{"summary":"Des effets ASCII et tramage pour les images et vidéos d’une page de connexion, d’un site ou d’une carte, avec une intensité réglable à tout moment.","notes":"<p>L’ASCII et le tramage redessinent une image en caractères ou en texture de pixels selon sa luminosité : ils gardent la structure des clairs et des sombres et suppriment les détails de la photo. Tous les paramètres sont réglables, si bien qu’une même image peut servir de visuel principal ou passer en arrière-plan comme simple texture. Voici les cinq points essentiels :</p>\n        <ol>\n        <li><strong>ASCII</strong> : pour un aspect terminal ou code, par exemple la page de connexion d’un outil pour développeurs. L’image est découpée en grille et chaque case est remplacée par un caractère selon sa luminosité, de plus en plus dense de « . » à « @ » ; la taille correspond à la taille des caractères, l’espacement à la densité de la grille et l’intensité d’échantillonnage au contraste.</li>\n        <li><strong>Tramage</strong> : pour un aspect de trame d’impression ou de grain de pixels. L’image est découpée en blocs de pixels, une matrice de Bayer 4×4 sert de seuil pour réduire chaque couleur à quelques niveaux, qui se fondent de loin en tons intermédiaires ; la taille correspond aux blocs de pixels, l’espacement à l’échelle de la texture, et plus l’intensité d’échantillonnage est élevée, moins il y a de niveaux de couleur et plus le grain est marqué.</li>\n        <li><strong>Masque</strong> : vaut pour les deux effets. Un dégradé vertical posé sur l’image la fait reculer, et l’effet ressemble alors à une texture plutôt qu’à du bruit ; si l’effet gêne la lecture du texte placé au-dessus, augmentez d’abord le masque.</li>\n        <li><strong>Tons échantillonnés</strong> : pour l’ASCII seulement ; détermine si les caractères dessinent les zones sombres ou les zones claires, selon la luminosité du sujet. Pour une ville la nuit, le sujet est la lumière : on échantillonne les tons clairs ; pour la photo d’un produit sombre sur fond blanc, on échantillonne les ombres.</li>\n        <li><strong>Couleur</strong> : on peut garder les couleurs d’origine ou choisir l’une des cinq palettes ; la luminosité de l’image est d’abord projetée sur cette palette, puis l’effet est généré, ce qui garde un style cohérent d’une image à l’autre.</li>\n        </ol>\n        <h2>Paramètres</h2>\n        <dl class=\"params\">\n        <dt>ASCII</dt><dd>Masque 90 ; tons échantillonnés Ombres ; intensité d’échantillonnage 100 ; taille 37 ; espacement 0</dd>\n        <dt>Tramage</dt><dd>Masque 88 ; intensité d’échantillonnage 100 ; taille 15 ; espacement 0</dd>\n        </dl>\n        <p>Le prompt associé demande d’abord quelle image ou quelle vidéo traiter, puis génère l’effet et ajoute à côté un panneau de paramètres réglables en temps réel.</p>","prompt":"Add an ASCII / ordered-dither effect to one specific image or video in\nthis project, with a live control panel next to it. Use the asset and the\nplace I name (an <img>, a <video>, or a file path or URL, and the page or\nsection it sits in); ask me if I have not named them. Apply the effect to\nthat asset only, in place: keep its size, position, rounded corners and\nanything layered on top of it. If the project already has an image or\ncanvas effect pipeline, or slider, segmented-control and panel components,\nbuild on them; otherwise create a reusable EffectLayer and EffectPanel.\n\nSettings, one object shared by the layer and the panel:\n{ filter: 'ascii' | 'dither', mask, palette, inkTone, strength, size,\n  spacing }\nKeep separate values per filter, so switching filters brings back that\nfilter's own settings.\n\nWhat each control does in each filter, its range, and the recommended\nvalue (ASCII / Dither):\n- Mask effect, 0-100 (90 / 88): a vertical veil over the picture so the\n  effect reads as texture rather than noise. Same in both filters.\n- Color, Original or one of five palettes (Original): gradient-maps the\n  asset's luminance into one color family before filtering.