Blender mesh 创建点线面 uOQ-CPcaqMo AxazJi3x6js

任务：创建一个半径为2圆环，圆环上均布32个点，显示为点或者线。

1. 点

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

# give Python access to Blender's functionality
import bpy
# extend Python's math functionality
import math
# extend Python's print functionality
import pprint

# initialize paramaters
vert_count = 32  # show with 16 and 64
angle_step = math.tau / vert_count
radius = 2

# create a list of vert coordinates
vert_coordinates = list()

# repeat code in a loop
for i in range(vert_count):

    # calculate current current_angle
    current_angle = angle_step * i

    # calculate coordinate
    x = radius * math.cos(current_angle)
    y = radius * math.sin(current_angle)

    # visualize what we are doing
    bpy.ops.mesh.primitive_ico_sphere_add(radius=0.05, location=(x, y, 0))

    # add current coordinate to list
    vert_coordinates.append((x, y, 0))

pprint.pprint(vert_coordinates)

2. 线

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

# give Python access to Blender's functionality
import bpy
# extend Python's math functionality
import math
import pprint


def get_circle_verts(vert_count, radius):
    # initialize paramaters
    angle_step = math.tau / vert_count

    # create a list of vert coordinates
    vert_coordinates = list()

    # repeat code in a loop
    for i in range(vert_count):

        # calculate current current_angle
        current_angle = angle_step * i

        # calculate coordinate
        x = radius * math.cos(current_angle)
        y = radius * math.sin(current_angle)

        # visualize what we are doing
        # bpy.ops.mesh.primitive_ico_sphere_add(radius=0.05, location=(x, y, 0))

        # add current coordinate to list
        vert_coordinates.append((x, y, 0))

    return vert_coordinates


def create_circle_mesh(coordinates, vert_count):
    verts = coordinates
    faces = []
    edges = []

    for i in range(vert_count - 1):
        current_vert_index = i
        next_vert_index = i + 1
        edges.append((current_vert_index, next_vert_index))

    edges.append((vert_count - 1, 0))

    # create a mesh from the vert, edge, and face data
    mesh_data = bpy.data.meshes.new("circle_data")
    mesh_data.from_pydata(verts, edges, faces)

    # create a object using the mesh data
    mesh_obj = bpy.data.objects.new("circle_object", mesh_data)

    bpy.context.collection.objects.link(mesh_obj)


# initialize paramaters
vert_count = 32
radius = 2

coordinates = get_circle_verts(vert_count, radius)

mesh_obj = create_circle_mesh(coordinates, vert_count)

Final Code:

circle_mesh_part_1 https://gist.github.com/CGArtPython/6e98beb424b6c58bbda0469b2c0434a9
circle_mesh_part_2 https://gist.github.com/CGArtPython/a4dff81d8c233601759dffd48917f296

ref:
Beginner Blender Python Exercise: Circle mesh from scratch
https://www.youtube.com/watch?v=uOQ-CPcaqMo
https://www.youtube.com/watch?v=AxazJi3x6js

Blender Python 配置 (on Windows) YUytEtaVrrc

1. 安装 vscode、 blender

1
2
3
4
5
6

scoop install vscode
scoop install blender

# blender 和内置的 python 路径如下：
# blender path : C:\Users\xxxx\scoop\apps\blender\4.2.3\blender.exe
# blender build-in python path : C:\Users\shenb\scoop\apps\blender\4.2.3\4.2\python\bin\python.exe

在 vscode 中选择 python 解释器，注意选择 blender 内置的 python 路径。

2. 安装自动补全 fake-bpy-module

安装时注意要选用 blender 内置的 python -m pip 安装。

ref: https://github.com/nutti/fake-bpy-module

1
2
3
4
5

C:\Users\shenb\scoop\apps\blender\4.2.3\4.2\python\bin\python.exe -m pip install fake-bpy-module

# 安装常用的其他包：
# C:\Users\shenb\scoop\apps\blender\4.2.3\4.2\python\bin\python.exe -m pip install pandas
# C:\Users\shenb\scoop\apps\blender\4.2.3\4.2\python\bin\python.exe -m pip install matplotlib

3. 安装 vscode 插件： Blender Development

ref: https://marketplace.visualstudio.com/items?itemName=JacquesLucke.blender-development

使用时打开命令面板 Ctrl+Shift+P:

