GeezerVison is showing motiongit.com — Clay Lab · Knowledge Explainer · @adunext · MotionGit
Clay Lab · Knowledge Explainer Track clay objects as they split, attach and move. For changes in objects and processes, with AI-guided icon and diagram selection. Optional presenter opening that shrinks into a circular window.
Full template A cohesive set of composable animation and motion components for creating an entire video around your content.
Beta Ask your AI assistant to check for updates before creating Updates and feedback All templates are currently in Beta. The library receives multiple updates each day. Ask your AI assistant to check for and apply available updates before you start. Save the new package separately and keep your old package and projects.
Ask your AI assistant to report a failure : review the sanitized conversation excerpts and errors before submission, and keep the receipt ID. Confirmed template defects are usually fixed in the following day’s release; complex cases may take longer. You can also comment below. ↓
default 1920×1080 · 60fps
Get and use this template Free Commercial use
View license details Finished videos may be used commercially. You may modify the template but not redistribute it.
♡
Support goals met — free download is open Douyin Your support helps more good templates grow.
Last synced: 10/29/2026, 1:40:29 PM
Download rules What does “goals met” mean?
The creator sets likes, favorites, comments and recommendation goals for their own social post. When supporters bring all four counts to those goals, the template opens for everyone. The figures below come from that original post.
Progress averages the four completion rates, each capped at 100%. Extra support in one metric cannot replace another goal. Downloads open only after all four goals are met.
Creator contributions ≈ 6,305 source lines
The author has published 13 free open-source templates, with 40,932 downloads in total.
Your support helps more good templates grow.
Full share text 0.76 复制打开抖音，看看【阿杜Next的作品】老规矩，全部免费开源，但制作不易，需要你的点赞留言... https://v.douyin.com/hjTyP8mfZes/
[email protected] :5pm 03/29 zgB:/
Scan for Douyin Scan with your phone camera or Douyin to view the original post, or copy the full share text into Douyin. Overall unlock progress 100%
Sign in to get template Free signup. Free template download. One account connects AI asset matching, music and more creative services.
View details · uses, materials and requirements Before you download
Suitable for 黏土实验台知识讲解：让对象、数据和关系随讲解发生变化，支持可选真人开场后缩成圆窗。 Prepare these inputs What question should this video make clear? Send an existing script if you have one. · Full script and viewpoint, without dropping negations, conditions, numbers, units or qualifiers. · With only a topic or outline, the assistant checks sources and drafts the full script; the text must be settled before production. Never treat examples as user content or fabricate voices or measured timings. Is there data, material or a source that must be kept? · This production's data, sources, original records and the line between assumption and illustration. · Build a fact sheet first; fictional data must be marked illustration, and facts need traceable sources. Use an existing voiceover, the presenter's original audio, or a voiceover generated from the script? · Narration at original speed, or the original audio of the live-action video. · Choose none only when there is explicitly no voiceover; otherwise keep needs-binding, and the assistant fetches the actual voice via the current MotionGit API. If you want to appear on camera, provide the full talking-head video. · Opens with the presenter full-frame, then the same carrier continuously shrinks into a small circular window. · No presenter by default; no sample avatar or cartoon character is added. Send a caption file too if you have one. · Keep measured sentence-level captions for the full text; word-level actions need real word evidence. · The assistant measures sentence boundaries from the real audio and saves the evidence; with copy only, it cannot claim measured timing. Send specific background music if you have it. · Prefer matching this production's background music via the current MotionGit music API, keeping the license and SHA; a track the user has licensed can also be bound. · Match a clearly licensed track first; keep needs-binding if none is chosen. Set none only when the user explicitly wants no music; never fall back to the author's old BGM. Send original records, photos of real objects or evidence images if you have them, to verify the knowledge basis. · Used by the assistant to verify and extract facts from source materials; this version has no shot that embeds original images, so showing them requires a separate extension. · Choose a pure-diagram mechanism based on this production's verified data; never fake evidence images. Full script · Keep the original meaning, negations, conditions, degree, numbers and units, and the order of argument; short labels don't replace full statements. Claims and knowledge basis · First label facts, assumptions, examples or analogies; data and diagrams must come from the same model. Visual changes and reasoning · Correlation lines, causal arrows, condition gates and completion states each mean something different; don't mix them.; Use object IDs to keep before/after states and sources; splits and merges must be conserved, with nothing added or lost out of nowhere.; When capacity is short, switch groups or split shots; don't silently drop sentences, shrink text or cram everything into one overview diagram.; For quantity operations, first separate mathematically referenced records from explicitly additive materials. Inputs, operations, graphic scale and results come from the same model; only same units can be added, subtracted or turned into ratios; multiplying or dividing constants are dimensionless; units are not converted automatically. Derived fractions keep their true values; displayed decimals are not exact results; across