Add fixers for glitch and audio replacing

.gitignore

@@ -2,3 +2,5 @@
 *.mp4
 out/*
 !out/keepme
+*.wav
+.DS_Store

add-glitch.sh

@@ -0,0 +1 @@
+ffmpeg -hwaccel videotoolbox -i glitch.mp4 -i video-with-new-audio.mp4 -i glitch.mp4 -filter_complex "[0:v:0][0:a:0][1:v:0][1:a:0][2:v:0][2:a:0]concat=n=3:v=1:a=1[vv][aa]" -map "[vv]" -map "[aa]" -c:v libx264 -r 50 -c:a aac -strict experimental -b:a 320k output.mp4

demux.sh

@@ -1,9 +0,0 @@
-echo "generating the demuxers..."
-
-# generuje ścieżki do złożenia przez ffmpega:
-ts-node generate-demuxer.ts >out/demuxer.txt
-
-# używa demuxer.txt żeby skleić końcowe video z dźwiękiem:
-echo generowanie całości
-ffmpeg -y -f concat -safe 0 -i out/demuxer.txt -r $framerate -video_track_timescale $timescale -tune stillimage -pix_fmt yuv420p out/video.mp4
-#                   ^ daję safe 0 aby przyjmowało bezwzględne ścieżki

fix-audio.sh

@@ -0,0 +1 @@
+ffmpeg -hwaccel videotoolbox -i out/video.mp4 -i icdw-11-final.wav -c:v copy -c:a aac -b:a 320k -strict experimental video-with-new-audio.mp4

generate-demuxer.ts

@@ -2,9 +2,7 @@ import findLoudness, { SwapPoint } from "./find-loudness";
 
 const graph_density = 8000;
 
-const threshold_at_point = parseInt(
+const threshold_at_point = 2;
-  process.env.demuxer_volume_threshold || "1"
-);
 
 const inertia_s = 0.3;
 const inertia_samples = inertia_s * graph_density;
@@ -16,65 +14,75 @@ const minutes = (units: number) => Math.floor(s(units) / 60);
 const hours = (units: number) => Math.floor(units / graph_density / 60 / 60);
 
 const formatTime = (units: number) =>
-  `${hours(units)}:${minutes(units)}:${Math.floor(s(units) % 60)}`;
+	`${hours(units)}:${minutes(units)}:${Math.floor(s(units) % 60)}`;
 
 type Mode = { left: boolean; right: boolean };
 
 async function run() {
-  const [left_breaks, right_breaks] = await Promise.all([
+	const [left_breaks, right_breaks] = await Promise.all([
-    findLoudness("/tmp/leftraw", threshold_at_point, inertia_samples, "left"),
+		findLoudness(
-    findLoudness("/tmp/rightraw", threshold_at_point, inertia_samples, "right"),
+			"/tmp/leftraw",
-  ]);
+			threshold_at_point,
+			inertia_samples,
+			"left"
+		),
+		findLoudness(
+			"/tmp/rightraw",
+			threshold_at_point,
+			inertia_samples,
+			"right"
+		),
+	]);
 
-  const merged = [...left_breaks, ...right_breaks].sort((a, b) =>
+	const merged = [...left_breaks, ...right_breaks].sort((a, b) =>
-    a.position_start < b.position_start
+		a.position_start < b.position_start
-      ? -1
+			? -1
-      : a.position_start > b.position_start
+			: a.position_start > b.position_start
-      ? 1
+				? 1
-      : 0
+				: 0
-  );
+	);
 
-  // console.log("left breaks:", left_breaks);
+	// console.log("left breaks:", left_breaks);
-  // console.log(`right_breaks`, right_breaks);
+	// console.log(`right_breaks`, right_breaks);
-  // console.log(`merged`, merged);
+	// console.log(`merged`, merged);
 
-  function new_mode(m: Mode, s: SwapPoint): Mode {
+	function new_mode(m: Mode, s: SwapPoint): Mode {
-    return { ...m, [s.label]: s.loud };
+		return { ...m, [s.label]: s.loud };
-  }
+	}
 
