@lincyaw/skills 1.0.0-rc.2 → 1.0.0-rc.21
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +1 -1
- package/package.json +2 -3
- package/skills/index.json +9 -18
- package/skills/presmium-animate/SKILL.md +4 -2
- package/skills/presmium-animate/gallery/algorithms/lecture.json +171 -115
- package/skills/presmium-animate/gallery/anime/scenes/kmeans-seeded.ts +2 -1
- package/skills/presmium-animate/gallery/craft/lecture.json +109 -57
- package/skills/presmium-animate/gallery/data/lecture.json +711 -175
- package/skills/presmium-animate/gallery/emphasis/lecture.json +321 -79
- package/skills/presmium-animate/gallery/films/lecture.json +36 -10
- package/skills/presmium-animate/gallery/lengths/lecture.json +23 -88
- package/skills/presmium-animate/gallery/math-ml/lecture.json +238 -159
- package/skills/presmium-animate/gallery/paths-3d/lecture.json +74 -64
- package/skills/presmium-animate/gallery/product/lecture.json +27 -3
- package/skills/presmium-animate/gallery/story/lecture.json +23 -90
- package/skills/presmium-animate/gallery/structure/lecture.json +104 -65
- package/skills/presmium-animate/gallery/text/lecture.json +115 -63
- package/skills/presmium-animate/gallery/timeline/lecture.json +106 -187
- package/skills/presmium-animate/gallery/timeline/scenes/bubble.ts +1 -1
- package/skills/presmium-animate/gallery/transitions/lecture.json +103 -52
- package/skills/presmium-animate/references/lengths.md +1 -1
- package/skills/presmium-animate/references/libraries.md +12 -6
- package/skills/presmium-animate/references/procedural.md +1 -0
- package/skills/presmium-animate/references/story.md +21 -7
- package/skills/presmium-animate/references/team.md +176 -0
- package/skills/presmium-author/SKILL.md +53 -19
- package/skills/presmium-author/references/assistance.md +3 -1
- package/skills/{presmium-present → presmium-author}/references/classroom.md +2 -2
- package/skills/presmium-author/references/document.md +1 -1
- package/skills/{presmium-present → presmium-author}/references/lecture.md +3 -1
- package/skills/presmium-author/references/publishing.md +161 -13
- package/skills/presmium-author/references/writing.md +170 -7
- package/src/cli.ts +4 -4
- package/src/install.ts +42 -5
- package/src/messages.ts +5 -5
- package/src/resources.ts +5 -4
- package/skills/presmium-present/SKILL.md +0 -35
- package/skills/presmium-present/references/delivery.md +0 -122
|
@@ -494,7 +494,8 @@
|
|
|
494
494
|
"level": 2,
|
|
495
495
|
"children": [
|
|
496
496
|
"carry-film--notes",
|
|
497
|
-
"carry-film--shot"
|
|
497
|
+
"carry-film--shot",
|
|
498
|
+
"carry-film--closing"
|
|
498
499
|
],
|
|
499
500
|
"stepOrder": [
|
|
500
501
|
"carry-film--step-0",
|
|
@@ -525,7 +526,17 @@
|
|
|
525
526
|
"children": [
|
|
526
527
|
{
|
|
527
528
|
"type": "text",
|
|
528
|
-
"text": "
|
|
529
|
+
"text": "转场要有理由:新画面从引出它的那个东西里出来,观众才知道两个画面是什么关系。这份参考收录转场、剪辑与镜头运动,每种效果写明它表达的关系、节拍和参数,最后用一段短片把这些手法串起来。"
|
|
530
|
+
}
|
|
531
|
+
]
|
|
532
|
+
},
|
|
533
|
+
{
|
|
534
|
+
"id": "wipes--intro--group",
|
|
535
|
+
"type": "paragraph",
|
|
536
|
+
"children": [
|
|
537
|
+
{
|
|
538
|
+
"type": "text",
|
|
539
|
+
"text": "第一组是转场:光圈、划变、色场淡变、方格和翻卡,分别用于进入细节、前进一步、换一段内容、分隔章节和翻到答案。"
|
|
529
540
|
}
|
|
530
541
|
]
|
|
531
542
|
}
|
|
@@ -541,7 +552,7 @@
|
|
|
541
552
|
"children": [
|
|
542
553
|
{
|
|
543
554
|
"type": "text",
|
|
544
|
-
"text": "
|
|
555
|
+
"text": "用于“换一种看法”或“进入细节”:一个圆从画面 A 的某一点打开,露出画面 B。圆心放在引出 B 的东西上(按钮、关键词、图上的点),转场就有了原因。参数:"
|
|
545
556
|
},
|
|
546
557
|
{
|
|
547
558
|
"type": "text",
|
|
@@ -779,7 +790,7 @@
|
|
|
779
790
|
"children": [
|
|
780
791
|
{
|
|
781
792
|
"type": "text",
|
|
782
|
-
"text": "
|
|
793
|
+
"text": "停留时保持轻微运动,不定格"
|
|
783
794
|
}
|
|
784
795
|
]
|
|
785
796
|
}
|
|
@@ -808,7 +819,7 @@
|
|
|
808
819
|
"children": [
|
|
809
820
|
{
|
|
810
821
|
"type": "text",
|
|
811
|
-
"text": "
|
|
822
|
+
"text": "用于步骤之间和“下一个”:画面 B 沿一个方向替换画面 A,有三种方式。"
|
|
812
823
|
},
|
|
813
824
|
{
|
|
814
825
|
"type": "text",
