如何用ffmpeg为图片拼接的幻灯片添加对应创建日期的文本

To add a unique creation date overlay to each image in your concatenated video, you’ll need to process each image individually with the drawtext filter before combining them. Here’s a practical, automated solution using a Bash script tailored to your setup:

Prerequisites #

First, install exiftool (to reliably pull EXIF creation dates from images) if you don’t already have it:

sudo apt install libimage-exiftool-perl  # Debian/Ubuntu
# For RHEL/CentOS: sudo dnf install perl-Image-ExifTool

Bash Script to Generate and Run the FFmpeg Command #

This script reads your existing files.txt (with absolute image paths), extracts each image’s creation date, and builds an FFmpeg command that applies the overlay to every image before concatenation:

#!/bin/bash

# Read the list of images from your existing files.txt
mapfile -t images < files.txt

# Initialize variables for FFmpeg inputs and filter complex
ffmpeg_inputs=""
filter_complex=""
concat_streams=""

# Configure drawtext settings (customize these as needed)
FONT_PATH="/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf"
FONT_SIZE=24
FONT_COLOR="white"
TEXT_POSITION="x=10:y=10"  # Top-left corner; use x=(w-text_w)/2:y=(h-text_h)/2 for center
SHADOW_SETTINGS="shadowcolor=black:shadowx=2:shadowy=2"
IMAGE_DURATION="3"  # Seconds each image is displayed

# Loop through each image to build the filter chain
for idx in "${!images[@]}"; do
    img_path="${images[$idx]}"
    
    # Extract creation date (fallback to file modification time if EXIF data is missing)
    creation_date=$(exiftool -d "%Y-%m-%d %H:%M:%S" -CreateDate -s3 "$img_path")
    if [[ -z "$creation_date" ]]; then
        creation_date=$(stat -c "%y" "$img_path" | cut -d' ' -f1,2)
    fi
    
    # Add image input with duration
    ffmpeg_inputs+=" -i \"$img_path\" -t $IMAGE_DURATION"
    
    # Add drawtext filter for this image
    filter_complex+="[${idx}:v]drawtext=fontfile=$FONT_PATH:text='${creation_date}':fontsize=$FONT_SIZE:fontcolor=$FONT_COLOR:$TEXT_POSITION:$SHADOW_SETTINGS[v${idx}];"
    
    # Track the filtered stream for concatenation
    concat_streams+="[v${idx}]"
done

# Final concat filter to combine all processed streams
filter_complex+="${concat_streams}concat=n=${#images[@]}:v=1:a=0[v]"

# Build and execute the FFmpeg command
ffmpeg_cmd="ffmpeg $ffmpeg_inputs -filter_complex \"$filter_complex\" -map \"[v]\" -c:v libx264 -pix_fmt yuv420p out.mp4"
eval $ffmpeg_cmd

Customization Tips #

• Font Settings: Adjust FONT_PATH to a font installed on your system (common options include /usr/share/fonts/truetype/ubuntu/Ubuntu-B.ttf or /usr/share/fonts/truetype/freefont/FreeSansBold.ttf).
• Text Position: Change TEXT_POSITION to place the date where you want it:
  • Bottom-right: x=w-text_w-10:y=h-text_h-10
  • Centered: x=(w-text_w)/2:y=(h-text_h)/2
• Duration: Modify IMAGE_DURATION to change how long each image is shown (in seconds).
• Resolution Consistency: If your images have different sizes, add a scale filter before drawtext to standardize them. For example:

  filter_complex+="[${idx}:v]scale=1920:1080,drawtext=...[v${idx}];"
  

  Replace 1920:1080 with your desired resolution.

How It Works #

• The script reads your files.txt to get all image paths.
• For each image, it extracts the creation date using exiftool (or falls back to file modification time if EXIF data is missing).
• It builds an FFmpeg filter complex where each image is processed with its own drawtext overlay.
• Finally, all processed image streams are concatenated into a single video stream, which is encoded to MP4 using H.264.

内容的提问来源于stack exchange，提问作者Hauke