October 18, 2025

Understanding FFMPEG

The basic structure of ffmpeg command is

• first provide the global flags
• the provide the flags for input file
• then use -i flag and provide the input file
• then provide the output options
• then provide the output file

Example

ffmpeg -loglevel error -y \
        -i "before/test.jpeg" \
        -vf "scale=512:512:force_original_aspect_ratio=decrease:flags=lanczos,pad=512:512:(ow-iw)/2:(oh-ih)/2:black,setsar=1:1" \
        -q:v 2 \
        -update 1 \
        "after/test.jpeg"

there is no separate flag for output in ffmpeg the output file is added last in the command

October 21, 2025

FFMPEG commands explation

• -y and -loglevel error global flag for yes to overwrite and only output the logs on error

• -i name of the input image is enter followed by this flag

• -vf vedio filter chain flag and we define chain next, each filter is separated by a commna here we have (scale=…, -> pad=…, -> setsar=…)

• scale=512:512:force_original_aspect_ratio=decrease:flags=lanczos

this is the first chain of transformation in the (vf), here we have taking the scale and providing it 3 input, one is required, the remaining two are optional so 512:512 define the dimenstion of the output image

• force_original_aspect_ratio=decrease

this defined that is if the image is greater than 512 * 512 scale it down put preserve the ratio, for smaller image this would do anything,
hence padding is needed to fill the when the image is not exactly 512 * 512

• flags=lanczos

is this scaling algorthm that ffmpeg will use wiki

• pad=512:512:(ow-iw)/2:(oh-ih)/2:black

this is next filter in the chain, this is where we define the padding, when the image is not exactly 512 * 512 the format is pading=x_dim:y_dim:start_x_coord:start_y_coord:paddingcolor so the dim of the output image is 512 * 512 and the starting coordinate in the x and y direction are

• ow - iw

these are prefined variable in ffmpeg , ow is the output width, iw in the input width so to find the starting x coord we do (ow - iw) / 2 same for the y dim, and black is the padding color

• setsar=1:1

is the pixel ratio of width / height, so 1:1 means that pixel defined here are square

• -q:v 2

-q means quality, here :v mean the vedio channel, since ffmpeg has many channel like :a (audio) etc

• 1

means the compression quality from range of 1 to 31 with 1 means best quality, meaning low compression but higher image size and 31 highly compression low quality we have choosing 2 here ie reasonable quality and image size

• update 1

overwrite the to the same file if present it has for conformation but we have set -y as a global flag to say yes to any such requests