Checklist the per-video bar — engine/sim
98.0/100
quantitative quality · weights learn from your reviews (
engine.sim.memory review longform-W30-Mon good|bad)⚠ 1 flag(s) — not yet ship-ready: copy_generic · see docs/strategy/VIDEO-CHECKLIST.md
Composition comp · template family · expected output
9:16 Reelpending1:1 Squarepending16:9 Widepending9:16 4Kpending1:1 4Kpending16:9 4KpendingGIF (SMS)pending
render pending — silent master not yet on disk
expected output: 0/7 rendered — same matrix the /media preview surfaces for this asset.
Composition layer × scene 6 scenes · 150.8s · comp_id + rendered still + tier + the script
| # | Layer (comp_id · still · tier) | Beat | Timecode | Motion | Logo | Audio | VO / on-screen / caption |
|---|---|---|---|---|---|---|---|
| 1 |  matches intent shared field signature-3d | open | 0–26.6s | spatial-parallax | icon·white-knockout | ♪ bed_in | One idea, nine outputs. Today we zoom into the one thing that decides if any of those outputs are good: the script. People skip it. They point a camera and hope. We are going to do the boring thing that wins, which is write it down first. By the end you will know how to write a short script with the help of a model, without letting the model sound like a robot. on-screen: The script is the blueprint |
expected on screen: red ground · icosa hero in the shared Signal Field · Signum leads · node-graph · spatial-parallax · icon·white-knockout logo · caption bottom-left spec (the prompt): comp_id shared Signal Fieldvisual node-graphshape icosaground redtreatment white-knockoutmotion spatial-parallaxpower summoninstrument summon→The script is the blueprint | |||||||
| 2 |  matches intent NumberedList template | teach | 26.6–53.2s | kinetic-build | icon·white-knockout | ♪ node_lock | Start with the spine. One sentence that says what the viewer should feel and do. Not a paragraph, one line. We literally type it at the top of the file and tell the model, every scene must serve this line or it gets cut. When you have a spine, you can throw away a clever joke that does not fit. Without a spine, every clever joke survives, and the script dies of clutter. on-screen: Write the spine in one line |
expected on screen: red ground · a NumberedList panel over a dimmed Signal Field · Signum leads · node-graph · kinetic-build · icon·white-knockout logo · caption bottom-left spec (the prompt): comp_id NumberedListvisual node-graphshape icosaground redtreatment white-knockoutmotion kinetic-buildpower morphinstrument morph+laser→Write the spine in one linecurate items, nodes | |||||||
| 3 |  matches intent ChecklistCard template | teach | 53.2–78.4s | kinetic-build | icon·liquid-chrome | ♪ node_lock | Now use the machine the right way. Do not ask it to write the script. Ask it to write three openings for the spine. Models are fast at options and bad at taste. So you take the options, you keep the one that sounds like a person, and you delete the two that sound like a press release. The model proposes, you decide. That split is the whole skill. on-screen: Make the model draft, you cut |
expected on screen: red ground · a ChecklistCard panel over a dimmed Signal Field · Signum leads · node-graph · kinetic-build · icon·liquid-chrome logo · caption bottom-left spec (the prompt): comp_id ChecklistCardvisual node-graphshape icosaground redtreatment liquid-chromemotion kinetic-buildpower morphinstrument morph+laser→Make the model draft, you cutcurate items, nodes | |||||||
| 4 |  matches intent BuildLog template | teach | 78.4–104.30000000000001s | kinetic-build | icon·white-knockout | ♪ node_lock | Here is the step nobody does. Read the draft out loud, on camera, before you trust it. Your mouth finds the lumps your eyes miss. A sentence that looks fine reads like a stack of bricks. So we feed the model the recording, or just the lines we tripped on, and say, shorten these until I can say them in one breath. A script is for the ear, not the page. on-screen: Read it out loud, fix the lumps |
expected on screen: red ground · a BuildLog panel over a dimmed Signal Field · Signum leads · node-graph · kinetic-build · icon·white-knockout logo · caption bottom-left spec (the prompt): comp_id BuildLogvisual node-graphshape icosaground redtreatment white-knockoutmotion kinetic-buildpower morphinstrument morph+laser→Read it out loud, fix the lumpscurate lines, nodes | |||||||
| 5 |  matches intent KpiGrid template | proof | 104.3–125.6s | receipts-count | icon·liquid-chrome | ♪ node_lock | This script you are hearing went through that exact loop. One spine, three openings from the model, one chosen by a human, the lumps read out and cut down. No invented numbers, no hype words, just a line that survives being spoken. That is the proof. If the method made this watchable, it can make yours watchable too. on-screen: We wrote this one this way |
expected on screen: red ground · a KpiGrid panel over a dimmed Signal Field · Signum leads · receipts · receipts-count · icon·liquid-chrome logo · caption bottom-left spec (the prompt): comp_id KpiGridvisual receiptsshape icosaground redtreatment liquid-chromemotion receipts-countpower receiptsinstrument spotlight→We wrote this one this waycurate kpis, statsLabels | |||||||
| 6 |  matches intent shared field signature-3d | resolve | 125.6–150.79999999999998s | coalescence | icon·white-knockout | ♪ bed_out | So the takeaway is simple. The script is the blueprint, and the blueprint is one spine plus lines you can say in one breath. Let the model give you options, you keep taste for yourself, and read it out loud before the camera ever rolls. You can see the scripts behind every video this week at office dot temerarii dot xyz. Write the spine first. The camera can wait. on-screen: Write the spine, then press record |
expected on screen: red ground · icosa hero in the shared Signal Field · Signum leads · coalescence · coalescence · icon·white-knockout logo · caption bottom-left spec (the prompt): comp_id shared Signal Fieldvisual coalescenceshape icosaground redtreatment white-knockoutmotion coalescencepower coalescenceinstrument coalescence→Write the spine, then press record | |||||||
Format stack 1 aspects · same scenes[], re-cropped
16:9
1920×1080
X/Twitter · YouTube · LinkedIn video
Channels 2 destinations
YouTubeBlog
Social captions supplemental published copy · per channel (comp_id level)
youtubeThe Script Is the Blueprint: One Spine, Model Drafts, and Reading It Out Loud
The script decides if any of your outputs are good, and most people skip it. They point a camera and hope. This walkthrough does the boring thing that wins: write it down first, the right way, with a model speeding up the parts that should be fast.
The method:
- Write the spine in one line. One sentence that says what the viewer should feel and do, not a paragraph. Type it at the top of the file and tell the model every scene must serve this line or it gets cut. A spine is a knife.
- Make the model draft, you cut. Do not ask it to write the script. Ask it to write three openings for the spine. Models are fast at options and bad at taste. You keep the one that sounds like a person and delete the rest.
- Read it out loud before you trust it. Read the draft out loud, on camera, before you trust it. Your mouth finds the lumps your eyes miss. A sentence that looks fine reads like a stack of bricks. Feed the recording back, or just cut the lines that tripped you, until it can be said in one breath.
This script went through that exact loop, one spine, three openings from the model, one chosen by a human, the lumps read out and cut. No invented numbers, no hype words, just a line that survives being spoken.
The takeaway: the script is one spine plus lines you can say in one breath. Let the model give you options, keep taste for yourself, and read it out loud before the camera rolls. See it at office.temerarii.xyz.
Keywords: scriptwriting, video script, content strategy, plain language, model drafting, voiceover.