-  function mode_to_string(mode: Mode) {
+	function mode_to_string(mode: Mode) {
-    if (mode.left && mode.right) {
+		if (mode.left && mode.right) {
-      return "both";
+			return "both";
-    }
+		}
-    for (const side of ["left", "right"]) {
+		for (const side of ["left", "right"]) {
-      if (mode[side as keyof Mode]) {
+			if (mode[side as keyof Mode]) {
-        return side;
+				return side;
-      }
+			}
-    }
+		}
-    return "none";
+		return "none";
-  }
+	}
 
-  console.log("file", `${process.cwd()}/pics/none.png`);
+	console.log("file", `${process.cwd()}/pics/none.png`);
-  let last_point = 0;
+	let last_point = 0;
-  let mode: Mode = { left: false, right: false };
+	let mode: Mode = { left: false, right: false };
-  let last_file;
+	let last_file;
-  let total = 0;
+	let total = 0;
-  for (let i = 2; i < merged.length; i++) {
+	for (let i = 2; i < merged.length; i++) {
-    const point = merged[i];
+		const point = merged[i];
-    mode = new_mode(mode, point);
+		mode = new_mode(mode, point);
-    const file = `${process.cwd()}/pics/${mode_to_string(mode)}.png`;
+		const file = `${process.cwd()}/pics/${mode_to_string(mode)}.png`;
-    const duration = (point.position_start - last_point) / graph_density;
+		const duration = (point.position_start - last_point) / graph_density;
-    console.log(
+		console.log(
-      "duration",
+			"duration",
-      (point.position_start - last_point) / graph_density
+			(point.position_start - last_point) / graph_density
-    );
+		);
-    console.log("file", file);
+		console.log("file", file);
-    last_point = point.position_start;
+		last_point = point.position_start;
-    last_file = file;
+		last_file = file;
-    total += duration * graph_density;
+		total += duration * graph_density;
-  }
+	}
-  console.log("duration", merged[merged.length - 1].duration / graph_density);
+	console.log("duration", merged[merged.length - 1].duration / graph_density);
-  console.log("file", last_file);
+	console.log("file", last_file);
-  console.error(total, formatTime(total));
+	console.error(total, formatTime(total));
 }
 run();

generate.sh

@@ -6,18 +6,43 @@
 #  W katalogu z tym skryptem musisz mieć katalog "pics", w którym są pliki "left.png", "right.png", "none.png" i "both.png"
 #
 
-export input=~/Downloads/icdw5/icdw5-stereo.wav                  # tutaj dajemy ścieżkę do pliku mp3 z Arkiem w jednym kanale i Kubą w drugim
+framerate=60
-export input_mono=~/Downloads/icdw5/icdw5-stereo.wav             # tutaj dajemy ścieżkę do pliku mp3 z Arkiem w jednym kanale i Kubą w drugim
+timescale=25000
-export intro=~/projects/midline/podcast-visualizer/out/intro.mp4 # glitch
+
-export outro=~/projects/midline/podcast-visualizer/out/intro.mp4 # to samo na końcu, co na początku
+aresample=8000   # to bez zmian
-export final_output=~/Downloads/icdw5-viz.mp4
+
-export framerate=25
+# echo dzielimy mp3 na dwa osobne wav
-export timescale=25000
+# ffmpeg -i $input -map_channel 0.0.0 /tmp/left.wav -map_channel 0.0.1 /tmp/right.wav
-export demuxer_volume_threshold=15 #od 0 do 128
+
+
+ffmpeg -i ./left.wav -ac 1 -filter:a aresample=$aresample -map 0:a -c:a pcm_u8 -f data -  > /tmp/leftraw &
+ffmpeg -i ./right.wav -ac 1 -filter:a aresample=$aresample -map 0:a -c:a pcm_u8 -f data -  > /tmp/rightraw &
+
+# czekamy aż obydwa wątki się zakończą
+wait;
+
+echo "generating the demuxers...";
+
+# generuje ścieżki do złożenia przez ffmpega:
+
+ts-node generate-demuxer.ts > out/demuxer.txt
 
