x81\x5b' # 'センター' bf.frame = frame_num bf.position = pos[7] frames.append(bf) # 上半身 bf = VmdBoneFrame() bf.name = b'\x8f\xe3\x94\xbc\x90\x67' # '上半身' bf.frame = frame_num direction = pos[8] - pos[7] up = QVector3D.crossProduct(direction, (pos[14] - pos[11])).normalized() upper_body_orientation = QQuaternion.fromDirection(direction, up) initial = QQuaternion.fromDirection(QVector3D(0, 1, 0), QVector3D(0, 0, 1)) bf.rotation = upper_body_orientation * initial.inverted() frames.append(bf) upper_body_rotation = bf.rotation # 下半身 bf = VmdBoneFrame() bf.name = b'\x89\xba\x94\xbc\x90\x67' # '下半身' bf.frame = frame_num direction = pos[0] - pos[7] up = QVector3D.crossProduct(direction, (pos[4] - pos[1])) lower_body_orientation = QQuaternion.fromDirection(direction, up) initial = QQuaternion.fromDirection(QVector3D(0, -1, 0), QVector3D(0, 0, 1)) bf.rotation = lower_body_orientation * initial.inverted() lower_body_rotation = bf.rotation frames.append(bf) # 首は回転させず、頭のみ回転させる # 頭 bf = VmdBoneFrame() bf.name = b'\x93\xaa' # '頭' bf.frame = frame_num if head_rotation is None: # direction = pos[10] - pos[9] direction = pos[10] - pos[8] up = QVector3D.crossProduct((pos[9] - pos[8]), (pos[10] - pos[9])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(0, 1, 0), QVector3D(1, 0, 0)) rotation = orientation * initial_orientation.inverted() bf.rotation = upper_body_rotation.inverted() * rotation else: bf.rotation = upper_body_rotation.inverted() * head_rotation frames.append(bf) # 左腕 bf = VmdBoneFrame() bf.name = b'\x8d\xb6\x98\x72' # '左腕' bf.frame = frame_num direction = pos[12] - pos[11] up = QVector3D.crossProduct((pos[12] - pos[11]), (pos[13] - pos[12])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(1.73, -1, 0), QVector3D(1, 1.73, 0)) rotation = orientation * initial_orientation.inverted() # 左腕ポーンの回転から親ボーンの回転を差し引いてbf.rotationに格納する。 # upper_body_rotation * bf.rotation = rotation なので、 bf.rotation = upper_body_rotation.inverted() * rotation left_arm_rotation = bf.rotation # 後で使うので保存しておく if vis[6]: # 左ひじが見えているなら frames.append(bf) # 左ひじ bf = VmdBoneFrame() bf.name = b'\x8d\xb6\x82\xd0\x82\xb6' # '左ひじ' bf.frame = frame_num direction = pos[13] - pos[12] up = QVector3D.crossProduct((pos[12] - pos[11]), (pos[13] - pos[12])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(1.73, -1, 0), QVector3D(1, 1.73, 0)) rotation = orientation * initial_orientation.inverted() # 左ひじポーンの回転から親ボーンの回転を差し引いてbf.rotationに格納する。 # upper_body_rotation * left_arm_rotation * bf.rotation = rotation なので、 bf.rotation = left_arm_rotation.inverted() * upper_body_rotation.inverted() * rotation # bf.rotation = (upper_body_rotation * left_arm_rotation).inverted() * rotation # 別の表現 if vis[6] and vis[7]: # 左ひじと左手首が見えているなら frames.append(bf) # 右腕 bf = VmdBoneFrame() bf.name = b'\x89\x45\x98\x72' # '右腕' bf.frame = frame_num direction = pos[15] - pos[14] up = QVector3D.crossProduct((pos[15] - pos[14]), (pos[16] - pos[15])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(-1.73, -1, 0), QVector3D(1, -1.73, 0)) rotation = orientation * initial_orientation.inverted() bf.rotation = upper_body_rotation.inverted() * rotation right_arm_rotation = bf.rotation if vis[3]: # 右ひじが見えているなら frames.append(bf) # 右ひじ bf = VmdBoneFrame() bf.name = b'\x89\x45\x82\xd0\x82\xb6' # '右ひじ' bf.frame = frame_num direction = pos[16] - pos[15] up = QVector3D.crossProduct((pos[15] - pos[14]), (pos[16] - pos[15])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(-1.73, -1, 0), QVector3D(1, -1.73, 0)) rotation = orientation * initial_orientation.inverted() bf.rotation = right_arm_rotation.inverted() * upper_body_rotation.inverted() * rotation if vis[3] and vis[4]: # 右ひじと右手首が見えているなら frames.append(bf) # 左足 bf = VmdBoneFrame() bf.name = b'\x8d\xb6\x91\xab' # '左足' bf.frame = frame_num direction = pos[5] - pos[4] up = QVector3D.crossProduct((pos[5] - pos[4]), (pos[6] - pos[5])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(0, -1, 0), QVector3D(-1, 0, 0)) rotation = orientation * initial_orientation.inverted() bf.rotation = lower_body_rotation.inverted() * rotation left_leg_rotation = bf.rotation if vis[12]: # 左ひざが見えているなら frames.append(bf) # 左ひざ bf = VmdBoneFrame() bf.name = b'\x8d\xb6\x82\xd0\x82\xb4' # '左ひざ' bf.frame = frame_num direction = pos[6] - pos[5] up = QVector3D.crossProduct((pos[5] - pos[4]), (pos[6] - pos[5])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(0, -1, 0), QVector3D(-1, 