| 金思雨,覃京燕.基于计算机图像风格迁移的音乐可视化智能设计研究[J].包装工程,2020,41(16):193-198. |
| 基于计算机图像风格迁移的音乐可视化智能设计研究 |
| Music Visualization Intelligent Design Based on Image Style Migration |
| 投稿时间：2020-03-28 修订日期：2020-08-20 |
| DOI：10.19554/j.cnki.1001-3563.2020.16.029 |
| 中文关键词: 通感 图像风格迁移 音乐可视化 智能设计 交互艺术 |
| 英文关键词: synesthesia image style migration music visualization intelligent design interactive art |
| 基金项目: |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 基于Processing程序识别音乐特征和图像特征，控制图像风格迁移的实时生成效果，尝试将人对音乐和图像的联觉进行可视化表达。方法 预处理阶段分别分析输入的音乐文件和图像文件，提取音乐文件中左右声道的实时强度等特征，识别图像文件中色彩空间及单位像素的亮度。数据交互阶段，建立随机生成效果，并以音乐特征和图像特征为因变量控制粒子生成的位置、时机等。画面生成阶段，通过蠕虫效果对每个有效粒子进行动态视觉展示，根据实时更新的有效特征值指导渲染效果，主要控制渲染过程中新粒子的位置和数量，以及粒子的形状、颜色和速度等。结果 音乐文件与图像文件同时对图像风格迁移的可视化表达产生影响，不同搭配方式生成视觉画面和风格属性是独特的唯一的；同时，图像重建过程动态可视，屏幕从空白到完成人像效果呈现的完整过程与音乐播放进程实时同步输出。 结论 将传统的音乐可视化和图像风格迁移相结合，为音乐可视化增加了具体内容表达，为图像风格迁移增加了时序表达；这种实时信息可视化是人工智能模仿人类通感的一种艺术表达形式，同时是一种表演艺术。这种可视化效果可以帮助用户用音乐生成独一无二的个性肖像；也可被广泛应用于艺术家表达音乐和视觉的关联性。 |
| 英文摘要: |
| The work aims to identify music features and image features based on the Processing program to control the real-time generation effect of image style migration and try to visualize the synesthesia of music and image. In the pre-processing stage, the input music file and image file were analyzed respectively, and the real-time intensity of left and right channels in the music file was extracted to identify the color space and the brightness of a unit pixel in the image file. In the data interaction stage, the effect of random generation was established, and the position and timing of particle generation were controlled by music feature and image feature as dependent variables. In the image generation stage, each effective particle was displayed dynamically by worm effect, and the rendering effect was guided by the real-time updated effective eigenvalues, mainly controlling the position and number of new particles in the rendering process, as well as the shape, color, and speed of particles. Both music files and image files influenced the visual expression of image style migration. The different collocation methods were distinct and unique in generating visual images and style attributes. At the same time, the process of image reconstruction was dynamically visible, and the complete process from the screen to the completion of the portrait effect was synchronized with the music playing process in real-time. The combination of traditional music visualization and image style migration adds specific content expression for music visualization and time sequence expression for image style migration. This real-time information visualization is an art expression form of artificial intelligence imitating human synesthesia, and also a performance art. This visualization effect can help users create the unique personalities of the music, and can also be widely applied to artists’ expression of the relevance of music and vision. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