|
|
@@ -821,7 +832,7 @@
|
|
|
821
832
|
},
|
|
822
833
|
{
|
|
823
834
|
"type": "text",
|
|
824
|
-
"text": "
|
|
835
|
+
"text": " 划变:一条边划过,B 露出,A 退后虚化。"
|
|
825
836
|
},
|
|
826
837
|
{
|
|
827
838
|
"type": "text",
|
|
@@ -834,7 +845,7 @@
|
|
|
834
845
|
},
|
|
835
846
|
{
|
|
836
847
|
"type": "text",
|
|
837
|
-
"text": "
|
|
848
|
+
"text": " 滑入:B 带着阴影盖过 A,A 以较小幅度后退变暗。"
|
|
838
849
|
},
|
|
839
850
|
{
|
|
840
851
|
"type": "text",
|
|
@@ -847,7 +858,7 @@
|
|
|
847
858
|
},
|
|
848
859
|
{
|
|
849
860
|
"type": "text",
|
|
850
|
-
"text": "
|
|
861
|
+
"text": " 推移:A、B 在一条带上一起移动一个画面,带速度模糊。"
|
|
851
862
|
},
|
|
852
863
|
{
|
|
853
864
|
"type": "text",
|
|
@@ -873,7 +884,7 @@
|
|
|
873
884
|
},
|
|
874
885
|
{
|
|
875
886
|
"type": "text",
|
|
876
|
-
"text": "
|
|
887
|
+
"text": " 时 B 从右边进来、向左运动,顺着阅读方向,表示前进;为 "
|
|
877
888
|
},
|
|
878
889
|
{
|
|
879
890
|
"type": "text",
|
|
@@ -886,7 +897,7 @@
|
|
|
886
897
|
},
|
|
887
898
|
{
|
|
888
899
|
"type": "text",
|
|
889
|
-
"text": "
|
|
900
|
+
"text": " 时表示回退。参数:"
|
|
890
901
|
},
|
|
891
902
|
{
|
|
892
903
|
"type": "text",
|
|
@@ -1149,7 +1160,7 @@
|
|
|
1149
1160
|
"children": [
|
|
1150
1161
|
{
|
|
1151
1162
|
"type": "text",
|
|
1152
|
-
"text": "A 淡出,画面经过一片色场,B
|
|
1163
|
+
"text": "用于讲次中段落之间:A 淡出,画面经过一片色场,B 从中淡入,是两个不相关画面之间安静的转场。色场可以是页面色("
|
|
1153
1164
|
},
|
|
1154
1165
|
{
|
|
1155
1166
|
"type": "text",
|
|
@@ -1162,7 +1173,7 @@
|
|
|
1162
1173
|
},
|
|
1163
1174
|
{
|
|
1164
1175
|
"type": "text",
|
|
1165
|
-
"text": "
|
|
1176
|
+
"text": ",亮色页面上“淡到白”)、"
|
|
1166
1177
|
},
|
|
1167
1178
|
{
|
|
1168
1179
|
"type": "text",
|
|
@@ -1188,7 +1199,7 @@
|
|
|
1188
1199
|
},
|
|
1189
1200
|
{
|
|
1190
1201
|
"type": "text",
|
|
1191
|
-
"text": "
|
|
1202
|
+
"text": ")。表达“这个变成那个”时改用匹配剪辑。参数:"
|
|
1192
1203
|
},
|
|
1193
1204
|
{
|
|
1194
1205
|
"type": "text",
|
|
@@ -1429,7 +1440,7 @@
|
|
|
1429
1440
|
"children": [
|
|
1430
1441
|
{
|
|
1431
1442
|
"type": "text",
|
|
1432
|
-
"text": "16 × 9 的方格按波浪铺满画面 A,再按相反的波浪退去,露出画面 B
|
|
1443
|
+
"text": "用于章节分隔和重新开始:16 × 9 的方格按波浪铺满画面 A,再按相反的波浪退去,露出画面 B。波从有意义的地方出发(中心、阅读顺序的起点),也可以按带种子的随机顺序。两个相关状态之间更适合用移动。参数:"
|
|
1433
1444
|
},
|
|
1434
1445
|
{
|
|
1435
1446
|
"type": "text",
|
|
@@ -1693,7 +1704,7 @@
|
|
|
1693
1704
|
"children": [
|
|
1694
1705
|
{
|
|
1695
1706
|
"type": "text",
|
|
1696
|
-
"text": "
|
|
1707
|
+
"text": "用于问题和答案、术语和定义、论断和证据:一张卡片翻到背面,正面是画面 A,背面是画面 B,像抽认卡从问题翻到答案。两面属于同一个物体,所以观众把 B 读成 A 的另一面。参数:"
|
|
1697
1708
|
},
|
|
1698
1709
|
{
|
|
1699
1710
|
"type": "text",
|
|
@@ -1856,7 +1867,7 @@
|
|
|
1856
1867
|
"children": [
|
|
1857
1868
|
{
|
|
1858
1869
|
"type": "text",
|
|
1859
|
-
"text": "以 power2.inOut
|
|
1870
|
+
"text": "以 power2.inOut 用 0.6 s 翻到 183°;半圈时抬起到 106%,影子随高度变大变淡"
|
|
1860
1871
|
}
|
|
1861
1872
|
]
|
|
1862
1873
|
},
|
|
@@ -1976,7 +1987,7 @@
|
|
|
1976
1987
|
"children": [
|
|
1977
1988
|
{
|
|
1978
1989
|
"type": "text",
|
|
1979
|
-
"text": "
|
|
1990
|
+
"text": "按顺序讲一个流程的几个阶段时用面板切换。第一张面板从右侧滑入:前向传播得到预测值。"
|
|
1980
1991
|
}
|
|
1981
1992
|
]
|
|
1982
1993
|
}
|
|
@@ -2033,7 +2044,7 @@
|
|
|
2033
2044
|
"children": [
|
|
2034
2045
|
{
|
|
2035
2046
|
"type": "text",
|
|
2036
|
-
"text": "
|
|
2047
|
+
"text": "表现同一组数据的前后变化时用擦除转场。先给出排序前的数组。"
|
|
2037
2048
|
}
|
|
2038
2049
|
]
|
|
2039
2050
|
}
|
|
@@ -2049,7 +2060,7 @@
|
|
|
2049
2060
|
"children": [
|
|
2050
2061
|
{
|
|
2051
2062
|
"type": "text",
|
|
2052
|
-
"text": "
|
|
2063
|
+
"text": "一条亮线从左往右扫过,扫过的部分换成排序后的样子。"
|
|
2053
2064
|
}
|
|
2054
2065
|
]
|
|
2055
2066
|
}
|
|
@@ -2074,7 +2085,7 @@
|
|
|
2074
2085
|
"children": [
|
|
2075
2086
|
{
|
|
2076
2087
|
"type": "text",
|
|
2077
|
-
"text": "
|
|
2088
|
+
"text": "比较处理前后的同一张图时用擦除对比。分割线从左边扫到中间:左边是去噪后的图,右边是原来带噪声的图。"
|
|
2078
2089
|
}
|
|
2079
2090
|
]
|
|
2080
2091
|
}
|
|
@@ -2090,7 +2101,7 @@
|
|
|
2090
2101
|
"children": [
|
|
2091
2102
|
{
|
|
2092
2103
|
"type": "text",
|
|
2093
|
-
"text": "
|
|
2104
|
+
"text": "分割线扫到最右边,整张图都去噪了。"
|
|
