AgentMarketMCP / SKILL 资产档案馆

目录 / dithered-motif-site

SKILL 未评级 已上架

dithered-motif-site

Use when building a "dithered particle-field" animated hero/background — a still or video subject made of many small discrete elements (sparks, falling ash, a rose window, a flock, crowds, falling particles, repeated architecture) rendered as an animated Bayer-dithered dot field on a canvas, with pointer-reactive warp, used as a full-bleed page background. Covers the whole manual pipeline: (1) writing Higgsfield `nano_banana_pro` image prompts and `seedance_2_0` (image-to-video) prompts whose SUBJECT is dither-compatible — the rule that a solid object (one face, one animal, one smooth mass) turns into an illegible grey blob and only "many small discrete elements" survives dithering, and that video prompts MUST demand a locked-off static camera or the loop becomes unusable; (2) ffmpeg frame extraction (~11fps, crop any painted border before scale, flatten background noise); (3) the canvas dither engine itself — downscale-to-buffer, Rec.709 luma, percentile auto-levels PINNED for the whole sequence (never recomputed per frame, or density visibly pulses), the invert flag (decided by dark-on-light vs light-on-dark SOURCE, independent of page color scheme), 8x8 Bayer ordered-dither quantization, and batched-fillStyle circle/square cell drawing; (4) crossfade-the-wrap looping for non-cyclic motifs (most natural motifs have no true loop point — do not hunt for a seam that doesn't exist) vs scored-seam looping for genuinely cyclic ones; (5) pointer-interaction by displacing sampling coordinates (not drawn output) — directional lean for wind/fall, rotational twist for radial subjects, gentle scatter for swarms, and why radial ripple is almost always wrong; (6) page composition and text legibility over a dot field via gradient scrim, mix-blend-mode:multiply (light grounds only), or a solid backing panel. Triggers on "dithered background", "dither animation", "Bayer dither canvas", "particle dither hero", "dot field animation", "dithered video loop", "ordered dither ordered-di

模型生成摘要(rules/v1 · 2026-11-12 19:41):Use when building a "dithered particle-field" animated hero/background — a still or video subject made of many small dis…
这是模型对公开材料的总结,不是官方声明,请以原链内容为准。

存档时间线

版本存档时间内容哈希内容
v12026-11-12 19:4053d56b5e 可取

版本索引永久保留;内容副本只保留最新 1 份,更早版本仅留索引与哈希(存档时间线的证据链不会因此断裂)。

下载存档内容副本

纠错与举报(发现条目失效、署名有误或涉及侵权?)
提交举报 / 纠错

侵权举报经核验成立后,我们会即时下线该条目并删除已存的内容副本。