• Blender: Start command 启动一个 Blender 程序实例。
• Blender: Run Script command 执行当前的 python 脚本。

ref:
5 Steps to setup VSCode for Blender Python (on Windows)
https://www.youtube.com/watch?v=YUytEtaVrrc

视频内字幕提取

video-subtitles-ocr

视频字幕提取，基于 opencv 和 paddleocr

视频内字幕提取

这里是针对内封了硬字幕的视频，字幕已经成为了画面的一部分。

思路：简单用 opencv 提取视频内的所有帧，现在可以使用 paddleocr，更加方便一点对图片进行 ocr 识别。

0. 首先需要配置一下

0.1 安装 python 库

• opencv-python
• scikit-image
• paddleocr

ref: https://paddlepaddle.github.io/PaddleOCR/latest/quick_start.html#1-paddlepaddle

word 不能直接导入 eps 格式的图片，所以先将 eps 图片转换成 png 格式。

这里使用 python 调用 PIL 读取 eps 图片，直接保存为 png 图片

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

# pip install pillow
# scoop install ghostscript

from PIL import Image

import os

for parent, dirnames, filenames in os.walk(os.getcwd()):
    for filename in filenames:
        ext = os.path.splitext(filename)[1]
        if ext != '.eps':
          continue

        eps_image = Image.open(filename)
        size = eps_image.size        
        print(filename, size)

        eps_image.load(scale=10)    # 分辨率增大到10倍
        eps_image.save(filename+'.png', quality=100)

• PIL 默认采用 Ghostscript 处理 eps 图片， 所以先确保安装了 Ghostscript。 scoop install ghostscript
• eps 为矢量图，默认情况下其尺寸较小，导致保存成图片式分辨率较低，eps_image.load(scale=10) 将分辨率增大到10倍后保存。

下载 bili 视频

1
2
3
4
5
6
7

pip install you-get
# you-get https://www.bilibili.com/video/BV1234567890

pip install yt-dlp
# yt-dlp https://www.bilibili.com/video/BV1234567890

# yt-dlp "https://www.youtube.com/watch?v=GGGGGGGGGGGG" --proxy http://127.0.0.1:7890

视频转音频

1
2
3
4
5
6

scoop install ffmpeg
ffmpeg -i xxx.mp4
# checkout the audio video format
ffmpeg -i xxx.mp4 -acodec copy xxx.aac
ffmpeg -i xxx.mp4 -acodec copy xxx.mp3
ffmpeg -i xxx.mkv -acodec copy xxx.opus

音频转文字 📘

https://github.com/openai/whisper

1. ** whisper **：

   1 2 3 4

   pip install -U openai-whisper whisper audio.flac audio.mp3 audio.wav --model small --device cpu --language zh

2. ** python **：
   使用 Python 运行 main.py 脚本。

   1 2 3 4 5 6

   import whisper model = whisper.load_model("small", device="cpu") result = model.transcribe(f"123.mp3") print(result["text"]))

cat-catch 设置 N_m3u8DL-RE

ref https://github.com/nilaoda/N_m3u8DL-RE
ref https://github.com/xifangczy/cat-catch
ref https://github.com/corbamico/m3u8dl-invoke/

1.1 安装 n-m3u8dl-re

1

scoop install n-m3u8dl-re_x

1.2 新建 m3u8dl 程序

先用 powershell 写一段脚本 n_m3u8dl-re_protocol.ps1 ，然后转换成可执行文件 n_m3u8dl-re_protocol.exe

• 脚本

  1 2 3 4 5 6 7 8 9 10 11 12 13 14

  # 接收m3u8://xxxxxx param($m3u8url) Write-Host "m3u8url: $m3u8url" $bs64 = $m3u8url.replace('m3u8dl://', '').replace('/', '') $httpurl = [Text.Encoding]::ASCII.GetString([Convert]::FromBase64String($bs64)) Write-Host "httpurl: $httpurl" $path = 'c:\Users\xxxxx\scoop\shims\N_m3u8DL-re.exe' $cmd = $path + ' ' + $httpurl Write-Host "$cmd" Start-Process "$path" "$httpurl" -NoNewWindow Start-Sleep 30