2094
2105
|
}
|
|
2095
2106
|
]
|
|
2096
2107
|
}
|
|
@@ -2106,7 +2117,7 @@
|
|
|
2106
2117
|
"children": [
|
|
2107
2118
|
{
|
|
2108
2119
|
"type": "text",
|
|
2109
|
-
"text": "
|
|
2120
|
+
"text": "最后说明文字从中心以圆形展开。"
|
|
2110
2121
|
}
|
|
2111
2122
|
]
|
|
2112
2123
|
}
|
|
@@ -2122,7 +2133,7 @@
|
|
|
2122
2133
|
"children": [
|
|
2123
2134
|
{
|
|
2124
2135
|
"type": "text",
|
|
2125
|
-
"text": "
|
|
2136
|
+
"text": "剪辑在一帧内换画面,靠运动或形状把视线带过切点:甩镜头用于快节奏换场,速度匹配剪辑连接两个不同的物体,形状匹配剪辑表达“这个变成了那个”。"
|
|
2126
2137
|
}
|
|
2127
2138
|
]
|
|
2128
2139
|
}
|
|
@@ -2138,7 +2149,7 @@
|
|
|
2138
2149
|
"children": [
|
|
2139
2150
|
{
|
|
2140
2151
|
"type": "text",
|
|
2141
|
-
"text": "
|
|
2152
|
+
"text": "用于快节奏地换功能、换场,一部短片最多用两三次。镜头从画面 A 快速甩到画面 B:两块画面在一条带上,一次 power3.inOut 移动一个画面宽,接缝两侧的速度天然一致;只沿运动方向模糊,最快的一帧(正中)模糊最大,切换藏在这一帧里。Remotion 版 "
|
|
2142
2153
|
},
|
|
2143
2154
|
{
|
|
2144
2155
|
"type": "text",
|
|
@@ -2151,7 +2162,7 @@
|
|
|
2151
2162
|
},
|
|
2152
2163
|
{
|
|
2153
2164
|
"type": "text",
|
|
2154
|
-
"text": "
|
|
2165
|
+
"text": " 用同一手法,但只接图标、方向固定;这一版接画面参数和方向。参数:"
|
|
2155
2166
|
},
|
|
2156
2167
|
{
|
|
2157
2168
|
"type": "text",
|
|
@@ -2366,7 +2377,7 @@
|
|
|
2366
2377
|
"children": [
|
|
2367
2378
|
{
|
|
2368
2379
|
"type": "text",
|
|
2369
|
-
"text": "
|
|
2380
|
+
"text": "用于两个不同物体之间(输入和输出、这一步和下一步),没有形状可以匹配时。卡片 A 沿一条直线加速飞走,在速度最快的一刻硬切,卡片 B 从同一条线的另一侧以同样的速度进来、减速停下,眼睛读成一个连续的动作。剪辑段的长度就是这一状态的时长。参数:"
|
|
2370
2381
|
},
|
|
2371
2382
|
{
|
|
2372
2383
|
"type": "text",
|
|
@@ -2666,7 +2677,7 @@
|
|
|
2666
2677
|
"children": [
|
|
2667
2678
|
{
|
|
2668
2679
|
"type": "text",
|
|
2669
|
-
"text": "
|
|
2680
|
+
"text": "用于“这个变成了那个”:输入和输出、问题和回答、符号和它代表的东西。场景 A 里的圆加速放大,一帧硬切到场景 B 里一个不同的圆;两个圆的大小和位置完全相同,形状带着视线跨过两个场景。参数:"
|
|
2670
2681
|
},
|
|
2671
2682
|
{
|
|
2672
2683
|
"type": "text",
|
|
@@ -2911,7 +2922,7 @@
|
|
|
2911
2922
|
"children": [
|
|
2912
2923
|
{
|
|
2913
2924
|
"type": "text",
|
|
2914
|
-
"text": "
|
|
2925
|
+
"text": "停留时保持轻微运动"
|
|
2915
2926
|
}
|
|
2916
2927
|
]
|
|
2917
2928
|
}
|
|
@@ -2940,7 +2951,7 @@
|
|
|
2940
2951
|
"children": [
|
|
2941
2952
|
{
|
|
2942
2953
|
"type": "text",
|
|
2943
|
-
"text": "
|
|
2954
|
+
"text": "镜头运动不换画面,只改变观众看的位置和范围:推镜给静止画面分量,拉镜揭示上下文,滑动变焦改变参照系,移焦转移注意力,穿越变焦进入细节,摇镜依次经过各个站点,分屏对比前后两种状态。"
|
|
2944
2955
|
}
|
|
2945
2956
|
]
|
|
2946
2957
|
}
|
|
@@ -2956,7 +2967,7 @@
|
|
|
2956
2967
|
"children": [
|
|
2957
2968
|
{
|
|
2958
2969
|
"type": "text",
|
|
2959
|
-
"text": "
|
|
2970
|
+
"text": "用于标题卡、定义、引文等讲解停留较久的画面:一次缓慢的推镜,让静止的画面在朗读时有分量。推的总距离固定,停留越长推得越慢,文字的移动始终比阅读慢。参数:"
|
|
2960
2971
|
},
|
|
2961
2972
|
{
|
|
2962
2973
|
"type": "text",
|
|
@@ -3093,7 +3104,7 @@
|
|
|
3093
3104
|
"children": [
|
|
3094
3105
|
{
|
|
3095
3106
|
"type": "text",
|
|
3096
|
-
"text": "
|
|
3107
|
+
"text": "停留时保持轻微运动;不设退场,由下一状态或下一节切走"
|
|
3097
3108
|
}
|
|
3098
3109
|
]
|
|
3099
3110
|
}
|
|
@@ -3122,7 +3133,7 @@
|
|
|
3122
3133
|
"children": [
|
|
3123
3134
|
{
|
|
3124
3135
|
"type": "text",
|
|
3125
|
-
"text": "
|
|
3136
|
+
"text": "用于要看到背景才能正确理解的数字:训练成绩旁边的测试成绩、增长率旁边的基数。镜头先只给一个数字,再一次拉远,露出改变它读法的上下文。数字始终是同一个元素,镜头以它为中心缩放。示例数值要标注“示例数据”或写出来源。参数:"
|
|
3126
3137
|
},
|
|
3127
3138
|
{
|
|
3128
3139
|
"type": "text",
|
|
@@ -3370,7 +3381,7 @@
|
|
|
3370
3381
|
"children": [
|
|
3371
3382
|
{
|
|
3372
3383
|
"type": "text",
|
|
3373
|
-
"text": "
|
|
3384
|
+
"text": "停留时保持轻微运动"
|
|
3374
3385
|
}
|
|
3375
3386
|
]
|
|
3376
3387
|
}
|
|
@@ -3399,7 +3410,7 @@
|
|
|
3399
3410
|
"children": [
|
|
3400
3411
|
{
|
|
3401
3412
|
"type": "text",
|
|
3402
|
-
"text": "
|
|
3413
|
+
"text": "用于“东西没变,参照系变了”:同一个词放进更长的上下文、同一个结果对照更大的语料。希区柯克变焦让镜头后退的同时焦距拉长,主体大小不变,背后的世界涌上来("
|
|
3403
3414
|
},
|
|
3404
3415
|
{
|
|
3405
3416
|
"type": "text",
|
|
@@ -3412,7 +3423,7 @@
|
|
|
3412
3423
|
},
|
|
3413
3424
|
{
|
|
3414
3425
|
"type": "text",
|
|
3415
|
-
"text": "
|
|
3426
|
+
"text": " 时远去)。效果来自一个约束:主体大小等于焦距除以距离,所以每一帧的焦距由距离求出(f = d),深度 z 处的层按 d/(d+z) 缩放,全部在 2D 中计算。参数:"
|
|
3416
3427
|
},
|
|
3417
3428
