设置: 构建语音架构

创作一部时长 20 秒的动态口语短片，名为“BUILD THE FLOOR”（搭建地板）。将其视为一段完全通过文字排版呈现的原创微型演讲。每一行文字都会改变画面的结构。最终的宣言必须建立在之前文字所构建的实体之上。提交一个独立的HTML文件，分辨率为 1080×1080，目标帧率为 60fps，使用SVG和/或 Canvas 技术。嵌入所有必需的素材。美术指导：背景 #102820；纸张 # F4E9D5；结构强调 # F2B544。营造一个具有公共纪念碑般规模和气势的文学编辑世界。演讲部分使用高对比度衬线字体，重要文字使用粗体无衬线字体，小型时间标记使用简洁的等宽字体。请勿使用破旧的抗议海报元素、扩音器、旗帜、人群、麦克风或素材库视频。原文——逐字逐句，顺序不变。“我们被告知要等待。”“于是我们体会到了等待的沉重。”“然后一个声音让出了空间。”“接着是另一个。”“现在，这空间属于我们了。”不要将这些话归于任何真实人物。不要添加任何额外的口号。完整的演讲必须在没有音频的情况下也能被理解。中心视觉规则：文字具有特定的结构角色。“等待”是门楣。“等待”是悬垂的重物。“声音”是支撑。“空间”是开口。“空间”是平台。这些角色必须体现在实际的字母形式中，而不是体现在放置在文字后面的单独插图中。故事板 0.00–3.00 “我们被告知要等待。”这句话占据了下方一个巨大的“等待”字样下方狭窄的水平开口。开口逐渐缩小，但句子仍然清晰可读。这种压力来自布局，而不是镜头抖动。3.00–6.50 “于是我们体会到了等待的沉重。”以口语短语组的形式展现这句话。 “等待”沿着两条细长的垂直线缓缓下降。这些线变成了悬索。基线在视觉负荷的作用下，通过可控的矢量变形而弯曲。6:50–10:00 “然后一个声音腾出了空间。”“声音”垂直进入，旋转成一个结构支撑。它抬起了基线的一端。“空间”中字母“O”的倒角变成了一个开口。下一个构图透过这个开口显现出来。10:00–12:50 “然后又一个。”第二个支撑出现。倾斜的结构恢复水平。这是影片中第一个完美的水平对齐时刻。用短暂的静止来使这种变化显得自然而然。12:50–16:00 “现在，地板属于我们了。”之前的结构性词语组合成一个平台。最后一句话落在上面。“我们”通过共同的对齐和比例得到强调，而不是通过单独的炫目效果。16:00–18:00 保持完整的宣告。支撑线中传递出一丝轻微的张力释放。没有新的文本。没有躁动的装饰性动作。 18:00–20:00 摄像机跟踪已完成的平台下方。平台底部充满画面，成为入口处的“WAIT”字样横梁。在完全遮挡的情况下，将隐藏元素恢复到初始位置。展现完整的开场构图。编排：不要强行将语音与舞蹈节拍结合起来。使用基于短语、停顿和重音的提示单。短语的呈现应遵循朗读顺序。至少留出一个400毫秒的绝对静止时刻。使用临界阻尼运动来模拟负载的沉降。仅对基线使用受控弹性形变。使可用空间的变化体现情感的推进。字体实现：使用实际的字形路径来表示结构词。在形变过程中保持字腔和清晰度。对于不相关词语之间的过渡，使用遮挡或匹配的结构边缘；不要插值任意路径点。在当前朗读的短语变成建筑结构之前，保持其可读性。工程：将 window.seek(t) 实现为确定性场景渲染器。不使用累积物理效果、自主CSS过渡或随机变化值。使用时间重映射曲线和解析变换。使用 20 秒周期性时间轴，并将 render(0) 设置为 render(20)。提供 n/60 的固定步长访问，导出 1200 帧。音频为可选，默认关闭；不依赖浏览器语音合成或自动播放。质量门：检查八个代表性帧。检查语音朗读顺序是否正确。检查结构变换是否会在观众阅读之前破坏文本行。检查最终平台是否由之前的元素清晰地组装而成。返回完整的HTML实现。