• 转换

  1 2 3

  Install-Module ps2exe ps2exe n_m3u8dl-re_protocol.ps1 n_m3u8dl-re_protocol.exe

matplotlib marker color 字符对照

ref: https://matplotlib.org/stable/api/colors_api.html
ref: https://matplotlib.org/stable/gallery/color/named_colors.html

1
2
3
4
5
6
7
8
9
10
11

import matplotlib.pyplot as plt
import matplotlib.colors as mcolors

for (k, v) in mcolors.BASE_COLORS.items():
    v = "#" + "".join(format(round(val * 255), "02x") for val in v)
    print(f'<font color="{v}"> {k:<10} ■  </font>')

for (k, v) in mcolors.TABLEAU_COLORS.items():
    print(f'<font color="{v}"> {k:<10} ■ </font>')

mcolors.BASE_COLORS (8)

████ b
████ g
████ r
████ c
████ m
████ y
████ k
████ w

mcolors.TABLEAU_COLORS (10)

████ tab:blue
████ tab:orange
████ tab:green
████ tab:red
████ tab:purple
████ tab:brown
████ tab:pink
████ tab:gray
████ tab:olive
████ tab:cyan

Python f-string 备忘单

ref https://fstring.help/cheat/

ref https://cheatography.com/brianallan/cheat-sheets/python-f-strings-basics/
ref https://cheatography.com/brianallan/cheat-sheets/python-f-strings-number-formatting/

数字

integer = 123
number_ = 9876.6789
percent = 0.98765

string = “Python”

Python 正则表达式备忘表

ref https://regexlearn.com/zh-cn/cheatsheet

ref https://www.w3schools.com/python/python_regex.asp

radioactivedecay 用于放射性核素衰变的简单分析计算的开源包，支持放射性核素、亚稳态和分支衰变的衰变链。

https://github.com/radioactivedecay/radioactivedecay

安装：pip install radioactivedecay

1. 绘制衰变链

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

#%%
import radioactivedecay as rd
import matplotlib.pyplot as plt

# 创建一个放射性核素： Nuclide
ncld = rd.Nuclide('U-235')
half = ncld.half_life('readable')    # 半衰期
prog = ncld.progeny()                # 衰变产物
frac = ncld.branching_fractions()    # 占比
mode = ncld.decay_modes()            # 衰变射线类型

print(ncld, mode, prog, frac, half, sep='\n')

# 绘制衰变链
ncld.plot()

#%%
# 列出多个核算衰变类型
NUCs = ['U-235',
        'Pu-239']

for name in NUCs:
    ncld = rd.Nuclide(name)
    half = ncld.half_life('readable')
    prog = ncld.progeny()
    frac = ncld.branching_fractions()
    mode = ncld.decay_modes()
    print(ncld, mode, prog, frac, half, sep='\n')

    fig, ax = ncld.plot()
    fig.savefig(f'decay_chains_{name}.png', dpi=300)

pandas 按条件筛选和设置表格样式。

https://pandas.pydata.org/docs/user_guide/indexing.html#boolean-indexing

https://pandas.pydata.org/docs/reference/style.html#style

pandas 合并数据常用的两个函数：

• pd.concat https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.concat.html

• pd.merge https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.merge.html

刚好有一个对比 RNN 和 LSTM 神经网络运行结果的场景分别用到了这两个函数：

• 采用 concat 合并两个csv表； csv表为两次运行结果的，列名基本一致。
• 采用 merge 合并模型运行条件类似的表， 用于对比数据

一、WSL 安装 Ubuntu22.04

1. 启用 WSL

2. 打开 Microsoft Store，安装 Ubuntu22.04

3. 设置源，更新
   ref： https://mirrors.tuna.tsinghua.edu.cn/help/ubuntu/

更新：sudo apt update

升级：sudo apt upgrade

4. 安装 ZSH、oh-my-zsh

参考 zsh & oh-my-zsh 安装配置

二、安装 openmc

ref： https://docs.openmc.org/en/stable/quickinstall.html

1. 参照 Installing from Source on Ubuntu，安装 openmc。

   安装完成后，运行 openmc 试一下, 提示错误： 找不到材料数据卡 cross_sections.xml。

2. 下载材料数据集

   1 2 3 4

   git clone https://github.com/openmc-dev/data openmc-dev_data cd openmc-dev_data python3 convert_nndc71.py