|
{
|
|
3418
3429
|
"type": "text",
|
|
@@ -3640,7 +3651,7 @@
|
|
|
3640
3651
|
"children": [
|
|
3641
3652
|
{
|
|
3642
3653
|
"type": "text",
|
|
3643
|
-
"text": "
|
|
3654
|
+
"text": "用于先看损失曲线再看验证曲线、先看结论再看它的前提:两张卡片在不同深度,焦点从近处的一张移到远处的一张,不切镜头就转移注意力。一个焦平面值 φ 同时驱动每张卡片的模糊、缩放、亮度和强调色边框,四者一起变化,读成同一个镜头。参数:"
|
|
3644
3655
|
},
|
|
3645
3656
|
{
|
|
3646
3657
|
"type": "text",
|
|
@@ -3842,7 +3853,7 @@
|
|
|
3842
3853
|
"children": [
|
|
3843
3854
|
{
|
|
3844
3855
|
"type": "text",
|
|
3845
|
-
"text": "
|
|
3856
|
+
"text": "用于从目录进入某一讲、从系统图进入一个模块:镜头从总览飞进其中一张卡片,直到这张卡片充满画面,成为画面 B。被推近的卡片从一开始就装着缩小的 B,同一个元素长到全尺寸,不需要替换。"
|
|
3846
3857
|
},
|
|
3847
3858
|
{
|
|
3848
3859
|
"type": "text",
|
|
@@ -3855,7 +3866,7 @@
|
|
|
3855
3866
|
},
|
|
3856
3867
|
{
|
|
3857
3868
|
"type": "text",
|
|
3858
|
-
"text": " 倒放,从 B
|
|
3869
|
+
"text": " 倒放,从 B 拉出,显示它在整体中的位置。参数:"
|
|
3859
3870
|
},
|
|
3860
3871
|
{
|
|
3861
3872
|
"type": "text",
|
|
@@ -4083,7 +4094,7 @@
|
|
|
4083
4094
|
"children": [
|
|
4084
4095
|
{
|
|
4085
4096
|
"type": "text",
|
|
4086
|
-
"text": "
|
|
4097
|
+
"text": "用于流水线、步骤序列和功能巡览:镜头沿一排站点横移,每站停一下。每个站点是一个状态,停留时长就是状态时长,可以按讲解调整;站数改变时,同步增减定义里的状态。邻站始终在画面边缘可见,整排读成一个连续的地方。参数:"
|
|
4087
4098
|
},
|
|
4088
4099
|
{
|
|
4089
4100
|
"type": "text",
|
|
@@ -4272,7 +4283,7 @@
|
|
|
4272
4283
|
"children": [
|
|
4273
4284
|
{
|
|
4274
4285
|
"type": "text",
|
|
4275
|
-
"text": "
|
|
4286
|
+
"text": "用于降噪、修复、渲染、重画等同一事物的前后对比:一条硬边把“之后”扫过“之前”,中途停下让两半并排对比,再扫完。默认画面是同一组带噪声的采样,之前是原始折线,之后是滤波后的曲线;给出 "
|
|
4276
4287
|
},
|
|
4277
4288
|
{
|
|
4278
4289
|
"type": "text",
|
|
@@ -4298,7 +4309,7 @@
|
|
|
4298
4309
|
},
|
|
4299
4310
|
{
|
|
4300
4311
|
"type": "text",
|
|
4301
|
-
"text": "
|
|
4312
|
+
"text": "(同样大小)可以对比两张图。参数:"
|
|
4302
4313
|
},
|
|
4303
4314
|
{
|
|
4304
4315
|
"type": "text",
|
|
@@ -4513,7 +4524,7 @@
|
|
|
4513
4524
|
"children": [
|
|
4514
4525
|
{
|
|
4515
4526
|
"type": "text",
|
|
4516
|
-
"text": "
|
|
4527
|
+
"text": "硬边停住 1.5 s"
|
|
4517
4528
|
}
|
|
4518
4529
|
]
|
|
4519
4530
|
},
|
|
@@ -4559,7 +4570,7 @@
|
|
|
4559
4570
|
"children": [
|
|
4560
4571
|
{
|
|
4561
4572
|
"type": "text",
|
|
4562
|
-
"text": "
|
|
4573
|
+
"text": "推近让结果落定"
|
|
4563
4574
|
}
|
|
4564
4575
|
]
|
|
4565
4576
|
}
|
|
@@ -4624,7 +4635,7 @@
|
|
|
4624
4635
|
"children": [
|
|
4625
4636
|
{
|
|
4626
4637
|
"type": "text",
|
|
4627
|
-
"text": "
|
|
4638
|
+
"text": "承接短片长 15 秒:每个画面边界都有一个元素延续到下一个画面,观众因此读成一个连续的故事。"
|
|
4628
4639
|
}
|
|
4629
4640
|
]
|
|
4630
4641
|
}
|
|
@@ -4640,7 +4651,7 @@
|
|
|
4640
4651
|
"children": [
|
|
4641
4652
|
{
|
|
4642
4653
|
"type": "text",
|
|
4643
|
-
"text": "一个点依次变成 Presmium 做出的每一样东西:文稿、步骤、幻灯片、视频的一帧,最后是片名。整片是一个 defineScene
|
|
4654
|
+
"text": "一个点依次变成 Presmium 做出的每一样东西:文稿、步骤、幻灯片、视频的一帧,最后是片名。整片是一个 defineScene 场景,每拍一个状态;每个边界用一种不同的承接动作(morphRect、ribbon、iris、镜头拉远加 hop、seal)。镜头会跟随、甩动、落地时震动,静止时没有漂移;背后是一层远处的点阵。动作来自 gallery/lengths/scenes/kit/fx/moves.ts 与 camera.ts。"
|
|
4644
4655
|
}
|
|
4645
4656
|
]
|
|
4646
4657
|
},
|
|
@@ -4913,6 +4924,22 @@
|
|
|
4913
4924
|
"scene": "carry/one-dot",
|
|
4914
4925
|
"alt": "一个强调色的点落下,展开成文稿的一页;三下连击依次点亮三个步骤;光圈从步骤 3 的圆点打开出幻灯片;镜头拉远,幻灯片是视频的一帧,节点跳上时间轴成为播放头;播放键长成转动的圆盘,片名 Presmium 落下。"
|
|
4915
4926
|
}
|
|
4927
|
+
},
|
|
4928
|
+
"carry-film--closing": {
|
|
4929
|
+
"id": "carry-film--closing",
|
|
4930
|
+
"type": "text",
|
|
4931
|
+
"content": [
|
|
4932
|
+
{
|
|
4933
|
+
"id": "carry-film--closing--p",
|
|
4934
|
+
"type": "paragraph",
|
|
4935
|
+
"children": [
|
|
4936
|
+
{
|
|
4937
|
+
"type": "text",
|
|
4938
|
+
"text": "每个转场、剪辑和镜头运动都回答同一个问题:观众的视线从哪里来、到哪里去。一个元素带着视线跨过边界,两个画面就读成一个连续的动作。"
|
|
4939
|
+
}
|
|
4940
|
+
]
|
|
4941
|
+
}
|
|
4942
|
+
]
|
|
4916
4943
|
}
|
|
4917
4944
|
},
|
|
4918
4945
|
"steps": {
|
|
@@ -5956,6 +5983,10 @@
|
|
|
5956
5983
|
"blockId": "carry-film--shot",
|
|
5957
5984
|
"visible": true,
|
|
5958
5985
|
"state": "dot"
|
|
5986
|
+
},
|
|
5987
|
+
{
|
|
5988
|
+
"blockId": "carry-film--closing",
|
|
5989
|
+
