本文介绍用AnimeGANv2实现李云龙视频二次元风格化的方法。先说明AnimeGANv2的改进，接着讲实现思路与目录结构，然后阐述素材准备、环境配置和代码实操过程，包括安装依赖、导入环境、用预训练模型进行视频风格迁移，最后说明合并音视频的步骤，还提供了相关展示和技术详解链接。
☞☞☞AI 智能聊天, 问答助手, AI 智能搜索, 免费无限量使用 DeepSeek R1 模型☜☜☜
李云龙二次元风格化！一键star、fork，你也可以生成这样的团长！
0.效果展示
打滚卖萌求star求fork!
- 视频效果如下，前往B站观看效果更佳：李云龙二次元风格化：
- github开源repo：李云龙二次元风格化
- CSDN技术详解：李云龙二次元风格化！
1.模型简介
1.1AnimeGANv2
本文使用了animeGANv2进行了视频的风格迁移。
animeGANv2，顾名思义，是其前作AnimeGAN的改进版，改进方向主要在以下4点：
- 解决了生成的图像中的高频伪影问题。
- 它易于训练，并能直接达到论文所述的效果。
- 进一步减少生成器网络的参数数量。(现在生成器大小 8.17Mb)
- 尽可能多地使用来自BD电影的新的高质量的风格数据。
效果图参考：
本文则是使用了paddlepaddle预训练好的animeGANv2模型对李云龙名场面视频进行了风格化迁移，详情请看下文分解。
2.实现思路
2.1目录结构
├── codes 代码、数据都在这里面 │ ├── PaddleGAN-develop 代码，基于paddleGAN进行了一些改动，鸣谢 │ ├── videos 所需要转换的视频及音频信息 │ ├── images notebook展示用的图片，没啥用
3.素材准备
首先要找到自己要操作的视频素材，将视频的音频单独提取出来备用
我自己找的资源放在了codes/videos/liyunlong文件夹下，是李云龙名场面：
你咋不敢跟旅长干一架呢！→旅长我给你跪下了
In [ ]
# 安装基本环境!pip install -r codes/PaddleGAN-develop/requirements.txt
Looking in indexes: https://mirror.baidu.com/pypi/simple/
Requirement already up-to-date: xgboost==1.1.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (1.1.0)
Requirement already satisfied, skipping upgrade: scipy in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from xgboost==1.1.0) (1.3.0)
Requirement already satisfied, skipping upgrade: numpy in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from xgboost==1.1.0) (1.16.4)
Looking in indexes: https://mirror.baidu.com/pypi/simple/
Requirement already satisfied: tqdm in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 1)) (4.36.1)
Requirement already satisfied: PyYAML>=5.1 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 2)) (5.1.2)
Requirement already satisfied: scikit-image>=0.14.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 3)) (0.17.2)
Requirement already satisfied: scipy>=1.1.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 4)) (1.3.0)
Requirement already satisfied: opencv-python in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 5)) (4.1.1.26)
Requirement already satisfied: imageio-ffmpeg in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 6)) (0.3.0)
Collecting librosa==0.7.0 (from -r codes/PaddleGAN-develop/requirements.txt (line 7))
Downloading https://mirror.baidu.com/pypi/packages/ad/6e/0eb0de1c9c4e02df0b40e56f258eb79bd957be79b918511a184268e01720/librosa-0.7.0.tar.gz (1.6MB)
|████████████████████████████████| 1.6MB 11.2MB/s eta 0:00:01
Requirement already satisfied: numba==0.48 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 8)) (0.48.0)
Requirement already satisfied: easydict in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from -r codes/PaddleGAN-develop/requirements.txt (line 9)) (1.9)
Collecting munch (from -r codes/PaddleGAN-develop/requirements.txt (line 10))
Downloading https://mirror.baidu.com/pypi/packages/cc/ab/85d8da5c9a45e072301beb37ad7f833cd344e04c817d97e0cc75681d248f/munch-2.5.0-py2.py3-none-any.whl
Requirement already satisfied: numpy>=1.15.1 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (1.16.4)
Requirement already satisfied: matplotlib!=3.0.0,>=2.0.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (2.2.3)
Requirement already satisfied: PyWavelets>=1.1.1 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (1.1.1)
Requirement already satisfied: imageio>=2.3.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (2.6.1)
Requirement already satisfied: pillow!=7.1.0,!=7.1.1,>=4.3.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (7.1.2)
Requirement already satisfied: tifffile>=2019.7.26 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (2021.3.16)
Requirement already satisfied: networkx>=2.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (2.4)
Requirement already satisfied: audioread>=2.0.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (2.1.8)
Requirement already satisfied: scikit-learn!=0.19.0,>=0.14.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (0.22.1)
Requirement already satisfied: joblib>=0.12 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (0.14.1)
Requirement already satisfied: decorator>=3.0.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (4.4.0)
Requirement already satisfied: six>=1.3 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (1.15.0)