-aresample=8000 # to bez zmian
 mkdir -p out
 
-source ./split.sh
+# używa demuxer.txt żeby skleić końcowe video z dźwiękiem:
-source ./demux.sh
+echo generowanie całości
-source ./merge.sh
+# ffmpeg -y -f concat -safe 0 -i out/demuxer.txt -r $framerate -video_track_timescale $timescale -tune stillimage -fps_mode vfr -pix_fmt yuv420p out/video.mp4
+
+# linux
+# ffmpeg -y -f concat -safe 0 -hwaccel vaapi -hwaccel_output_format vaapi -vaapi_device /dev/dri/renderD128 -i out/demuxer.txt -r $framerate -video_track_timescale $timescale -tune stillimage -pix_fmt yuv420p out/video.mp4
+
+# macos mod
+ffmpeg -y -f concat -safe 0 -hwaccel videotoolbox -i out/demuxer.txt -r $framerate -video_track_timescale $timescale -tune stillimage -pix_fmt yuv420p out/video.mp4
+
+
+
+#                   ^ daję safe 0 aby przyjmowało bezwzględne ścieżki
+
+
+# echo łączenie video z dźwiękiem:
+# ffmpeg  -i out/video.mp4 -i $input -ac 1 -video_track_timescale $timescale -tune stillimage out/video-and-audio.mp4

merge.sh

@@ -1,44 +0,0 @@
-processed_intro_path=$PWD/out/intro-processed.mp4
-processed_outro_path=$PWD/out/outro-processed.mp4
-
-echo łączenie video z dźwiękiem:
-ffmpeg -y -vaapi_device /dev/dri/renderD128 \
-	-i out/video.mp4 -i $input_mono \
-	-video_track_timescale $timescale \
-	-tune stillimage \
-	-c:a aac \
-	-ac 1 \
-	-strict experimental \
-	-vf 'format=nv12,hwupload' \
-	-c:v h264_vaapi \
-	out/video-and-audio.mp4
-
-echo reencoding intro to enable fast concat...
-
-function reencode_for_demux_compatibility() {
-	ffmpeg -y -i "$1" \
-		-pix_fmt yuv420p \
-		-r $framerate \
-		-video_track_timescale $timescale \
-		-ac 1 \
-		-b:a 320k \
-		"$2"
-
-}
-
-reencode_for_demux_compatibility "$intro" "$processed_intro_path"
-
-if [ "$intro" = "$outro" ]; then
-	processed_outro_path="$processed_intro_path"
-else
-	reencode_for_demux_compatibility "$outro" "$processed_outro_path"
-fi
-
-echo "" >out/demuxer-branding.txt
-echo "file $processed_intro_path" >>out/demuxer-branding.txt
-echo "file video-and-audio.mp4" >>out/demuxer-branding.txt
-echo "file $processed_outro_path" >>out/demuxer-branding.txt
-
-cat out/demuxer-branding.txt
-
-ffmpeg -y -f concat -safe 0 -i out/demuxer-branding.txt -c:v copy -b:a 320k "$final_output"

split.sh

@@ -1,9 +0,0 @@
-echo dzielimy mp3 na dwa osobne wav
-ffmpeg -y -i $input -map_channel 0.0.0 /tmp/left.wav -map_channel 0.0.1 /tmp/right.wav
-
-echo na dwóch wątkach generujemy surowe pliki
-ffmpeg -y -i /tmp/left.wav -ac 1 -filter:a aresample=$aresample -map 0:a -c:a pcm_u8 -f data - >/tmp/leftraw &
-ffmpeg -y -i /tmp/right.wav -ac 1 -filter:a aresample=$aresample -map 0:a -c:a pcm_u8 -f data - >/tmp/rightraw &
-
-# czekamy aż obydwa wątki się zakończą
-wait