shots use the public full model clock and don't reset objects.; Known input readings for static value categories and formulas keep the declared facts, and formula results are derived from those true values; bar/text entrances are only reveals. Real before/after changes, source accumulation, item-by-item arrival and material transfer use the existing explicit models; decimals=0 or rounding is not a discrete event; hidden presentation-only fields are not visible counts or semantic progress.; Numbers must be shown in full; a truncated prefix is not an interim reading; experiment only presents the given before/after snapshots, geometric interpolation is not an intermediate measurement, and fixed line observations are shown at their real coordinates.; secondBeat is off by default; choose it only when the copy truly has a two-item comparison, an adjacent observed change or a before/after difference; bind this segment's stable object IDs, same-unit real data and measured local focus/result/end, and never repeat comparisons just to add motion. When capacity or the reading window is short, keep all content and split the segment.; In the snapshot's second beat, highlights, condition bars and differences all reference the same object ID of the current explicit step, and the last object stays after it ends.; Whole relation labels and conditions containing numbers fade in as complete fields, avoiding wrong number prefixes.; Optional subject-first layout · Basic diagrams can choose content-first as needed; native mechanisms keep their original layout, and the film title never overpowers the explanatory content. Known conditions kept across segments · Up to 3 explicit claim-bound conditions, referenced via contextFactIds in segments; keep them in the same position throughout, never show future results early, and use only measured sentence starts for availableFrom. If capacity runs out, switch group or split the segment. Production route · Use knowledge-story/1 for new topics; source-performance only reproduces an explicitly bound original revision and cannot replace a new script. Source performance binding · You must provide the original clock declared by the current source performance, measured sentence/word evidence, all display text and the original model; the word-by-word route uses real word intervals, while the fixed-action route binds measured sentence boundaries and original action events separately, without estimating word timings. Missing items are rejected. Knowledge question · Concisely summarize the question this video answers, keeping its conditions and qualifiers; don't rewrite the conclusion as a sensational headline. Aspect ratio · Choose a landscape or a separate portrait layout based on where it will be published; the fixed source-performance route keeps only the original aspect ratio. Language · Use the same language as this video's script and audio; check full glyph coverage and reading capacity separately for Chinese and English. Run locally with · Runtime verification Requires file and command access plus the declared runtime dependencies.
Version and license 2.7.4 · licenses/TEMPLATE-LICENSE.txt
Run locally with Requires file and command access plus the declared runtime dependencies.
Runtime verification Not verified Summary from the template introduction (README.md) and this release’s motion.json, params.schema.json and guide.json. Missing details remain unverified.
How do I use the download? Download .motion and keep the original. Give it to an AI assistant with file and command access. Check runtime dependencies, preview, then export. Wait for the attachment upload to finish, then ask the assistant to confirm it can read the file. If it keeps spinning, retry or use the ZIP compatibility download. A .motion file is already a ZIP container; do not compress it again. If it remains unreadable, place the original in a local project folder the assistant can access. Read the file handoff guide →
Give it to your AI assistant Guide me through making a video with this template. Package details and handoff text
file knowledge-explainer-clay-2.7.4.motion
size 31 MB
sha256 63063d3c7a3c… Instructions for AI → Your download includesyour personal API key . Keep .motion files containing your personal key private. Share the template link; visits do not earn credits. Referral rewards require an eligible friend’s actual payment.
Finished videos may be used commercially. You may modify the template but not redistribute it.
Runtime details
Runtime native-python-node/1
Output 1920×1080
Frame rate 60 fps
Duration 03:27
Params 19
Styles 1
Network offline
Spec 0.1.0-draft.1 About Animated explainers 625625 downloadsUpdated 10 days ago Authoring tool Not declared Static checks Package structure and path safety Root manifest: motion.json Parameters and style branches User guidance and duration adaptation File integrity Personal information and credentials Instructions and licenses Authoring provenance
Passed: package structure and automated scans. These checks do not verify copyright ownership or guarantee safe code.
Free template Download free. Customize and create again. Free to download, with no top-up required. Reuse and customize it: replace text, images and audio, and adjust visuals and animation.
Personal and commercial video production · Individuals and businesses may make and publish videos under the template license. Unless authorized, you may not repost or resell the template package or source code, including modified versions, or use them to develop a third-party online service.
Free assets and music Image, video and sticker search plus 1,000+ music tracks your AI assistant can match. Asset guide → Optional generation Image, voice and video generation are optional and billed separately. Creator API → First time? Follow the tutorial from template to finished video. Open tutorial → Template code follows the license on this page; fonts, images and music keep their own licenses. Free asset services have a daily allowance.