   阅读全文 »

ref: https://github.com/ScoopInstaller/Scoop/issues/4390

0. 问题： scoop 本身不支持安装包有解压密码的自动解压。

scoop 本身不支持安装包有解压密码的自动解压。因此在安装 MAS 时会卡在解压处，无法继续安装。

1. 7z 命令手册

7z 命令是支持的： 7z x FILENAME.7z -pPWD -oOUTDIR

2. 查 Scoop 源码的解压函数

找到 Scoop 源码里的 “~\scoop\apps\scoop\current\lib\decompress.ps1” 文件。
文件封装了 Expand-7zipArchive 函数命令， 但参数没有能将密码传入的方法。

3. 试 Scoop 的 pre_install

Scoop Wiki 里提到了 pre_install 等说法。

在 MAS.json 中加入：

1

"pre_install": "7z x $dir\\$fname -p1234 $('-o' + $dir)"，

尝试了很多次还是不起作用，Scoop的策略是遇到压缩包先执行解压、再做后续的事情。所以这一句不会被执行，安装程序依旧卡在解压处，无法继续安装。

0. 设置

板子型号：ESP-12E

0.1 安装 windows 驱动：CP2102

• https://www.silabs.com/developers/usb-to-uart-bridge-vcp-drivers?tab=downloads
• CP210x_Universal_Windows_Driver

0.1 安装 esptool

1
2
3
4
5
6

pip install esptool

# usb 连接电脑，
# "我的电脑"->"管理"->"设备管理器"->"端口"
# 检查端口号
esptool chip_id 

1. 客户端 VSCode 安装 Remote 插件

安装插件, 略

添加远程连接 aaa@192.168.1.xxx， 连接、等待， 输入密码

2. 连接时可能连不上，并出现警告

警告： The remote host may not meet VS Code Server's prerequisitesfor glibc and libstdc++

从在线日历获取事件

Outlook 查看日历链接： 设置-日历-共享日历-发布日历，可以看到如下链接：

“https://outlook.live.com/owa/calendar/d82e6afe-XXXX-XXXX/cid-YYYY/calendar.ics“

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

from ics import Calendar
import requests
 
# Parse the URL
url = "https://outlook.live.com/owa/calendar/d82e6afe--XXXX-XXXX/cid-YYYY/calendar.ics"
cal = Calendar(requests.get(url).text)


#%%
# Print all the events
events = sorted(cal.events)
for event in events:
    print(event.begin , event.name )

#%%
timeline = cal.timeline
for tl in timeline:
    print(tl.begin , tl.name )

# %%
# 获取今天及之后的事件
from ics import Calendar
import requests
import arrow

cal = Calendar(requests.get(url).text)

today = arrow.now()
print(today)
today = arrow.now().replace(hour=0, minute=0, second=0, microsecond=0)
print(today)
today = arrow.now().span('day')[0]
print(today)

evts = cal.timeline.start_after(today)

for evt in evts:
    print(evt.begin.strftime("%Y-%m-%d %a %H:%M"), evt.name )

ref: (https://github.com/ScoopInstaller/Scoop/wiki/App-Manifests)

尝试自己制作一个安装配置文件， 软件是 N_m3u8DL-RE。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

{
    "version": "v0.2.0-beta",
    "description": "Cross-Platform, modern and powerful stream downloader for MPD/M3U8/ISM",
    "homepage": "https://github.com/nilaoda/N_m3u8DL-RE",
    "license": "MIT",
    "suggest": {
        "ffmpeg": [
            "ffmpeg",
            "ffmpeg-nightly"
        ]
    },
    "url": "https://github.com/nilaoda/N_m3u8DL-RE/releases/download/v0.2.0-beta/N_m3u8DL-RE_Beta_win-x64_20230628.zip",
    "hash": "9e3133f03e112cc57f34bf04234b7857ae7c997c214f0eb510a120739c388652",
    "bin": [
        "N_m3u8DL-RE_Beta_win-x64/N_m3u8DL-RE.exe",
        [
            "N_m3u8DL-RE.exe",
            "N_m3u8DL_RE"
        ]
    ],
    "checkver": {
        "url": "https://api.github.com/repos/nilaoda/N_m3u8DL-RE/releases",
        "regex": "releases/download/(?<tag>[vV]?[\\w-.]+)/N_m3u8DL-RE_Beta_win-x64_(?<date>[\\d.-]+).zip",
        "replace": "${tag}"
    },
    "autoupdate": {
        "url": "https://github.com/nilaoda/N_m3u8DL-RE/releases/download/$matchTag/N_m3u8DL-RE_Beta_win-x64_$matchDate.zip"
    }
}