"visible": false
|
|
5959
5990
|
}
|
|
5960
5991
|
],
|
|
5961
5992
|
"narrationIds": []
|
|
@@ -5972,6 +6003,10 @@
|
|
|
5972
6003
|
"blockId": "carry-film--shot",
|
|
5973
6004
|
"visible": true,
|
|
5974
6005
|
"state": "doc"
|
|
6006
|
+
},
|
|
6007
|
+
{
|
|
6008
|
+
"blockId": "carry-film--closing",
|
|
6009
|
+
"visible": false
|
|
5975
6010
|
}
|
|
5976
6011
|
],
|
|
5977
6012
|
"narrationIds": []
|
|
@@ -5988,6 +6023,10 @@
|
|
|
5988
6023
|
"blockId": "carry-film--shot",
|
|
5989
6024
|
"visible": true,
|
|
5990
6025
|
"state": "steps"
|
|
6026
|
+
},
|
|
6027
|
+
{
|
|
6028
|
+
"blockId": "carry-film--closing",
|
|
6029
|
+
"visible": false
|
|
5991
6030
|
}
|
|
5992
6031
|
],
|
|
5993
6032
|
"narrationIds": []
|
|
@@ -6004,6 +6043,10 @@
|
|
|
6004
6043
|
"blockId": "carry-film--shot",
|
|
6005
6044
|
"visible": true,
|
|
6006
6045
|
"state": "slide"
|
|
6046
|
+
},
|
|
6047
|
+
{
|
|
6048
|
+
"blockId": "carry-film--closing",
|
|
6049
|
+
"visible": false
|
|
6007
6050
|
}
|
|
6008
6051
|
],
|
|
6009
6052
|
"narrationIds": []
|
|
@@ -6020,6 +6063,10 @@
|
|
|
6020
6063
|
"blockId": "carry-film--shot",
|
|
6021
6064
|
"visible": true,
|
|
6022
6065
|
"state": "video"
|
|
6066
|
+
},
|
|
6067
|
+
{
|
|
6068
|
+
"blockId": "carry-film--closing",
|
|
6069
|
+
"visible": false
|
|
6023
6070
|
}
|
|
6024
6071
|
],
|
|
6025
6072
|
"narrationIds": []
|
|
@@ -6036,6 +6083,10 @@
|
|
|
6036
6083
|
"blockId": "carry-film--shot",
|
|
6037
6084
|
"visible": true,
|
|
6038
6085
|
"state": "title"
|
|
6086
|
+
},
|
|
6087
|
+
{
|
|
6088
|
+
"blockId": "carry-film--closing",
|
|
6089
|
+
"visible": false
|
|
6039
6090
|
}
|
|
6040
6091
|
],
|
|
6041
6092
|
"narrationIds": []
|
|
@@ -6047,7 +6098,7 @@
|
|
|
6047
6098
|
"content": [
|
|
6048
6099
|
{
|
|
6049
6100
|
"type": "text",
|
|
6050
|
-
"text": "
|
|
6101
|
+
"text": "训练的一轮分三步。第一张面板滑进来:前向传播,得到预测值。"
|
|
6051
6102
|
}
|
|
6052
6103
|
]
|
|
6053
6104
|
},
|
|
@@ -6065,7 +6116,7 @@
|
|
|
6065
6116
|
"content": [
|
|
6066
6117
|
{
|
|
6067
6118
|
"type": "text",
|
|
6068
|
-
"text": "
|
|
6119
|
+
"text": "第三张进来:反向传播,更新参数。一轮训练到这里结束。"
|
|
6069
6120
|
}
|
|
6070
6121
|
]
|
|
6071
6122
|
},
|
|
@@ -6074,7 +6125,7 @@
|
|
|
6074
6125
|
"content": [
|
|
6075
6126
|
{
|
|
6076
6127
|
"type": "text",
|
|
6077
|
-
"text": "
|
|
6128
|
+
"text": "先看排序前的数组,顺序是乱的。"
|
|
6078
6129
|
}
|
|
6079
6130
|
]
|
|
6080
6131
|
},
|
|
@@ -6083,7 +6134,7 @@
|
|
|
6083
6134
|
"content": [
|
|
6084
6135
|
{
|
|
6085
6136
|
"type": "text",
|
|
6086
|
-
"text": "
|
|
6137
|
+
"text": "一条亮线从左往右扫过,扫过的部分换成排序后的样子。"
|
|
6087
6138
|
}
|
|
6088
6139
|
]
|
|
6089
6140
|
},
|
|
@@ -6092,7 +6143,7 @@
|
|
|
6092
6143
|
"content": [
|
|
6093
6144
|
{
|
|
6094
6145
|
"type": "text",
|
|
6095
|
-
"text": "
|
|
6146
|
+
"text": "分割线扫到中间。左边去过噪,右边还带着噪声,两半可以直接比较。"
|
|
6096
6147
|
}
|
|
6097
6148
|
]
|
|
6098
6149
|
},
|
|
@@ -6101,7 +6152,7 @@
|
|
|
6101
6152
|
"content": [
|
|
6102
6153
|
{
|
|
6103
6154
|
"type": "text",
|
|
6104
|
-
"text": "
|
|
6155
|
+
"text": "分割线扫到最右边,整张图都去噪了。"
|
|
6105
6156
|
}
|
|
6106
6157
|
]
|
|
6107
6158
|
},
|
|
@@ -6110,7 +6161,7 @@
|
|
|
6110
6161
|
"content": [
|
|
6111
6162
|
{
|
|
6112
6163
|
"type": "text",
|
|
6113
|
-
"text": "
|
|
6164
|
+
"text": "最后,说明文字从中心以圆形展开。"
|
|
6114
6165
|
}
|
|
6115
6166
|
]
|
|
6116
6167
|
}
|
|
@@ -154,7 +154,7 @@ Not a slower 30-second film. Keep the skeleton (hook, brand, problem, answer, ca
|
|
|
154
154
|
- **Shape.** The loudest moment is the reveal, 30–60% of the way in, never the logo or the last 15%. Before a big hit the music drops 6–12 dB for 0.3–0.5 s and comes back on the hit (in Presmium a short, quieter bed clip in the step before). Many identical hits in lockstep turn into a tone: loosen their timing and vary each one. The ending: the music stops on a bar line, 0.2–0.7 s of near-silence, then one soft note in the track's key 8–20 dB under the peak, ringing 0.5–2.5 s, and no dead air after it; not a stock chord with a shimmer.