\n- Sample tones, Shadows | Highlights (Shadows; ASCII only, hidden for\n  Dither): which end of the tonal range becomes dense characters.\n- Sampling strength, 0-100 (100 / 100). ASCII: contrast of the tone curve\n  and how bright and opaque the characters are. Dither: how few levels\n  each color channel keeps (8 at 0, 2 at 100) and how fully the quantized\n  color replaces the original.\n- Size, 0-100 (37 / 15). ASCII: font size, 7-24px. Dither: pixel block,\n  2-9px.\n- Spacing, 0-100 (0 / 0). ASCII: grid step, 8-24px across and 9-26px\n  down, never tighter than the glyphs. Dither: scale of the Bayer\n  pattern, 1-4 blocks per matrix cell.\n\nHelpers used throughout:\n- range(v, a, b) = a + (b - a) * v / 100\n- luminance L = (0.299 R + 0.587 G + 0.114 B) / 255\n- s = strength / 100\n- \"The box\" is the asset's rendered size on the page.\n\nSampling:\n- Cover-fit the asset into the box and draw it into a small sampling\n  canvas (about 480px on the long side), then read it with getImageData.\n  A sample is the pixel under a point in the box, composited over white.\n- With a palette, recolor that canvas first: map each pixel's L through a\n  256-entry lookup built from the palette's four stops.\n- Per image or video frame, take about 12,000 evenly spaced samples:\n  low = 8th percentile of L, high = 92nd percentile.\n- Shadows means dark ink, Highlights light ink.\n\nASCII:\n- Characters, from empty to full: \"  .,:;-=+xX08S#&@\" (starts with two\n  spaces).\n- Font: round(range(size, 7, 24))px monospace (22 instead of 24 when the\n  box is under 720px wide), centered in each cell.\n- Grid: x step = max(range(spacing, 8, 24), widest glyph * 1.08); line\n  height = max(range(spacing, 9, 26), height of \"@\" * 1.12, font * 1.04).\n  The first cell sits half a step in from the edge.\n- Tone of a cell: raw = dark ink ? 1 - L : L; normalized = dark ink ?\n  (high - L) / (high - low) : (L - low) / (high - low), clamped 0-1 (use\n  raw alone when high - low < 0.025); mix = 0.72 + 0.14 s;\n  t = raw (1 - mix) + normalized mix. Then a tone curve: black = 0.22 s,\n  white = 1 - 0.08 s, lev = clamp((t - black) / (white - black), 0, 1),\n  tone = lev (1 - s) + smoothstep(lev) s.\n- Skip cells with tone <= 0.02. Character = chars[round(tone * 16)]; skip\n  spaces.\n- Ink: the sampled color lifted in HSV: value += (1 - value) *\n  (0.24 + 0.18 s + 0.24 tone); saturation *= 1 + 0.16 + 0.2 s (clamped).\n  Alpha = clamp((0.12 + 0.7 tone) (0.72 + 0.28 s), 0.06, 0.94).\n\nDither (ordered, Bayer 4x4):\n- pixel size = max(2, round(range(size, 2, 9))) (7 instead of 9 under\n  720px); pattern scale = max(1, round(range(spacing, 1, 4))); levels =\n  max(2, round(range(strength, 8, 2))) per channel; blend =\n  range(strength, 0.32, 1).\n- Matrix [[0,8,2,10],[12,4,14,6],[3,11,1,9],[15,7,13,5]]; threshold =\n  (m + 0.5) / 16, indexed by floor(x / pattern scale), floor(y / pattern\n  scale).\n- For each block sample its center, push saturation away from luma by\n  1 + 0.06 + 0.18 s, then per channel: scaled = c / 255 * (levels - 1);\n  q = floor(scaled) + (fraction > threshold ? 1 : 0);\n  out = c + (q / (levels - 1) * 255 - c) * blend.\n- Write the blocks into a canvas of box / pixel size and draw it scaled up\n  with image smoothing off.\n\nMask and layering:\n- The mask is a vertical gradient with stops at the top, 42% and the\n  bottom. Base opacities: on a light page white 54/20/38%, on a dark page\n  black 44/20/50%. For a mask value v: stop = v <= 50 ? base *\n  (v / 50)^1.08 : base + (100 - base) * ((v - 50) / 50)^1.08.\n- ASCII: draw the (recolored) picture, the mask, then the characters.\n  Dither: draw the dithered picture, then the mask on top.\n- Changing filter or palette cross-fades from the last frame over 150ms.\n\nPalettes (four stops at 0, 0.4, 0.74, 1):\n- meadow: #2B361F, #677D45, #AEBA78, #E5E1BD\n- blue: #192A4D, #326AC5, #A2C2F1, #EDF3FC\n- amber: #3C200F, #A74C15, #EF9A46, #F7E7CA\n- lavender: #292144, #6552AB, #B3A5E4, #F0ECFB\n- mono: #151515, #575757, #A6A6A6, #ECECEC\n\nRendering:\n- Draw into a canvas laid over the asset with the same size and corner\n  radius, at devicePixelRatio capped at 1.6x and about 2048 x 2048.