1 安装 zsh

1
2
3
4
5

sudo apt install zsh

# echo $SHELL           # 显示当前 shell 类型
# cat /etc/shells       # 显示可使用的 shell
# chsh -s /bin/zsh      # 将 zsh 设为默认终端

2 安装 ohmyzsh

ref: https://ohmyz.sh/

• 用 wget 安装。

  1	sh -c "$(wget https://raw.github.com/ohmyzsh/ohmyzsh/master/tools/install.sh -O -)"

有时候会出现 443 网路错误，可手动下载、安装 install.sh

1	nano zsh-install.sh

打开 https://github.com/ohmyzsh/ohmyzsh/blob/master/tools/install.sh
将安装代码全部复制进去

1 2	sudo chmod +x zsh-install.sh ./zsh-install.sh

3 安装 ohmyzsh （gitee 源）

ref: https://gitee.com/mirrors/oh-my-zsh.git

1

sh -c "$(wget https://gitee.com/mirrors/oh-my-zsh/raw/master/tools/install.sh -O -)"

4 配置 ohmyzsh （gitee 源）

• 打开配置文件

  1	nano ~/.zshrc

• 修改如下：

  1 2 3 4 5

  # 主题 ZSH_THEME="ys" # 插件 plugins=(git zsh-autosuggestions zsh-syntax-highlighting )

• 也可以用命令修改：

  1 2	sed -i 's/^ZSH_THEME=.*/ZSH_THEME="ys"/g' .zshrc sed -i 's/^plugins=.*/plugins=(git zsh-autosuggestions zsh-syntax-highlighting)/g' .zshrc

• 下载插件

  1 2 3

  git clone https://gitee.com/mirrors/zsh-autosuggestions ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/zsh-autosuggestions git clone https://gitee.com/mirrors/zsh-syntax-highlighting ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/zsh-syntax-highlighting

  ref: https://github.com/zsh-users/zsh-autosuggestions

ref: https://github.com/zsh-users/zsh-syntax-highlighting

• 加载配置文件， 使生效

  1	source ~/.zshrc

===

1
2
3
4
5
6
7
8
9
10
11
12

sudo apt install zsh -y
sudo apt install git -y

sh -c "$(wget https://gitee.com/mirrors/oh-my-zsh/raw/master/tools/install.sh -O -)"

sed -i 's/^ZSH_THEME=.*/ZSH_THEME="ys"/g' .zshrc
sed -i 's/^plugins=.*/plugins=(git zsh-autosuggestions zsh-syntax-highlighting)/g' .zshrc

git clone https://gitee.com/mirrors/zsh-autosuggestions ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/zsh-autosuggestions
git clone https://gitee.com/mirrors/zsh-syntax-highlighting ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/zsh-syntax-highlighting

source ~/.zshrc

用 raspberry pico w 和墨水屏做一个 mnist-clock

1. 墨水屏

• 型号：Pico-ePaper-2.9
• https://www.waveshare.net/wiki/Pico-ePaper-2.9
• 工作电压：3.3V/5V
• 通信接口：SPI
• 分辨率：296 x 128
• 显示颜色：黑、白
• 局部刷新：0.3s
• 全局刷新 ：2s

2. pico w 基本配置

2.1 刷固件

• https://micropython.org/download/rp2-pico-w/

• 下载类似rp2-pico-w-20221107-unstable-v1.19.1-616-g5987130af.uf2

• 首次将 pico w 连接电脑之后，会以U盘形式弹出来，把固件拖进去。

• 已经刷过固件的，按着板子上的 BOOTTSEL 按钮连接电脑。

重新上电，固件就自动刷好了。

2.2 开发环境 thonny

• 安装 thonny, scoop install thonny
• Thonny IED： 打开 ‘运行’ - ‘配置解释器’， 配置一下
• 运行一下板载的 LED 测试程序：

1
2
3
4
5
6
7
8
9

from machine import Pin 
import time

led = Pin('LED', Pin.OUT)
while(True):
    led.on()
    time.sleep(1)
    led.off()
    time.sleep(1)

可以看到板载的 LED 在闪烁。