Tune slideshow timing and title overlays

.gitignore

@@ -12,9 +12,10 @@ out/
 *.avi
 *.webm
 
-# Local input media
+# Local input media and bundled local tools
-input/
+input*/
 media/
+ffmpeg/
 
 # OS/editor
 .DS_Store

AGENTS.md

@@ -0,0 +1,10 @@
+# Project context for movmaker
+
+## Notes
+- This project will be a simple Python + ffmpeg movie maker.
+- Desired workflow: user dumps all files into one flat input directory and runs a script; no manual YAML and no required subdirectories.
+- Script should auto-detect images, videos, audio files, and an optional simple text data file for titles/dates/places. Default/simple data file content applies generally to the whole movie; optional per-file entries may be supported but must not be required.
+
+## Guidelines
+- Do not store sensitive secrets, passwords, API keys, or private credentials here.
+- Never push changes to git unless explicitly asked.

README.md

@@ -85,16 +85,18 @@ IMG_002.jpg: Louvre
 Basic:
 
 ```bash
-python3 movmaker.py input --output movie.mp4
+python3 movmaker.py input
 ```
 
 With options:
 
 ```bash
 python3 movmaker.py input \
-  --output out/movie.mp4 \
+  --out-dir out \
-  --image-duration 3 \
+  --image-duration 6 \
-  --fade 1 \
+  --fade 3 \
+  --audio-fade 10 \
+  --first-transition-at 12 \
   --resolution 1920x1080 \
   --fps 30
 ```
@@ -103,12 +105,17 @@ python3 movmaker.py input \
 
 - scans one flat input directory
 - auto-detects media/audio/data files by extension
-- shows each image for 3 seconds by default
+- shows each image fully visible for 6 seconds by default, not counting crossfades
+- keeps the first image longer so the first transition starts at 12 seconds by default
 - includes video clips inline
-- crossfades between media items
+- crossfades between media items over 3 seconds by default
 - loops/trims music to match the video
-- overlays title/date/place at the beginning
+- fades music in at the beginning and out at the end over 10 seconds by default
-- writes an MP4 file
+- overlays the opening title for the first 6 seconds: large bold title, then smaller non-bold date/location on one line
+- adds persistent bottom-left title/date/place and bottom-right `marijo@novosel.dk` stamps
+- writes an MP4 file named like `YYYYMMDD_title.mp4` by default
+- takes `YYYYMMDD` from the first picture/video metadata, falling back to file modification date
+- uses local `ffmpeg/ffmpeg` and `ffmpeg/ffprobe` when present, otherwise system tools
 
 ## Notes
 

movmaker.py

@@ -13,17 +13,26 @@ import argparse
 import json
 import math
 import os
+import re
 import shlex
 import subprocess
 import sys
 import tempfile
 from dataclasses import dataclass
+from datetime import datetime
 from pathlib import Path
 
 IMAGE_EXTS = {".jpg", ".jpeg", ".png", ".webp", ".bmp", ".tif", ".tiff"}
 VIDEO_EXTS = {".mp4", ".mov", ".mkv", ".avi", ".webm", ".m4v"}
 AUDIO_EXTS = {".mp3", ".wav", ".m4a", ".aac", ".flac", ".ogg"}
 DATA_NAMES = {"data.txt", "info.txt", "title.txt"}
+BASE_DIR = Path(__file__).resolve().parent
+FFMPEG = BASE_DIR / "ffmpeg" / "ffmpeg"
+FFPROBE = BASE_DIR / "ffmpeg" / "ffprobe"
+
+
+def tool_path(local_path: Path, fallback: str) -> str:
+    return str(local_path) if local_path.exists() else fallback
 
 
 @dataclass
@@ -52,7 +61,8 @@ def capture(cmd: list[str]) -> str:
 
 def require_tool(name: str) -> None:
     try:
-        subprocess.run([name, "-version"], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
+        executable = tool_path(FFMPEG, "ffmpeg") if name == "ffmpeg" else tool_path(FFPROBE, "ffprobe") if name == "ffprobe" else name
+        subprocess.run([executable, "-version"], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
     except Exception:
         raise SystemExit(f"Error: required tool not found: {name}")
 
@@ -68,7 +78,7 @@ def parse_resolution(value: str) -> tuple[int, int]:
 def ffprobe_duration(path: Path) -> float:
     try:
         out = capture([
-            "ffprobe", "-v", "error",
+            tool_path(FFPROBE, "ffprobe"), "-v", "error",
             "-show_entries", "format=duration",
             "-of", "json",
             str(path),
@@ -79,6 +89,136 @@ def ffprobe_duration(path: Path) -> float:
         return 3.0
 