Requirement already satisfied: resampy>=0.2.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (0.2.2)
Requirement already satisfied: soundfile>=0.9.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (0.10.3.post1)
Requirement already satisfied: llvmlite<0.32.0,>=0.31.0dev0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from numba==0.48->-r codes/PaddleGAN-develop/requirements.txt (line 8)) (0.31.0)
Requirement already satisfied: setuptools in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from numba==0.48->-r codes/PaddleGAN-develop/requirements.txt (line 8)) (41.4.0)
Requirement already satisfied: python-dateutil>=2.1 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (2.8.0)
Requirement already satisfied: kiwisolver>=1.0.1 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (1.1.0)
Requirement already satisfied: cycler>=0.10 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (0.10.0)
Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (2.4.2)
Requirement already satisfied: pytz in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image>=0.14.0->-r codes/PaddleGAN-develop/requirements.txt (line 3)) (2019.3)
Requirement already satisfied: cffi>=1.0 in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from soundfile>=0.9.0->librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (1.14.0)
Requirement already satisfied: pycparser in /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages (from cffi>=1.0->soundfile>=0.9.0->librosa==0.7.0->-r codes/PaddleGAN-develop/requirements.txt (line 7)) (2.19)
Building wheels for collected packages: librosa
Building wheel for librosa (setup.py) ... done
Created wheel for librosa: filename=librosa-0.7.0-cp37-none-any.whl size=1598346 sha256=f77d09d330640410b176d36a3c3f9f5108aade35b81da8a49d2046ac494bd691
Stored in directory: /home/aistudio/.cache/pip/wheels/81/7e/60/c27574fffbf2f28075dbf4b28c00d3fe3240fefb51d597932e
Successfully built librosa
Installing collected packages: librosa, munch
Found existing installation: librosa 0.7.2
Uninstalling librosa-0.7.2:
Successfully uninstalled librosa-0.7.2
Successfully installed librosa-0.7.0 munch-2.5.0In [ ]
# 导入基本环境import paddle import os import sys sys.path.insert(0,'codes/PaddleGAN-develop')from ppgan.apps import AnimeGANPredictor
codes/PaddleGAN-develop/ppgan/modules/dense_motion.py:116: DeprecationWarning: invalid escape sequence \h """
5.GAN它！
友情提示：此处最好使用GPU环境，cpu推理属实是有点点慢 进行模型的推理：
In [ ]
# 使用paddlepaddle预训练好的animeGANv2模型对视频进行风格迁移：from ppgan.apps import AnimeGANPredictorimport cv2
predictor = AnimeGANPredictor('',None,)
video_src = 'codes/videos/liyunlong/格式工厂混流 亮剑-03+亮剑-03+亮剑-04 00_00_23-.mp4'video_ = cv2.VideoCapture(video_src)
video_name_ = os.path.basename(video_src)
total_frames = video_.get(cv2.CAP_PROP_FRAME_COUNT)
fps_ = video_.get(cv2.CAP_PROP_FPS)print("video {}, fps:{}, total frames:{}...".format(video_name_, fps_, total_frames))
frame_count_ = 0save_per_frames = 1dst_dir = 'codes/videos/liyunlong/'
out_video = cv2.VideoWriter('{}/hayao_{}'.format(dst_dir, video_name_),
cv2.VideoWriter_fourcc(*'DIVX'), int(fps_),
(int(video_.get(3)), int(video_.get(4))))print('now begin...')while True:
ret_, frame_ = video_.read() if not ret_: # or len(fps_list_) == 0:
print('end of video...') break
result_frame = predictor.anime_image_only(frame_) if frame_count_ % save_per_frames == 0:
out_video.write(result_frame)
frame_count_ = frame_count_ + 1
if frame_count_ % 100 == 0: print("{}/{} processed...".format(frame_count_, int(total_frames)), flush=False)[03/17 14:08:11] ppgan INFO: Found /home/aistudio/.cache/ppgan/animeganv2_hayao.pdparams video 格式工厂混流 亮剑-03+亮剑-03+亮剑-04 00_00_23-.mp4, fps:23.0, total frames:5739.0... now begin... 100/5739 processed... 200/5739 processed... 300/5739 processed...
代码解释
6.最终视频
合成最终所需要的视频：
In [ ]
# 合并生成的视频和之前分离的音频：!ffmpeg -i codes/videos/liyunlong/hayao_格式工厂混流 亮剑-03+亮剑-03+亮剑-04 00_00_23-.mp4 -i codes/videos/liyunlong/音频1.aac -c:v copy -c:a aac -strict experimental codes/videos/liyunlong/李云龙二次元化.mp4
这样就大功告成啦~~~
你可以在此基础上：
- 更换你喜欢的视频
- 更换其他paddle预训练好的模型
- 甚至可以尝试自己动手训练定制化的模型！
打滚卖萌求star、fork！
- 视频前往B站观看效果更佳：李云龙二次元风格化：
- github开源repo：李云龙二次元风格化
- CSDN技术详解：李云龙二次元风格化！