|
|
155
155
|
- **Sync per kind.** Viewers notice sound 45 ms early or 125 ms late. A click's or impact's attack lands on the contact frame (10–30 ms early is fine); a whoosh peaks on the move's fastest frame, starting 100–450 ms before; a pop at half the element's travel; a riser ends 30–250 ms before its hit. Most files peak after their start, so measure the offset. Take sync times from the scene's own state durations, never retype them, and place sounds only once the picture is locked: a retimed state moves every sound after it. Measure the export's audio offset once (a sharp effect against its frame) and subtract it. A run of effects (ticks on a list) alternates two takes, gets quieter, closes up as the motion speeds up, and becomes one swoosh once it is too dense to count.
|
|
156
156
|
- **Levels.** Under a voice the bed sits about −18 dB (volume 0.12–0.14) and ducks further while anyone speaks; keep effects out of 1–4 kHz under narration and never put a hit on a key word. An effect is audible when it is 3 dB (clear at 6 dB) over the bed in its own strongest band, not by overall level; nothing in 2–6 kHz more than 6 dB above the music's loudest there (phones turn harsh first). A bed whose energy is mostly under 150 Hz is inaudible on a phone. Deliver −16 to −14 LUFS for social video and a true peak of −1 dBTP measured on the final MP4 (AAC adds up to 1 dB; master at about −1.5): `ffmpeg -i <file> -af ebur128=peak=true -f null -`; `silencedetect=n=-50dB:d=0.3` finds dead air at the start and end and the longest gap.
|
|
157
|
-
- **In Presmium** sound is `production.mediaClips` (see `presmium-
|
|
157
|
+
- **In Presmium** sound is `production.mediaClips` (see `presmium-author`, lecture production): each clip lies within one step with its own volume, so a bed is cut into per-step clips at bar lines, and ducking is the lower volume of the clips under narration.
|
|
158
158
|
|
|
159
159
|
## Build and export in Presmium
|
|
160
160
|
|
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
# Choosing a library and writing the moves
|
|
4
4
|
|
|
5
|
-
Every library here is driven the same way: the scene file builds a timeline (or a drawing function) once, and Presmium pauses it and seeks it to the time it needs. What differs is what each library draws well. The pitfalls below are the ones met while building this skill's gallery (`gallery/<library>/`, see "The gallery" in `SKILL.md`), each checked by seeking the scene forward and back in a browser, in light and dark.
|
|
5
|
+
Every library here is driven the same way: the scene file builds a timeline (or a drawing function) once, and Presmium pauses it and seeks it to the time it needs. What differs is what each library draws well. The pitfalls below are the ones met while building this skill's gallery (`gallery/<library>/`, see "The gallery" in `SKILL.md`) and a 19-page explainer made by several agents, each checked by seeking the scene forward and back in a browser, in light and dark.
|
|
6
6
|
|
|
7
7
|
## What a scene receives
|
|
8
8
|
|
|
@@ -16,22 +16,27 @@ A scene file default-exports one call to a function from `@lincyaw/addon-motion/
|
|
|
16
16
|
| `total` | Their sum, the length of the timeline in milliseconds |
|
|
17
17
|
| `width`, `height` | The layout size of `root` in CSS pixels before ancestor transforms (including slide scaling) when the scene is built |
|
|
18
18
|
| `color(name)` | A design-system color as `rgb()`/`rgba()`, such as `color('accent')` (names in "Rules for every library" below) |
|
|
19
|
+
| `parameters`, `state` | The block's parameters and the current step's declared state, read live. A scene whose timeline covers every step reads neither while it builds ("Build once" below) |
|
|
19
20
|
| the library | `gsap` and `plugins`; `anime`; `motion`; `d3`; `THREE`, `renderer` and `addons`; `React` and `remotion` |
|
|
20
21
|
|
|
21
22
|
Each library counts time in its own unit: GSAP and Motion in seconds (`at(k) / 1000`), Anime.js, D3, Three.js and `defineScene` in milliseconds, Remotion in frames.
|
|
22
23
|
|
|
23
24
|
## Rules for every library
|
|
24
25
|
|
|
25
|
-
- **Colors** come from `color(name)` or `var(--pm-color-<name>)`. Names: `bg`, `surface`, `sunk`, `code`, `ink`, `muted`, `line`, `line-strong`, `accent`, `accent-soft`, `accent-line`, `on-accent`, `ok`, `ok-soft`, `warn`, `warn-soft`, `danger`, `danger-soft`, `syntax-key`, `syntax-string`, `syntax-keyword`, `pointer-fill`, `pointer-stroke`, and for light and shade `sheen` (stays light in both themes) and `shadow` (stays dark). Mix with `color-mix(in oklab, var(--pm-color-accent) 40%, transparent)`.