\n- Image: redraw when a setting changes (debounced 72ms).\n- Video: redraw on every new frame (requestVideoFrameCallback, falling\n  back to requestAnimationFrame) while it plays and is on screen; keep\n  the video element underneath as the source.\n\nControl panel:\n- Place it next to the asset, below it when there is room, otherwise\n  beside it, never covering it. About 420px wide, centered on the asset.\n- Rows, in order: Mask effect, Filter (two tabs: ASCII | Dither), Color\n  (round swatches), Sample tones (ASCII only), Sampling strength, Size,\n  Spacing (shown with %).\n- Each row 46px tall: a 136px label column, then the control. Slider: a\n  22px track with 8px radius and a filled part, a 9x30 rounded-rectangle\n  thumb (3px radius), a number box with the value centered, and a snap\n  mark at 50. Swatches: 22px circles, each filled with its palette's two\n  middle stops at 135deg (Original: a tiny crop of the asset); the\n  selected one gets a ring with a small gap.\n- Every change applies live while dragging. Add a \"Reset to recommended\"\n  action that restores the values above for the current filter.\n- Follow the page's light or dark theme and the project's tokens.","skill":"/skills/kubai-ascii-dither/SKILL.md","skillText":"# ASCII与像素抖色效果\n\n《999 个小技巧，让你 Vibe 的产品更有质感》第 003 期 · @kubai087\n\n在用户的项目里按下面的要求改造。要求以用户的口吻写成：文中的 I、me、my 指用户；写着要先问的地方，先停下来问用户再继续。\n\n## 改造要求\n\nAdd an ASCII / ordered-dither effect to one specific image or video in\nthis project, with a live control panel next to it. Use the asset and the\nplace I name (an <img>, a <video>, or a file path or URL, and the page or\nsection it sits in); ask me if I have not named them. Apply the effect to\nthat asset only, in place: keep its size, position, rounded corners and\nanything layered on top of it. If the project already has an image or\ncanvas effect pipeline, or slider, segmented-control and panel components,\nbuild on them; otherwise create a reusable EffectLayer and EffectPanel.\n\nSettings, one object shared by the layer and the panel:\n{ filter: 'ascii' | 'dither', mask, palette, inkTone, strength, size,\n  spacing }\nKeep separate values per filter, so switching filters brings back that\nfilter's own settings.\n\nWhat each control does in each filter, its range, and the recommended\nvalue (ASCII / Dither):\n- Mask effect, 0-100 (90 / 88): a vertical veil over the picture so the\n  effect reads as texture rather than noise. Same in both filters.\n- Color, Original or one of five palettes (Original): gradient-maps the\n  asset's luminance into one color family before filtering.\n- Sample tones, Shadows | Highlights (Shadows; ASCII only, hidden for\n  Dither): which end of the tonal range becomes dense characters.\n- Sampling strength, 0-100 (100 / 100). ASCII: contrast of the tone curve\n  and how bright and opaque the characters are. Dither: how few levels\n  each color channel keeps (8 at 0, 2 at 100) and how fully the quantized\n  color replaces the original.\n- Size, 0-100 (37 / 15). ASCII: font size, 7-24px. Dither: pixel block,\n  2-9px.\n- Spacing, 0-100 (0 / 0). ASCII: grid step, 8-24px across and 9-26px\n  down, never tighter than the glyphs. Dither: scale of the Bayer\n  pattern, 1-4 blocks per matrix cell.\n\nHelpers used throughout:\n- range(v, a, b) = a + (b - a) * v / 100\n- luminance L = (0.299 R + 0.587 G + 0.114 B) / 255\n- s = strength / 100\n- \"The box\" is the asset's rendered size on the page.\n\nSampling:\n- Cover-fit the asset into the box and draw it into a small sampling\n  canvas (about 480px on the long side), then read it with getImageData.\n  A sample is the pixel under a point in the box, composited over white.\n- With a palette, recolor that canvas first: map each pixel's L through a\n  256-entry lookup built from the palette's four stops.\n- Per image or video frame, take about 12,000 evenly spaced samples:\n  low = 8th percentile of L, high = 92nd percentile.