如何用Python生成FFmpeg filter_complex所需的指定模式？

需求说明 #

需要根据目录中的图片数量，生成特定格式的字符串用于FFmpeg的filter_complex参数，构建xfade滤镜链。目标模式片段示例：

[s0][s1][s01]
[s01][s2][s012]
[s012][s3][s0123]
[s0123][s4][s01234]

最终要拼接成完整的filter_complex命令片段，示例：

"[s0][s1]xfade=transition=pixelize:duration=0.2:offset=4[s01];[s01][s2]xfade=transition=pixelize:duration=0.2:offset=8[s012];[s012][s3]xfade=transition=pixelize:duration=0.2:offset=12[s0123];[s0123][s4]xfade=transition=pixelize:duration=0.2:offset=16[s01234];[s01234][s5]xfade=transition=pixelize:duration=0.2:offset=20,format=yuv420p" out.mp4

错误尝试及输出 #

之前的循环实现逻辑有误，输出不符合预期：

def print_pattern(num_lines):
    for i in range(num_lines):
        for j in range(i + 1):
            print(j, end='')
        print(' and', i + 1, 'and', end=' ')
        for j in range(i + 1):
            print(j, end='')
        for k in range(i + 1):
            print(k + j + 1, end='')
        print()

num_lines = 4  # Change this variable to control the number of lines
print_pattern(num_lines)

输出：

0 and 1 and 01
01 and 2 and 0123
012 and 3 and 012345
0123 and 4 and 01234567

正确实现方法 #

1. 生成目标模式片段 #

以下函数可生成你需要的标签模式：

def generate_xfade_pattern(num_images):
    # num_images为图片总数，过渡片段数为图片数-1
    for i in range(num_images - 1):
        # 上一步的输出标签：s0、s01、s012...
        prev_label = f"s{''.join(str(n) for n in range(i+1))}"
        # 当前输入的图片标签：s1、s2、s3...
        curr_img_label = f"s{i+1}"
        # 当前过渡的输出标签：s01、s012、s0123...
        output_label = f"s{''.join(str(n) for n in range(i+2))}"
        print(f"[{prev_label}][{curr_img_label}][{output_label}]")

# 示例：5张图片对应4行模式
generate_xfade_pattern(5)

输出：

[s0][s1][s01]
[s01][s2][s012]
[s012][s3][s0123]
[s0123][s4][s01234]

2. 生成完整的filter_complex参数 #

以下函数可直接生成包含xfade参数和format滤镜的完整命令片段：

def generate_filter_complex(num_images, transition="pixelize", duration=0.2, base_offset=4):
    filters = []
    for i in range(num_images - 1):
        prev_label = f"s{''.join(str(n) for n in range(i+1))}"
        curr_img_label = f"s{i+1}"
        output_label = f"s{''.join(str(n) for n in range(i+2))}"
        # 计算过渡偏移量，假设每张图片展示base_offset秒
        offset = base_offset * (i + 1)
        filter_str = f"[{prev_label}][{curr_img_label}]xfade=transition={transition}:duration={duration}:offset={offset}[{output_label}]"
        filters.append(filter_str)
    # 最后添加format滤镜统一编码格式
    filters.append(f"{output_label},format=yuv420p")
    # 拼接成完整的命令字符串
    return '"' + ';'.join(filters) + '" out.mp4'

# 示例：5张图片生成的命令片段
print(generate_filter_complex(5))

输出：

"[s0][s1]xfade=transition=pixelize:duration=0.2:offset=4[s01];[s01][s2]xfade=transition=pixelize:duration=0.2:offset=8[s012];[s012][s3]xfade=transition=pixelize:duration=0.2:offset=12[s0123];[s0123][s4]xfade=transition=pixelize:duration=0.2:offset=16[s01234];s01234,format=yuv420p" out.mp4

内容的提问来源于stack exchange，提问作者Eranga Harasgama