0, 0)) rotation = orientation * initial_orientation.inverted() bf.rotation = left_leg_rotation.inverted() * lower_body_rotation.inverted() * rotation if vis[12] and vis[13]: # 左ひざと左足首が見えているなら frames.append(bf) # 右足 bf = VmdBoneFrame() bf.name = b'\x89\x45\x91\xab' # '右足' bf.frame = frame_num direction = pos[2] - pos[1] up = QVector3D.crossProduct((pos[2] - pos[1]), (pos[3] - pos[2])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(0, -1, 0), QVector3D(-1, 0, 0)) rotation = orientation * initial_orientation.inverted() bf.rotation = lower_body_rotation.inverted() * rotation right_leg_rotation = bf.rotation if vis[9]: # 右ひざが見えているなら frames.append(bf) # 右ひざ bf = VmdBoneFrame() bf.name = b'\x89\x45\x82\xd0\x82\xb4' # '右ひざ' bf.frame = frame_num direction = pos[3] - pos[2] up = QVector3D.crossProduct((pos[2] - pos[1]), (pos[3] - pos[2])) orientation = QQuaternion.fromDirection(direction, up) initial_orientation = QQuaternion.fromDirection(QVector3D(0, -1, 0), QVector3D(-1, 0, 0)) rotation = orientation * initial_orientation.inverted() bf.rotation = right_leg_rotation.inverted() * lower_body_rotation.inverted() * rotation if vis[9] and vis[10]: # 右ひざと右足首が見えているなら frames.append(bf) return frames
def read_vmd_file(self, filename): # VMDファイルをバイナリ読み込み self.buffer = open(filename, "rb").read() # vmdバージョン signature = self.unpack(30, "30s") logger.debug("signature %s", signature) motion = VmdMotion() # モデル名 model_bname, model_name = self.read_text(20) logger.debug("model_bname %s", model_bname) motion.model_name = model_name # モーション数 motion.motion_cnt = self.read_uint(4) logger.debug("motion.motion_cnt %s", motion.motion_cnt) # 1F分のモーション情報 for _ in range(motion.motion_cnt): frame = VmdBoneFrame() # ボーン ---------------------- # ボーン名 bone_bname, bone_name = self.read_text(15) frame.name = bone_bname frame.format_name = bone_name logger.debug("name: %s, format_name %s", bone_bname, bone_name) # フレームIDX frame.frame = self.read_uint(4) logger.debug("frame.frame %s", frame.frame) # 位置X,Y,Z frame.position = self.read_Vector3D() logger.debug("frame.position %s", frame.position) # 回転X,Y,Z,scalar frame.rotation = self.read_Quaternion() logger.debug("frame.rotation %s", frame.rotation) logger.debug("frame.rotation.euler %s", frame.rotation.toEulerAngles()) # 補間曲線 frame.complement = self.unpack(64, "64B", True) logger.debug("complement %s", frame.complement) if bone_name not in motion.frames: # まだ辞書にない場合、配列追加 motion.frames[bone_name] = [] # 辞書の該当部分にボーンフレームを追加 motion.frames[bone_name].append(frame) if frame.frame > motion.last_motion_frame: # 最終フレームを記録 motion.last_motion_frame = frame.frame # ソート for k, v in motion.frames.items(): motion.frames[k] = sorted(v, key=lambda u: u.frame) # モーフ数 motion.morph_cnt = self.read_uint(4) logger.debug("motion.morph_cnt %s", motion.morph_cnt) # 1F分のモーフ情報 for _ in range(motion.morph_cnt): morph = VmdMorphFrame() # ボーン ---------------------- # ボーン名 morph_bname, morph_name = self.read_text(15) morph.name = morph_bname morph.format_name = morph_name logger.debug("name: %s, format_name %s", morph_bname, morph_name) # フレームIDX morph.frame = self.read_uint(4) logger.debug("morph.frame %s", morph.frame) # 度数 morph.ratio = self.read_float(4) logger.debug("morph.ratio %s", morph.ratio) if morph_name not in motion.morphs: # まだ辞書にない場合、配列追加 motion.morphs[morph_name] = [] # 辞書の該当部分にモーフフレームを追加 motion.morphs[morph_name].append(morph) # ソート for k, v in motion.morphs.items(): motion.morphs[k] = sorted(v, key=lambda u: u.frame) # カメラ数 motion.camera_cnt = self.read_uint(4) logger.debug("motion.camera_cnt %s", motion.camera_cnt) # 1F分のカメラ情報 for _ in range(motion.camera_cnt): camera = VmdCameraFrame() # フレームIDX camera.frame = self.read_uint(4) logger.debug("camera.frame %s", camera.frame) # 距離 camera.length = self.read_float(4) logger.debug("camera.length %s", camera.length) # 位置X,Y,Z camera.position = self.read_Vector3D() logger.debug("camera.position %s", camera.position) # 角度(オイラー角) camera.euler = self.read_Vector3D() logger.debug("camera.euler %s", camera.euler) # 補間曲線 camera.complement = self.unpack(24, "24B", True) logger.debug("camera.complement %s", camera.complement) # 視野角 camera.angle = self.read_uint(4) logger.debug("camera.angle %s", camera.angle) # パース有無 camera.perspective = self.unpack(1, "B") # カメラを追加 motion.cameras.append(camera) # ソート motion.cameras = sorted(motion.cameras, key=lambda u: u.frame) return motion