ffmpeg录制窗口应用时叠加滤镜、混音等处理的问题求助

Let's get your overlay and audio processing sorted out—you can choose between a one-step corrected command or a two-step workflow (record first, then apply effects) depending on what feels easier to debug and use.

Why Your Complex Command Failed #

Your original command had a few critical issues that broke it:

• You never referenced the captured desktop video stream ([1:v] from the gdigrab input) in your filter chain—right now you were only manipulating the filter.mp4 video, ignoring the app recording entirely.
• Typos like -b:30 (should be -b:v for video bitrate) and mixing output format (flv) with file extension (mp4) caused conflicts.
• The audio filter chain only processed filter.mp4's audio, not mixing it with your captured app audio.

Solution 1: One-Step Corrected Command #

This fixes stream references, typos, and adds proper mixing of your captured audio with the processed filter audio:

ffmpeg -rtbufsize 1500M \
-f dshow -i audio="virtual-audio-capturer" \
-f gdigrab -framerate 30 -draw_mouse 0 -i title="Main" \
-stream_loop 999 -i "filter/filter.mp4" \
-loop 999 -i "filter/nds_bg.mp4" \
-filter_complex "
# Crop and resize the captured app video to final output size
[1:v]crop=850:480:156:100,scale=640:360[v_captured];
# Process filter.mp4 for overlay blending
[2:v]scale=385:216,setdar=16/9[v1];
[2:v]scale=385:216,setdar=16/9[v2];
[v1][v2]blend=shortest=1:all_opacity=1[v3];
# Prepare the background overlay video
[3:v]setpts=N/(FRAME_RATE*TB),scale=640:360[v4];
# Layer overlays onto the captured app video
[v_captured][v4]overlay=shortest=1:x=20:y=130[v_temp];
[v_temp][v3]overlay=shortest=1[v_final];
# Process filter audio and mix with captured app audio
[2:a]aformat=sample_fmts=fltp:sample_rates=44100:channel_layouts=stereo,asetrate=8.5/10*44100,atempo=10/8.5,lowpass=f=2500,highpass=f=400,volume=3,bass=g=-30,equalizer=f=10.5:width_type=o:width=1:g=-30,equalizer=f=31.5:width_type=o:width=1:g=-30,equalizer=f=63:width_type=o:width=1:g=-10,equalizer=f=125:width_type=o:width=1:g=-20,equalizer=f=250:width_type=o:width=1:g=-1.5,equalizer=f=500:width_type=o:width=1:g=-20,equalizer=f=1000:width_type=o:width=1:g=-20,equalizer=f=8000:width_type=o:width=3:g=1,equalizer=f=16000:width_type=o:width=3:g=1[v_filter_a];
[0:a][v_filter_a]amix=inputs=2:duration=shortest[a_final]
" \
-map "[v_final]" -map "[a_final]" \
-vcodec libx264 -pix_fmt yuv420p -r 26 -g 30 -b:v 800k \
-acodec libmp3lame -b:a 128k -ar 44100 \
-preset:v ultrafast -bufsize 3000k -s 640x360 \
-shortest -y out.mp4

Key fixes here:

• Explicitly references your captured app video ([1:v]) and audio ([0:a]) in the filter chain
• Separates each filter step for readability
• Adds proper audio mixing between your captured audio and processed filter audio
• Corrects typos and format mismatches

Solution 2: Two-Step Workflow (Record First, Process Later) #

If you prefer to split recording and processing (great for debugging), follow these steps:

Step 1: Record the Raw App & Audio #

Use your working simple command to capture the base video:

ffmpeg -rtbufsize 1500M -f dshow -i audio="virtual-audio-capturer" -f gdigrab -framerate 30 -draw_mouse 0 -i title="Main" -vf crop=850:480:156:100 -pix_fmt yuv420p -profile:v baseline -y raw_recording.mp4

Step 2: Apply Overlays & Audio Effects #

Take the raw recording and add your overlays/audio processing:

ffmpeg -i raw_recording.mp4 \
-stream_loop 999 -i "filter/filter.mp4" \
-loop 999 -i "filter/nds_bg.mp4" \
-filter_complex "
# Resize raw recording to final output size
[0:v]scale=640:360[v_captured];
# Process filter.mp4 for overlay blending
[1:v]scale=385:216,setdar=16/9[v1];
[1:v]scale=385:216,setdar=16/9[v2];
[v1][v2]blend=shortest=1:all_opacity=1[v3];
# Prepare background overlay video
[2:v]setpts=N/(FRAME_RATE*TB),scale=640:360[v4];
# Layer overlays onto raw recording
[v_captured][v4]overlay=shortest=1:x=20:y=130[v_temp];
[v_temp][v3]overlay=shortest=1[v_final];
# Process filter audio and mix with raw recording audio
[1:a]aformat=sample_fmts=fltp:sample_rates=44100:channel_layouts=stereo,asetrate=8.5/10*44100,atempo=10/8.5,lowpass=f=2500,highpass=f=400,volume=3,bass=g=-30,equalizer=f=10.5:width_type=o:width=1:g=-30,equalizer=f=31.5:width_type=o:width=1:g=-30,equalizer=f=63:width_type=o:width=1:g=-10,equalizer=f=125:width_type=o:width=1:g=-20,equalizer=f=250:width_type=o:width=1:g=-1.5,equalizer=f=500:width_type=o:width=1:g=-20,equalizer=f=1000:width_type=o:width=1:g=-20,equalizer=f=8000:width_type=o:width=3:g=1,equalizer=f=16000:width_type=o:width=3:g=1[v_filter_a];
[0:a][v_filter_a]amix=inputs=2:duration=shortest[a_final]
" \
-map "[v_final]" -map "[a_final]" \
-vcodec libx264 -pix_fmt yuv420p -r 26 -g 30 -b:v 800k \
-acodec libmp3lame -b:a 128k -ar 44100 \
-preset:v ultrafast -bufsize 3000k -s 640x360 \
-shortest -y out.mp4

This way you can verify the raw recording is correct before adding effects, making it easier to spot issues if something goes wrong.

内容的提问来源于stack exchange，提问作者user3181176