\n- Shadows means dark ink, Highlights light ink.\n\nASCII:\n- Characters, from empty to full: \"  .,:;-=+xX08S#&@\" (starts with two\n  spaces).\n- Font: round(range(size, 7, 24))px monospace (22 instead of 24 when the\n  box is under 720px wide), centered in each cell.\n- Grid: x step = max(range(spacing, 8, 24), widest glyph * 1.08); line\n  height = max(range(spacing, 9, 26), height of \"@\" * 1.12, font * 1.04).\n  The first cell sits half a step in from the edge.\n- Tone of a cell: raw = dark ink ? 1 - L : L; normalized = dark ink ?\n  (high - L) / (high - low) : (L - low) / (high - low), clamped 0-1 (use\n  raw alone when high - low < 0.025); mix = 0.72 + 0.14 s;\n  t = raw (1 - mix) + normalized mix. Then a tone curve: black = 0.22 s,\n  white = 1 - 0.08 s, lev = clamp((t - black) / (white - black), 0, 1),\n  tone = lev (1 - s) + smoothstep(lev) s.\n- Skip cells with tone <= 0.02. Character = chars[round(tone * 16)]; skip\n  spaces.\n- Ink: the sampled color lifted in HSV: value += (1 - value) *\n  (0.24 + 0.18 s + 0.24 tone); saturation *= 1 + 0.16 + 0.2 s (clamped).\n  Alpha = clamp((0.12 + 0.7 tone) (0.72 + 0.28 s), 0.06, 0.94).\n\nDither (ordered, Bayer 4x4):\n- pixel size = max(2, round(range(size, 2, 9))) (7 instead of 9 under\n  720px); pattern scale = max(1, round(range(spacing, 1, 4))); levels =\n  max(2, round(range(strength, 8, 2))) per channel; blend =\n  range(strength, 0.32, 1).\n- Matrix [[0,8,2,10],[12,4,14,6],[3,11,1,9],[15,7,13,5]]; threshold =\n  (m + 0.5) / 16, indexed by floor(x / pattern scale), floor(y / pattern\n  scale).\n- For each block sample its center, push saturation away from luma by\n  1 + 0.06 + 0.18 s, then per channel: scaled = c / 255 * (levels - 1);\n  q = floor(scaled) + (fraction > threshold ? 1 : 0);\n  out = c + (q / (levels - 1) * 255 - c) * blend.\n- Write the blocks into a canvas of box / pixel size and draw it scaled up\n  with image smoothing off.\n\nMask and layering:\n- The mask is a vertical gradient with stops at the top, 42% and the\n  bottom. Base opacities: on a light page white 54/20/38%, on a dark page\n  black 44/20/50%. For a mask value v: stop = v <= 50 ? base *\n  (v / 50)^1.08 : base + (100 - base) * ((v - 50) / 50)^1.08.\n- ASCII: draw the (recolored) picture, the mask, then the characters.\n  Dither: draw the dithered picture, then the mask on top.\n- Changing filter or palette cross-fades from the last frame over 150ms.\n\nPalettes (four stops at 0, 0.4, 0.74, 1):\n- meadow: #2B361F, #677D45, #AEBA78, #E5E1BD\n- blue: #192A4D, #326AC5, #A2C2F1, #EDF3FC\n- amber: #3C200F, #A74C15, #EF9A46, #F7E7CA\n- lavender: #292144, #6552AB, #B3A5E4, #F0ECFB\n- mono: #151515, #575757, #A6A6A6, #ECECEC\n\nRendering:\n- Draw into a canvas laid over the asset with the same size and corner\n  radius, at devicePixelRatio capped at 1.6x and about 2048 x 2048.\n- Image: redraw when a setting changes (debounced 72ms).\n- Video: redraw on every new frame (requestVideoFrameCallback, falling\n  back to requestAnimationFrame) while it plays and is on screen; keep\n  the video element underneath as the source.\n\nControl panel:\n- Place it next to the asset, below it when there is room, otherwise\n  beside it, never covering it. About 420px wide, centered on the asset.\n- Rows, in order: Mask effect, Filter (two tabs: ASCII | Dither), Color\n  (round swatches), Sample tones (ASCII only), Sampling strength, Size,\n  Spacing (shown with %).\n- Each row 46px tall: a 136px label column, then the control. Slider: a\n  22px track with 8px radius and a filled part, a 9x30 rounded-rectangle\n  thumb (3px radius), a number box with the value centered, and a snap\n  mark at 50. Swatches: 22px circles, each filled with its palette's two\n  middle stops at 135deg (Original: a tiny crop of the asset); the\n  selected one gets a ring with a small gap.\n- Every change applies live while dragging. Add a \"Reset to recommended\"\n  action that restores the values above for the current filter.\n- Follow the page's light or dark theme and the project's tokens.\n"}