 
+def ffprobe_creation_date(path: Path) -> datetime | None:
+    try:
+        out = capture([
+            tool_path(FFPROBE, "ffprobe"), "-v", "error",
+            "-show_entries", "format_tags=creation_time:stream_tags=creation_time",
+            "-of", "json",
+            str(path),
+        ])
+        data = json.loads(out)
+        candidates: list[str] = []
+        if isinstance(data.get("format"), dict):
+            candidates.append(data["format"].get("tags", {}).get("creation_time", ""))
+        for stream in data.get("streams", []):
+            candidates.append(stream.get("tags", {}).get("creation_time", ""))
+        for value in candidates:
+            if not value:
+                continue
+            try:
+                return datetime.fromisoformat(value.replace("Z", "+00:00"))
+            except ValueError:
+                pass
+    except Exception:
+        pass
+    return None
+
+
+def exif_image_date(path: Path) -> datetime | None:
+    """Read common JPEG/TIFF EXIF date tags using only the stdlib."""
+    data = path.read_bytes()[:1024 * 1024]
+    if data.startswith(b"\xff\xd8"):
+        pos = 2
+        tiff: bytes | None = None
+        while pos + 4 <= len(data):
+            if data[pos] != 0xFF:
+                break
+            marker = data[pos + 1]
+            pos += 2
+            if marker in {0xD8, 0xD9}:
+                continue
+            if pos + 2 > len(data):
+                break
+            size = int.from_bytes(data[pos:pos + 2], "big")
+            segment = data[pos + 2:pos + size]
+            pos += size
+            if marker == 0xE1 and segment.startswith(b"Exif\x00\x00"):
+                tiff = segment[6:]
+                break
+        if tiff is None:
+            return None
+    else:
+        tiff = data
+
+    if tiff[:2] == b"II":
+        endian = "little"
+    elif tiff[:2] == b"MM":
+        endian = "big"
+    else:
+        return None
+
+    def u16(off: int) -> int:
+        return int.from_bytes(tiff[off:off + 2], endian)
+
+    def u32(off: int) -> int:
+        return int.from_bytes(tiff[off:off + 4], endian)
+
+    def read_ascii(offset: int, count: int) -> str:
+        return tiff[offset:offset + count].split(b"\x00", 1)[0].decode("ascii", "ignore").strip()
+
+    def parse_ifd(ifd_off: int) -> tuple[str | None, int | None]:
+        if ifd_off <= 0 or ifd_off + 2 > len(tiff):
+            return None, None
+        entries = u16(ifd_off)
+        found_date = None
+        exif_ifd = None
+        for i in range(entries):
+            off = ifd_off + 2 + i * 12
+            if off + 12 > len(tiff):
+                break
+            tag = u16(off)
+            typ = u16(off + 2)
+            count = u32(off + 4)
+            value = u32(off + 8)
+            if tag in {0x0132, 0x9003, 0x9004} and typ == 2:
+                found_date = read_ascii(value, count) if count > 4 else tiff[off + 8:off + 8 + count].split(b"\x00", 1)[0].decode("ascii", "ignore")
+            elif tag == 0x8769:
+                exif_ifd = value
+        return found_date, exif_ifd
+
+    try:
+        if u16(2) != 42:
+            return None
+        date_text, exif_ifd = parse_ifd(u32(4))
+        if exif_ifd:
+            exif_date, _ = parse_ifd(exif_ifd)
+            date_text = exif_date or date_text
+        if date_text:
+            return datetime.strptime(date_text[:19], "%Y:%m:%d %H:%M:%S")
+    except Exception:
+        return None
+    return None
+
+
+def media_date(path: Path) -> datetime:
+    ext = path.suffix.lower()
+    dt = None
+    if ext in IMAGE_EXTS:
+        try:
+            dt = exif_image_date(path)
+        except Exception:
+            dt = None
+    if dt is None:
+        dt = ffprobe_creation_date(path)
+    if dt is None:
+        dt = datetime.fromtimestamp(path.stat().st_mtime)
+    return dt
+
+
+def slugify(value: str) -> str:
+    value = value.strip().lower()
+    value = re.sub(r"[^a-z0-9]+", "_", value)
+    value = value.strip("_")
+    return value or "movie"
+
+
+def default_output_path(first_media: Path, data: MovieData, out_dir: Path) -> Path:
+    date = media_date(first_media).strftime("%Y%m%d")
+    title = slugify(data.title or first_media.parent.name or "movie")
+    return out_dir / f"{date}_{title}.mp4"
+
+
 def natural_key(path: Path) -> list[object]:
     # Simple natural-ish sort: IMG_2 before IMG_10.
     import re
@@ -167,7 +307,7 @@ def normalize_clip(item: MediaItem, out: Path, width: int, height: int, fps: int
 