Create a 20.00-second kinetic spoken-word film titled “BUILD THE FLOOR.”  Treat this as an original miniature speech staged entirely through typography. Every line changes the architecture of the frame. The final declaration must stand on something the earlier words physically constructed.  Deliver one self-contained HTML file, 1080×1080, targeting 60fps, using SVG and/or Canvas. Embed all required assets.  ART DIRECTION Background #102820. Paper # F4E9D5. Structural accent # F2B544. A literary editorial world with the scale and confidence of a public monument. Use a high-contrast serif for the speech, a heavy grotesque for load-bearing words, and a restrained monospace for small timing marks. No distressed protest-poster clichés, megaphones, flags, crowds, microphones, or stock footage.  ORIGINAL SPEECH — EXACT WORDS, EXACT ORDER “We were told to wait.” “So we learned the weight of waiting.” “Then one voice made room.” “Then another.” “Now the floor belongs to us.”  Do not attribute these words to any real person. No extra slogans. The complete speech must be understandable without audio.  CENTRAL VISUAL RULE Words have assigned architectural roles. WAIT is a lintel. WAITING is a suspended load. VOICE is a support. ROOM is an opening. FLOOR is a platform. These roles must emerge from the actual letterforms, not from separate illustrations placed behind text.  STORYBOARD 0.00–3.00 “We were told to wait.” The sentence occupies a narrow horizontal opening beneath a massive WAIT. The opening slowly compresses while the sentence remains readable. The pressure comes from the layout, not camera shake.  3.00–6.50 “So we learned the weight of waiting.” Reveal the sentence in spoken phrase groups. WAITING lowers on two fine vertical rules. Those rules become suspension cables. A baseline bends under the visual load using a controlled vector deformation.  6.50–10.00 “Then one voice made room.” VOICE enters vertically and rotates into a structural support. It lifts one end of the baseline. The counter of the O in ROOM becomes an aperture. The next composition is revealed through that aperture.  10.00–12.50 “Then another.” A second support arrives. The tilted structure levels. This is the first moment of perfect horizontal alignment in the film. Use a brief stillness to make the change feel earned.  12.50–16.00 “Now the floor belongs to us.” Earlier structural words assemble into a platform. The final sentence settles onto it. US receives emphasis through shared alignment and scale, never through a separate flashy effect.  16.00–18.00 Hold the complete declaration. A slight release of tension travels through the supporting rules. No new text. No restless decorative motion.  18.00–20.00 The camera tracks beneath the completed platform. Its underside fills the frame and becomes the opening WAIT lintel. Under full occlusion, return hidden elements to their initial positions. Reveal the exact opening composition.  CHOREOGRAPHY Do not force the speech onto a dance beat. Use a cue sheet based on phrases, pauses, and emphasis. Phrase reveals should follow reading order. Allow at least one 400ms moment of absolute stillness. Use critically damped motion for settling loads. Use controlled elastic deformation for the baseline only. Make the changes in available space carry the emotional progression.  TYPE IMPLEMENTATION Use actual glyph paths for structural words. Preserve counters and legibility during deformation. For transitions between unrelated words, use occlusion or matched structural edges; do not interpolate arbitrary path points. Keep the currently spoken phrase readable before it becomes architecture.  ENGINEERING Implement window. seek(t) as a deterministic scene renderer. No accumulated physics, autonomous CSS transitions, or changing random values. Use time-remapping curves and analytic transforms. Use a 20-second periodic timeline with render(0)=render(20). Provide fixed-step access at n/60 for 1200 export frames. Audio is optional and off by default; no browser speech synthesis or autoplay dependency.  QUALITY GATE Inspect eight representative frames. Check that the speech reads in the correct order. Check that structural transformations never destroy a line before the audience can read it. Check the final platform is visibly assembled from earlier elements. Return the complete HTML implementation.

提示取自 邮政 @Gdgtify.