About this template 黏土实验台 · 知识讲解模板
制作前提醒｜模型能力与 Token 消耗
模板已内置大量风格化图标、动效与动画组件。制作新主题时，通常仍需根据文案补绘大量专用图示，并编写或调整动画代码，对大模型的代码能力要求较高。仅建议使用 GPT Sol 6.1 或具备同等及以上代码能力的大模型制作。
复杂图示、长文案和多轮修改会增加 Token 消耗；请留意所用模型的 Token 用量、额度与费用。建议先确认文稿和分镜，制作短样片后再分章推进，减少反复重做。
像在桌面上做一次可看懂的小实验。柔软的黏土对象被拆开、聚合、附着或沿通道移动，观众能追踪“原来是什么、受到了什么作用、后来去了哪里”。适合讲对象拆分合并、界面附着、受力变化和搬运路径，尤其适合需要保留对象来龙去脉的过程说明。
可使用面积守恒的拆分与合并、头尾对象在界面附着、受力后分裂、后代路径追踪，以及局部放大后原位返回等原生表达。可换本期对象标签、有限数量与比例、通道条件及旁白。这里的黏土形变是目录限定的教学表达，不等同于任意化学反应、真实流体或材料仿真；类比与事实应分别说清。
助手可以从102个基础与科学图标、13类基础图示中补齐概念定义、步骤、对比和证据；需要根据数值判断去向时，可选条件分流，需要展示数量关系时可选数量运算。应先确认对象和变化规则，再决定采用黏土机制还是基础图示，而不是把不同概念都画成同一团物质。
把完整文稿或已有配音交给AI助手即可开始。助手先梳理条件、对象和出处，再安排前后状态与动作，按实际声音时长分配阅读空间，预览后检查完整成片。是否补文稿、配音、真人素材和数据来源，取决于本期知识内容。可按发布位置制作横屏或独立竖屏；默认以纯图解呈现；若选择真人，可用本期连续口播全屏开场，再连续缩成圆形小窗陪讲。
本页预览采用所选完整中文原片。原片中的固定实验演出保留原有内容边界，新稿只选择目录已说明的可变机制，不把演示效果当作新主题的科学证据。
给AI助手的起始提示： “用黏土实验台模板讲解这份材料。先核对对象、变化条件与来源，告诉我哪些地方适合拆分、附着或搬运，选择相应图标和图示，再按本期声音制作。”
模板包持续完善。
支持达标开放 ：查看本批知识讲解支持作品 。
开放条件：点赞严格超过5000、收藏严格超过500、评论严格超过500、推荐严格超过100，四项均满足后开放模板下载。当前下载未开放。
A template is a starting point Ask your AI assistant to adjust timing, colors, text, or aspect ratio while keeping the template’s visual style.
Developer resources What you can replace Parameter Description Type Constraints Default creationProfileRequired — enum knowledge-story/1 / knowledge-source-performance/1 —sourcePerformance— object — —durationSeconds— number ≥ 4，≤ 600 —fps— enum 24 / 25 / 30 / 50 / 60 —layout— enum landscape / portrait —audio— object — —title— string ≤ 100 characters，≥ 1 characters —script— string ≤ 50000 characters，≥ 1 characters —language— enum zh-CN / en —claims— array — —sources— array — —captions— array — —segments— array — —presenter— — — —composition— object — —contextFacts— array — —timeCorner— object — —sentenceEvidenceRequired for spoken narration. path: evidence file relative to --inputs; sha256: SHA256 of that file; audioSha256: SHA256 of the narration file (presenter narration: the presenter video). The file is JSON {audioSha256, method, captions} with captions identical to the production captions. See SENTENCE-EVIDENCE.md and sentence-evidence.schema.json. object — —wordEvidence— object — —
Update this template with your AI assistant Give the old template to the assistant and send the following sentence. The new version is saved as a separate file, and the old packages and projects are retained.
Please press https://motiongit.com/update.md to update this MotionGit template directly, keep the old package and existing project, and tell me which new file to use when you're done. v2.7.4 Latest Nov 2, 2026
出片速度提升：Mac 上约为原来的 1.3–3 倍，带真人出镜视频时约 2 倍，实际速度随设备、电脑繁忙程度和内容变化；成片画面与之前完全一致。出片中途出错时会及时报错退出，不再一直卡住。
v2.7.3 Nov 2, 2026
使用已有配音时，句级时间证据的格式有了公开说明和示例，字段排列顺序不同不再被误判为不符，校验不通过时会指出具体哪一项不符；分镜 recipe 补充了从画廊样例到生产参数的写法示例；绑定较大的人物视频时，预览和出片不再中途崩溃。
v2.7.2 Oct 27, 2026
No release notes have been provided for this version.
Show earlier updates (1) v2.5.2 Oct 26, 2026
No release notes have been provided for this version.
More templates by @adunext 
16:900:19View template
16:903:29View template
16:903:12View template
16:900:23View template
3:402:14View template
4:302:19View template
4:303:50View template
16:903:20View template