|
|
26
|
+
- **Colors** come from `color(name)` or `var(--pm-color-<name>)`. Names: `bg`, `surface`, `sunk`, `code`, `ink`, `muted`, `line`, `line-strong`, `accent`, `accent-soft`, `accent-line`, `on-accent`, `ok`, `ok-soft`, `warn`, `warn-soft`, `danger`, `danger-soft`, `syntax-key`, `syntax-string`, `syntax-keyword`, `pointer-fill`, `pointer-stroke`, and for light and shade `sheen` (stays light in both themes) and `shadow` (stays dark). Mix with `color-mix(in oklab, var(--pm-color-accent) 40%, transparent)`. Never type a raw hex or `rgb()` value into a scene or a shared kit: it stays the same in the dark theme and outside the design system. A piece's own colour roles (truth, wrong, the user) are names for design tokens, mapped once in its shared kit. For identities (roles, actors, series, categories) use `category-1` to `category-8`, in order, each with a `-soft` fill: text on a soft fill is `ink` and the category colour draws the edge, and past four categories label them as well; they never mean status, which stays with `ok`, `warn` and `danger`, and `accent` stays the emphasis.
|
|
26
27
|
- **Light and shade by theme.** The scene is built again when the theme changes, so decide at build from `color('bg')`. A white highlight (sheen, hot core, sparkle) reads as light on the dark page but as a gap on a white card: on the light page use the accent mixed with `shadow` instead. A halo in the accent color muddies the light page and reads as light only on the dark one: make it about 0.3 strong on light and 0.5 on dark. To darken a color in both themes, mix it with `shadow`, not `ink` (which is light in dark mode).
|
|
27
28
|
- **Fonts that ship:** IBM Plex Sans 400/500/600, Source Serif 4 400/600, IBM Plex Mono 400/500, Latin only. Heavier weights are synthesized by the browser; there is no variable axis, so a weight tween snaps. Chinese falls back to system fonts whose widths differ by machine: never measure text at build.
|
|
28
|
-
- **SVG ids are page-wide.** A document can show the same scene twice; give every filter, mask, clip path, gradient and
|
|
29
|
+
- **SVG ids are page-wide.** A document can show the same scene twice, and the reading page shows many scenes at once; give every filter, mask, clip path, gradient, pattern and marker an id made when the scene is built (a counter), never a fixed one. Filters and gradients on a zero-height line need `userSpaceOnUse` regions, or nothing renders.
|
|
30
|
+
- **Arrowheads are shapes, not markers.** A `marker-end` arrowhead shows at the end before its line has drawn there, and a marker with a fixed id (`id="arrow"`) is shared by every scene on the reading page, so hiding or unmounting one scene takes the other scenes' arrowheads with it. Draw one arrowhead element per edge, without an id, and show it when its line arrives.
|
|
31
|
+
- **Drawing a stroke with `pathLength="1"`.** A dash array of `1 1` leaves a stub at the end of a path not yet drawn, because the next dash starts exactly there: use `1 2`. A closed shape (a rounded rect) drawn this way keeps a notch at its starting corner: set `stroke-dasharray` back to `none` once it is drawn. `vector-effect: non-scaling-stroke` measures dashes in screen pixels instead of the path length, so the draw breaks into dashes: leave it off strokes that draw. GSAP's DrawSVG writes its own dash array over any in the CSS, so a dashed line drawn with it turns solid.
|
|
29
32
|
- **Nothing else animates what a scene moves:** no CSS transitions or `@keyframes` on those elements. A filter or clip path tweens only between the same function lists (`blur(0px)` to `blur(8px)`, never from `none`); remove the filter when it reaches zero.
|
|
30
33
|
- **Speed effects from speed.** Trails, smears and blur are earlier poses of the same function of time or a function of the velocity of its ease, so they are exact at any seek and vanish at rest.
|
|
31
34
|
- **A state owns its end.** A step rests at the end of its state (`ms = at(k + 1)`), so that frame must show this state's result. Clamped interval progress (`at(k)` to `at(k + 1)`) gets this for free; a scene that picks a discrete state or round from the time counts the boundary in the earlier one (`ms > at(k + 1)`), or every step rests on the first frame of the next.
|
|
32
35
|
- **Rest frames stay still.** Springs keep creeping after they look settled and re-rasterize text; snap to the exact end within 0.001. Keep an identity 3D transform at rest instead of removing it, and give `preserve-3d` elements `will-change: transform`: Chromium rasterizes a layer that once had a 3D transform differently, so a frame would depend on how it was reached.
|
|
33
36
|
- **Pictures** are `<img>` elements (a file in the document's `public/` or a URL): the viewer waits for every `<img>` in the scene and for the fonts before the first frame, so an export opens on the picture. A CSS background image or a picture drawn on a canvas is not waited for; a Three.js scene loads its textures in an async build.
|
|
34
|
-
- **Styles:** `frame()` (`gallery/lengths/scenes/kit/frame.ts`) puts the scene's CSS in `@scope`, so two scenes on a page can both style `.card`; a `<style>` written by hand is page-wide. Inside `@scope` a selector matches only below the frame's root. `cqw`/`cqh` inside an element that is itself a size container are that element's units, not the frame's.
|
|
37
|
+
- **Styles:** `frame()` (`gallery/lengths/scenes/kit/frame.ts`) puts the scene's CSS in `@scope`, so two scenes on a page can both style `.card`; a `<style>` written by hand is page-wide. Inside `@scope` a selector matches only below the frame's root. `cqw`/`cqh` inside an element that is itself a size container are that element's units, not the frame's. The page gives every `svg` `max-width: 100%`, which squashes an SVG placed inside a zero-size anchor (a label or icon pinned at a point): give such an SVG `max-width: none`.
|
|
38
|
+
- **Shared modules.** A scene imports other project files with relative `.ts` imports, so the scenes of one piece can share a kit (the frame, the type scale, the colour roles) and a module that draws the central picture and holds the example data: every page then draws the same picture at the same place from the same numbers (`gallery/lengths/scenes/kit/` is such a kit). Changing a shared module changes every page that imports it, so one person owns it ([Team](team.md)). Prefix its class names (`.sm-mark--ok`, not `.ok`): a scene's own `.ok` would take the shared styles, and a tween on a bare selector colours every matching mark on the board. Give it selector functions (`sel.slot(id, i)`) and coordinate functions (`slotCenter(id, i)`) so that pages don't compute positions that break when the shared drawing is resized.