     if item.kind == "image":
         cmd = [
-            "ffmpeg", "-y",
+            tool_path(FFMPEG, "ffmpeg"), "-y",
             "-loop", "1", "-t", f"{item.duration:.3f}", "-i", str(item.path),
             "-an", "-vf", vf,
             "-c:v", "libx264", "-preset", "veryfast", "-crf", "20",
@@ -175,7 +315,7 @@ def normalize_clip(item: MediaItem, out: Path, width: int, height: int, fps: int
         ]
     else:
         cmd = [
-            "ffmpeg", "-y",
+            tool_path(FFMPEG, "ffmpeg"), "-y",
             "-i", str(item.path),
             "-an", "-vf", vf,
             "-c:v", "libx264", "-preset", "veryfast", "-crf", "20",
@@ -195,7 +335,7 @@ def concat_or_mix_audio(audio_files: list[Path], out: Path) -> Path | None:
         return str(p.resolve()).replace("'", "'\\''")
     list_file.write_text("".join(f"file '{esc(p)}'\n" for p in audio_files), encoding="utf-8")
     run([
-        "ffmpeg", "-y", "-f", "concat", "-safe", "0", "-i", str(list_file),
+        tool_path(FFMPEG, "ffmpeg"), "-y", "-f", "concat", "-safe", "0", "-i", str(list_file),
         "-vn", "-c:a", "aac", "-b:a", "192k", str(out),
     ])
     return out
@@ -208,13 +348,14 @@ def build_video(
     output: Path,
     data: MovieData,
     fade: float,
+    audio_fade: float,
     width: int,
     height: int,
     fps: int,
 ) -> None:
     output.parent.mkdir(parents=True, exist_ok=True)
 
-    cmd = ["ffmpeg", "-y"]
+    cmd = [tool_path(FFMPEG, "ffmpeg"), "-y"]
     for clip in clips:
         cmd += ["-i", str(clip)]
     if audio:
@@ -223,38 +364,74 @@ def build_video(
     filters: list[str] = []
     n = len(clips)
 
+    total_duration = durations[0]
     if n == 1:
         current = "0:v"
     else:
         current = "v01"
         offset = max(0.0, durations[0] - fade)
         filters.append(f"[0:v][1:v]xfade=transition=fade:duration={fade:.3f}:offset={offset:.3f}[{current}]")
-        combined_duration = durations[0] + durations[1] - fade
+        total_duration = durations[0] + durations[1] - fade
         for i in range(2, n):
             next_label = f"v{i:02d}"
-            offset = max(0.0, combined_duration - fade)
+            offset = max(0.0, total_duration - fade)
             filters.append(f"[{current}][{i}:v]xfade=transition=fade:duration={fade:.3f}:offset={offset:.3f}[{next_label}]")
             current = next_label
-            combined_duration += durations[i] - fade
+            total_duration += durations[i] - fade
 
     title_lines = [x for x in [data.title, data.date, data.place] if x]
     final_label = "vout"
+    title_label = "titled"
+    bold_font = "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf"
+    normal_font = "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"
     if title_lines:
-        text = drawtext_escape("\\n".join(title_lines))
+        title_font_size = max(44, height // 14)
-        # Draw centered title for first 5 seconds with a translucent box.
+        subtitle_font_size = max(28, height // 24)
+        main_title = data.title or title_lines[0]
+        subtitle = " | ".join(x for x in [data.date, data.place] if x)
+        if subtitle:
+            filters.append(
+                f"[{current}]drawtext=fontfile='{bold_font}':text='{drawtext_escape(main_title)}':x=(w-text_w)/2:y=(h-text_h)/2-{title_font_size // 2}:"
+                f"fontsize={title_font_size}:fontcolor=white:borderw=3:bordercolor=black:"
+                f"enable='between(t,0,6)'[title0]"
+            )
+            filters.append(
+                f"[title0]drawtext=fontfile='{normal_font}':text='{drawtext_escape(subtitle)}':x=(w-text_w)/2:y=(h-text_h)/2+{subtitle_font_size}:"
+                f"fontsize={subtitle_font_size}:fontcolor=white:borderw=3:bordercolor=black:"
+                f"enable='between(t,0,6)'[{title_label}]"
+            )
+        else:
+            filters.append(
+                f"[{current}]drawtext=fontfile='{bold_font}':text='{drawtext_escape(main_title)}':x=(w-text_w)/2:y=(h-text_h)/2:"
+                f"fontsize={title_font_size}:fontcolor=white:borderw=3:bordercolor=black:"
+                f"enable='between(t,0,6)'[{title_label}]"
+            )
+    else:
+        filters.append(f"[{current}]copy[{title_label}]")
+
+    stamp_size = max(14, height // 60)
+    stamp_lines = [x for x in [data.title, data.date, data.place] if x]
+    stamp_text = drawtext_escape(" | ".join(stamp_lines))
+    stamp_left_label = "stamp_left"
+    if stamp_text:
         filters.append(
-            f"[{current}]drawtext=text='{text}':x=(w-text_w)/2:y=(h-text_h)/2:"
+            f"[{title_label}]drawtext=text='{stamp_text}':x=24:y=h-text_h-18:"
-            f"fontsize={max(32, height // 18)}:fontcolor=white:box=1:boxcolor=black@0.45:boxborderw=24:"
+            f"fontsize={stamp_size}:fontcolor=white@0.75:borderw=2:bordercolor=black@0.75[{stamp_left_label}]"
-            f"enable='between(t,0,5)'[{final_label}]"
         )
     else:
-        filters.append(f"[{current}]copy[{final_label}]")
+        filters.append(f"[{title_label}]copy[{stamp_left_label}]")
+    filters.append(
+        f"[{stamp_left_label}]drawtext=text='marijo@novosel.dk':x=w-text_w-24:y=h-text_h-18:"
+        f"fontsize={stamp_size}:fontcolor=white@0.75:borderw=2:bordercolor=black@0.75[{final_label}]"
+    )
 
