ArisuSay

Like momoisay, but instead of Momoi saying things, it's Tendou Aris (天童アリス — the dark-haired Blue Archive girl) doing her famous teabag, rendered as monochrome Braille dots in your terminal. Aris is drawn alone — background removed — halo and all.

▶ teabag   ·   q / Esc to quit

Variants

Pick a variant with animate <variant> — by number or name:

Install

cargo build --release          # → target/release/arisusay
# or install to ~/.cargo/bin:
cargo install --path .

Usage

arisusay say "정의 실현!"       # static Aris + speech bubble (prints once)

arisusay animate               # teabag, default speed
arisusay animate x2            # twice as fast        (same as: animate 2)
arisusay animate a             # hands-together bob   (same as: animate 3)
arisusay animate b             # both-hands-up bob    (same as: animate 4)

arisusay freestyle             # random variant each loop
arisusay freestyle -t "ㅋㅋㅋ"  # ...with a speech bubble

Controls: q, Esc, or Ctrl-C to quit. The animation runs in the alternate screen and restores your terminal on exit (even on panic).

Terminal size: the canvas is ~34×22, so a window of ~36×26 is enough (wider when using -t/--text). Too small → it prints a friendly message and exits.

How the frames are made

Frames are pre-rendered monochrome Braille baked into the binary via include_str! — no image files or codecs at runtime, exactly like momoisay. They live in frames/*.txt (one file per variant; individual frames joined by a form-feed).

The source media is not committed (it's the original creator's content, and the build doesn't need it — only frames/*.txt does). To regenerate the frames, first fetch the sources into refs/:

mkdir -p refs
# teabag loop (transparent background) — `teabag` / `x2`
curl -L -o refs/aris2.gif "https://media.tenor.com/oA5ClfmykW8AAAAi/alice-aris.gif"
# green-screen clips from https://www.youtube.com/watch?v=T9F1Wk8DQdg
yt-dlp --download-sections "*00:01:05-00:01:27" -f "bv*[height<=720]/b" \
       -o "refs/yt_src.%(ext)s"  "https://www.youtube.com/watch?v=T9F1Wk8DQdg"
yt-dlp --download-sections "*00:01:14-00:01:30" -f "bv*[height<=720]/b" \
       -o "refs/yt_src3.%(ext)s" "https://www.youtube.com/watch?v=T9F1Wk8DQdg"

• aris2.gif — teabag / x2
• yt_src.webm — a is cut from it (8.5–13 s: the single-Aris hands-together bob, after the clone intro ends)
• yt_src3.mp4 — b is cut from it (5.75–11.3 s: the both-hands-up bob, ending before a second Aris walks in)

Then regenerate:

python3 tools/gen_frames.py          # needs Python, Pillow, numpy, scipy, ffmpeg

Note: yt-dlp's section download seeks by keyframe, so cut windows in tools/gen_frames.py may need a small nudge if you re-download.

tools/gen_frames.py removes each source's background (alpha for the gif, chroma-key for the green screen), builds a union bounding box per source so Aris is never clipped or jittery, then fills a Braille dot wherever the pixel is opaque and dark.

Layout

src/main.rs        # tokio entry, clap dispatch, animation loop
src/cli.rs         # say / animate / freestyle ; Motion (teabag, x2, a, b)
src/frames.rs      # embedded frames + loader (yields every frame, verbatim lines)
src/display.rs     # speech bubble, RAII terminal guard, exit listener, render loop
frames/*.txt       # generated Braille frames (aris, motion_a, motion_b, static)
tools/gen_frames.py  # frame generator (sources go in refs/, not committed)

Differences from momoisay

Same feel, three small fixes: frame lines are loaded verbatim (momoisay slices off each line's last char), the iterator yields every frame (momoisay drops the last one), and a RAII guard restores the terminal even if the program panics.

Credits

Original animations of Aris teabagging by 청세치 / 세치혀. This is a fan-made terminal-art homage; all character/animation rights belong to the original creators and Nexon (Blue Archive). Built as a companion to momoisay.