What it does
Every frame is rebuilt on a fixed 7×9 pixel character grid. Each cell picks its own glyph from what the picture is doing at that spot:
| In the picture | On the grid |
|---|
| A strong edge | Bold strokes that follow it: / \ ‖ = |
| Fine texture with a direction | Thin hatching along that direction |
| Flat areas | A brightness ramp: . - = √ π 0 ◎ ● ■ |
| The outline of the subject | A thin contour one cell outside it |
| Parts that are still appearing | Flickering digits |
On top of the grid: bloom, a slight colour fringe, film grain, and live readouts (a tracker box, centroid X/Y, the subject's main axis θ, the turn angle ∠).
Features
- Your own files. Drop, paste or open a PNG, JPG, WebP, GIF or a short MP4/WebM. GIFs and videos stay animated.
- Type a word. Any text becomes a glowing object that grows in, and it works in 3D too. Use
| for a line break.
- Live camera. Your webcam as live glyph art. With the soundtrack on, movement plays notes.
- 3D object. Tick 3D object and the picture becomes a rounded relief with a front and a back. Drag to turn it, scroll to zoom, or let it turn by itself.
- Subject lift. On a picture with a real background, 3D finds the subject by itself, turns only the subject and keeps the background flat and still behind it. The area under the subject is painted in, so no double shows when it turns.
- AI depth. One button runs a small Depth Anything V2 model in the browser (27 MB, downloaded once) and gives the 3D object real depth. The picture never leaves the device.
- Depth map. Add a grey depth picture (white is near) and the 3D object gets real depth instead of a rounded guess.
- Nine templates. Iris, Planet, Jellyfish, Helix, Knot, Heart, Crystal, Amphora and Mushroom. All are generated in code, so there are no asset files.
- Grow-in intro. The subject grows from its lowest point upward, with a ruler counting 0.05 to 0.96 beside the tip.
- Background tools. Auto removes a plain background by colour and finds the subject by itself on a busy picture. Or pick the mode: find the subject, remove dark, remove light, remove a colour picked from the picture, keep everything. Cut strength, edge softness, grow or shrink, speck cleanup, cut only from the outside, and a preview of what is kept.
- Editor. Paint over what must stay or go, on top of any cut, with undo. The same editor sets a focus point and places your own marks.
- Focus point. Click the spot that matters: the close-up, the tracker, the pulse and a spotlight aim at it.
- Light and dark. Dark drawings on light paper are noticed and swapped automatically, colours kept; or force it either way.
- Sixteen looks. Signal, Terminal, Matrix, Amber CRT, Blueprint, Halftone, Blocks, Vapor, Glitch, Prism, Thermal, Gold, Kaleido, Newsprint, Marble and Typewriter. One click sets characters, colour and screen effects. Surprise me (or the R key) rolls a random one, and you can save your own.
- Character sets. Mixed symbols, classic ASCII, the full 70-character ramp, halftone dots, shade blocks, lines only, digits, binary, random letters, oldskool
_/\-+=.()<>:, newskool $#Xxo, Braille dots, or your own characters (sorted from light to heavy automatically).
- Colour. From the file, thirteen palettes or your own three colours, plus a backdrop colour, hue shift, rainbow spread and colour drift over time.
- Effects. Glow, scanlines, grain, lens fringe, screen curve, digit rain, mirror and kaleidoscope, ink on paper. Five intro styles and a speed control.
- Lens and light. Exposure, star glints on the brightest spots and a lens halo. On a 3D object the light can stay put, travel around it or sweep side to side, with a shadow depth control.
- Glitch controls. Amount, row shift, row height, speed, colour split, jumping blocks, static, picture roll and character scramble.
- Every overlay is editable. Tracker, readouts, ruler, caption, sparks, pulse, rain and your own marks each have a switch, a shape, a colour and speed and size sliders. Place marks by hand with your own labels when the automatic ones land in the wrong spot.
- Glass interface. A top bar with Open and Save, the stage, and one inspector with a rail of seven tabs: Source, Look, Colour, Motion, Effects, Sound, Save. Every slider shows its value and resets on double-click, and a help line at the bottom explains the control under the pointer.
- Soundtrack. A generative score composed from the art: the dominant colour picks the key, brightness the mode, detail the tempo, the file name the melody. Pads, bass, plucks, bells and optional drums run through a reverb and a ping-pong delay. The intro, ruler ticks, trackers and pulses each trigger a sound. Five moods, endless variations, WAV export.
- Video. MP4 (H.264 + AAC) drawn and encoded frame by frame with WebCodecs, so nothing is dropped on a slow machine: 720 px, 1080 px or full size, 30 or 60 fps, 6 to 57 seconds, with the soundtrack mixed in offline. Also a PNG up to 2048 px wide.
- GIF and share link. A looping GIF up to 800 px and 25 fps whose end dissolves back into its start, so there is no jump; unchanged pixels are skipped to keep the file small. Or copy a link that carries every setting and the template or text.
- Real text out. Copy the frame as plain characters, or save it as .txt, a coloured .html page or an .ans file with ANSI colour codes for a terminal.
- Transparent background. PNG, GIF and WebM video without the black, ready to lay over other footage.
- Private. Files never leave the browser. There is nothing to upload to.
Use it
- Open https://winchxyz.github.io/afterglyph/
- Drop a file on the stage, or pick a template.
- Tune density, glow and colour. Tick 3D object to turn it.
- Save PNG, Save GIF or Save video.
여기까지가 README 앞부분입니다. 나머지는 원문에서 읽을 수 있습니다. README 전문 보기 ↗