     cmd += ["-filter_complex", ";".join(filters), "-map", f"[{final_label}]"]
 
     if audio:
         audio_index = len(clips)
-        cmd += ["-map", f"{audio_index}:a:0", "-c:a", "aac", "-b:a", "192k", "-shortest"]
+        audio_fade_out_start = max(0.0, total_duration - audio_fade)
+        audio_filter = f"afade=t=in:st=0:d={audio_fade:.3f},afade=t=out:st={audio_fade_out_start:.3f}:d={audio_fade:.3f}"
+        cmd += ["-map", f"{audio_index}:a:0", "-af", audio_filter, "-c:a", "aac", "-b:a", "192k", "-shortest"]
     else:
         cmd += ["-an"]
 
@@ -265,9 +442,12 @@ def build_video(
 def main(argv: list[str] | None = None) -> int:
     parser = argparse.ArgumentParser(description="Make a video from a flat folder of images/videos/audio.")
     parser.add_argument("input", type=Path, help="flat input directory")
-    parser.add_argument("--output", "-o", type=Path, default=Path("movie.mp4"), help="output mp4 path")
+    parser.add_argument("--output", "-o", type=Path, help="output mp4 path; defaults to out/YYYYMMDD_title.mp4")
-    parser.add_argument("--image-duration", type=float, default=3.0, help="seconds each image is visible")
+    parser.add_argument("--out-dir", type=Path, default=Path("out"), help="directory for auto-named output videos")
-    parser.add_argument("--fade", type=float, default=1.0, help="crossfade duration in seconds")
+    parser.add_argument("--image-duration", type=float, default=6.0, help="seconds each image is visible")
+    parser.add_argument("--fade", type=float, default=3.0, help="crossfade duration in seconds")
+    parser.add_argument("--audio-fade", type=float, default=10.0, help="music fade-in/fade-out duration in seconds")
+    parser.add_argument("--first-transition-at", type=float, default=12.0, help="seconds before first transition starts")
     parser.add_argument("--resolution", type=parse_resolution, default=(1920, 1080), help="output resolution, e.g. 1920x1080")
     parser.add_argument("--fps", type=int, default=30, help="output frames per second")
     args = parser.parse_args(argv)
@@ -278,8 +458,25 @@ def main(argv: list[str] | None = None) -> int:
     width, height = args.resolution
     media, audio_files, data = scan_input(args.input, args.image_duration)
     fade = min(args.fade, max(0.0, args.image_duration - 0.1))
+    audio_fade = max(0.0, args.audio_fade)
+
+    output = args.output or default_output_path(media[0].path, data, args.out_dir)
+
+    # Image duration means fully visible time, not including crossfades.
+    # Middle images need both a fade-in and fade-out added to their clip length.
+    if len(media) > 1:
+        for idx, item in enumerate(media):
+            if item.kind != "image":
+                continue
+            if idx == 0:
+                item.duration = max(item.duration, args.first_transition_at + fade)
+            elif idx == len(media) - 1:
+                item.duration = item.duration + fade
+            else:
+                item.duration = item.duration + (2 * fade)
 
     print(f"Found {len(media)} media file(s), {len(audio_files)} audio file(s).")
+    print(f"Output: {output}")
     if data.title or data.date or data.place:
         print("Title data:", " | ".join(x for x in [data.title, data.date, data.place] if x))
 
@@ -296,9 +493,9 @@ def main(argv: list[str] | None = None) -> int:
             durations.append(ffprobe_duration(out))
 
         audio = concat_or_mix_audio(audio_files, tmp / "music.m4a")
-        build_video(clips, durations, audio, args.output, data, fade, width, height, args.fps)
+        build_video(clips, durations, audio, output, data, fade, audio_fade, width, height, args.fps)
 
-    print(f"Done: {args.output}")
+    print(f"Done: {output}")
     return 0