|
|
39
|
+
- **Build once.** The reading page builds a scene when it comes within a screen of the window, while the page is idle (slides build theirs at once), and keeps it from step to step. Reading `state`, or a parameter that changes from step to step, while building makes the scene build again at every step that changes it, and the whole build then stops the page as the reader arrives; read them where the frame is drawn (`seek` in `defineScene`), or not at all when the timeline already covers every step. A build that sets many layout properties in `cqw`/`cqh` makes the browser work out styles once for each: keep a long piece's build to about a tenth of a second (Performance panel, or the long tasks while scrolling the reading page).
|
|
35
40
|
- **A list that makes steps** (clicks, messages, screens) needs as many declared states as items; nothing checks the two agree.
|
|
36
41
|
- **Seeking a one-state scene:** its reading view shows no step buttons (they appear only with more than one state); check seeking with the scrubber (`.pm-scene__scrub`).
|
|
37
42
|
|
|
@@ -130,7 +135,9 @@ Presmium sets `force3D: false` (so a frame looks the same however it was reached
|
|
|
130
135
|
- **Later steps:** `from()` and `fromTo()` apply their start values when the timeline is built. For something that appears in step k > 0, set its hidden state with `gsap.set` (e.g. `drawSVG: 0`, `opacity: 0`, `motionPath: { …, end: 0 }`) and tween it with `to()` or `fromTo(…, { immediateRender: false })` at `s(k)`. Without the `set`, a path drawn in step 2 shows fully drawn during steps 0 and 1.
|
|
131
136
|
- **A look computed from the time alone** (every element placed by a function of `t`) is `defineScene`, with eases from `gallery/lengths/scenes/kit/fx/time.ts`; a GSAP tween that only drives a clock adds nothing.
|
|
132
137
|
- **GSAP can't tween `var()` or `color-mix()`:** tween to `color('accent')`; GSAP 3 has no `className` tween. SVG shapes scaled or rotated about their centre need `transform-box: fill-box; transform-origin: center`.
|
|
133
|
-
- **Callbacks never fire** (`onUpdate`, `onComplete`, `tl.call`). Numbers: `textContent` with `snap: { textContent: 1 }` or `modifiers: { textContent: v => (+v).toFixed(2) }
|
|
138
|
+
- **Callbacks never fire** (`onUpdate`, `onComplete`, `tl.call`), and neither do timers the scene starts (`setTimeout`, `setInterval`): a step reached by seeking runs none of them, so a counter written in `onUpdate` stays frozen. Numbers: `textContent` with `snap: { textContent: 1 }` or `modifiers: { textContent: v => (+v).toFixed(2) }`, or one element per value, each shown with `autoAlpha` at its time. Text swaps: `tl.set(el, { textContent: 'lo' }, time)`, which seeking back undoes.
|
|
139
|
+
- **`x` and `y` don't take `cqw` or `cqh`:** move a frame-relative element with `left`/`top` in percentages, or with `xPercent`/`yPercent`.
|
|
140
|
+
- **`autoRound: false` in a small viewBox.** GSAP rounds pixel-like values to whole numbers, so a `strokeWidth` tween in a `0 0 160 90` viewBox jumps between 0 and 1; set `autoRound: false` on the tween.
|
|
134
141
|
- **Stagger with `repeat` or `yoyo`:** put them inside `stagger: { …, yoyo: true, repeat: 1 }`; on the tween they repeat the whole wave and run into the next step.
|
|
135
142
|
- **TextPlugin:** `preserveSpaces: true` keeps indentation in code.
|
|
136
143
|
- **Randomness:** `"random(...)"`, `stagger.from: 'random'` and `utils.shuffle` use `Math.random`; compute values with a seeded generator and pass function-based values.
|
|
@@ -165,7 +172,6 @@ export default animeScene(({ root, anime, at, color }) => {
|
|
|
165
172
|
- **SVG transforms:** `translateX`, `scale`, `rotate` are written as CSS `transform`. Put the position on an outer `<g transform>` and animate an inner `<g>` with `transform-box: fill-box; transform-origin: center`.
|
|
166
173
|
- **A from-value shows from time 0**, before its tween starts. Elements that join later need an opacity tween that keeps them hidden.
|
|
167
174
|
- **`loop` with a stagger `delay`** repeats the whole staggered group as one iteration; add one tween per element instead.
|
|
168
|
-
- **Arrowheads:** a `marker-end` shows before its line is drawn; draw arrowheads as separate shapes that fade in.
|
|
169
175
|
- **A motion path drawn right to left** turns its label upside down through `rotate`; use only `translateX`/`translateY` from it.
|
|
170
176
|
- **A head streak is not a drawable with string keyframes:** `['0 0', '0.92 1', '1 1']` splits its duration evenly and runs ahead of a `'0 0'`→`'0 1'` core; draw values past 1 wrap to the start. Drive the streak from one clock (head `E(p)`, tail `E((p − lag) / (1 − lag))`) and write it with the drawable's `setAttribute('draw', …)`.
|
|
171
177
|
- **A spring from rest rushes through a slow phase:** ease the slow part with a keyframe, then finish with `{ to, ease: spring(…) }`; its settling time may run past the state.
|
|
@@ -81,6 +81,7 @@ The Fourier fit, the de Casteljau scene and the universal-approximation storyboa
|
|
|
81
81
|
- **Strokes and writing**: Make Me a Hanzi (9,000+ characters with stroke outlines and medians for stroke-order animation; Arphic Public License graphics, LGPL data; <https://github.com/skishore/makemeahanzi>), KanjiVG (CC BY-SA 3.0), Quick, Draw! (CC BY 4.0; strokes keep their timing, so a drawing replays at its author's speed), Hershey single-stroke fonts.
|
|
82
82
|
- **Logos**: theSVG first, then the organisation's GitHub avatar, then the site's favicon; never redraw a logo. Use a brand's mark only for that exact brand.
|
|
83
83
|
- **Sound and music**: see `lengths.md`, "Sound and music".
|
|
84
|
+
- **Images in a scene** are imported, `import figure from '../figures/x.png'` (or `x.png?url`), not built with `new URL(…, import.meta.url)`: an imported PNG or JPEG of 48 KB or more is published as WebP, no wider than 2400 px.
|
|
84
85
|
- **Where each asset came from** goes in the reading view's notes: the source and what was changed, so the next author can replace it.
|
|
85
86
|
- **Motion data**: the CMU Graphics Lab motion capture database ("free for all uses", not for resale), for a stick figure or to check an IK gait.
|
|
86
87
|
- **Already in the kit**: seeded `mulberry32` and `hash`, `noise1`/`noise2`, `spring01`, `wobble`, `shake`, `smoothPath`, `samplePath`, `trail`, and the Fourier series with `bezierPath`.
|