engine.sim.memory review longform-W25-Thu good|bad)| # | Layer (comp_id · still · tier) | Beat | Timecode | Motion | Logo | Audio | VO / on-screen / caption |
|---|---|---|---|---|---|---|---|
| 1 | ![]() matches intent shared field signature-3d | open | 0–23.4s | spatial-parallax | icon·3d-extrude | ♪ bed_in | Today is illustration, and we are showing how we keep a consistent visual world without a human redrawing every frame. We point it at ourselves first, our own characters and shapes. By the end you will know how we built figures in code, how we keep them on model every time, and the honest limits of doing it this way. Real method, no overselling. on-screen: Illustration with a steady hand |
expected on screen: white ground · knot hero in the shared Signal Field · Magister leads · node-graph · spatial-parallax · icon·3d-extrude logo · caption bottom-left spec (the prompt): comp_id shared Signal Fieldvisual node-graphshape knotground whitetreatment 3d-extrudemotion spatial-parallaxpower summoninstrument summon→Illustration with a steady hand | |||||||
| 2 | ![]() matches intent NumberedList template | teach | 23.4–48.9s | kinetic-build | icon·wireframe | ♪ node_lock | Our main figure, an astronaut, we traced once into a clean vector path and locked it. The rule we wrote for ourselves is delete only, never re parse. We never re trace it, because every re trace drifts. We keep that one SVG and only ever remove the stray bits. A locked vector is an illustration that stays exactly itself across a year of videos, which hand drawing cannot promise. on-screen: Trace once, keep the vector forever |
expected on screen: white ground · a NumberedList panel over a dimmed Signal Field · Magister leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left spec (the prompt): comp_id NumberedListvisual node-graphshape knotground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Trace once, keep the vector forevercurate items, nodes | |||||||
| 3 | ![]() matches intent ChecklistCard template | teach | 48.9–75.9s | kinetic-build | icon·wireframe | ♪ node_lock | We have a small cast of characters, each defined in code, not as a one off drawing. They live in a Python file with their traits and roles. Because they are code, we can place them into any scene, any video, and they come out identical. The tool is plain. Define the character once as data, then summon it anywhere. The drawing stops being a thing you redo and becomes a thing you call. on-screen: Build the cast in code |
expected on screen: white ground · a ChecklistCard panel over a dimmed Signal Field · Magister leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left spec (the prompt): comp_id ChecklistCardvisual node-graphshape knotground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Build the cast in codecurate items, nodes | |||||||
| 4 | ![]() matches intent AnnotatedDiagram template | teach | 75.9–101.4s | kinetic-build | icon·wireframe | ♪ node_lock | Each figure carries one meaning so the illustration teaches, not just fills space. One character stands for measuring, one for signaling, and so on. When a video is about a certain idea, we bring out the character that fits it. The model picks the match from our cast file, we approve. So the picture is doing work, telling you which idea is on screen, not just sitting there looking nice. on-screen: Give each character one job |
expected on screen: white ground · a AnnotatedDiagram panel over a dimmed Signal Field · Magister leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left spec (the prompt): comp_id AnnotatedDiagramvisual node-graphshape knotground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Give each character one jobcurate callouts, nodes | |||||||
| 5 | ![]() matches intent WireframeMock template | teach | 101.4–127.30000000000001s | kinetic-build | icon·wireframe | ♪ node_lock | Here is the honest proof, and its honest limit. Because our figures are locked vectors, the same astronaut shows up identical across the whole week, no drift, no off model frame. That consistency is real and you can see it. The limit is just as real. This style is clean and code native, not painterly. We are not pretending it is fine art. It is a steady system that never wanders. on-screen: Same figure, every single time |
expected on screen: white ground · a WireframeMock panel over a dimmed Signal Field · Magister leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left spec (the prompt): comp_id WireframeMockvisual node-graphshape knotground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Same figure, every single timecurate nodes | |||||||
| 6 | ![]() matches intent RankList template | proof | 127.3–152.1s | receipts-count | icon·wireframe | ♪ node_lock | So illustration, our way. Trace the figure once and lock the vector, delete only. Build your cast as code so you can place it anywhere. Give each character one job so the picture teaches. Accept the trade, steady and consistent over painterly. That is a visual world that holds for a year. See our cast live at office dot temerarii dot xyz, then go lock your first figure. on-screen: Lock it once, summon it anywhere |
expected on screen: white ground · a RankList panel over a dimmed Signal Field · Magister leads · receipts · receipts-count · icon·wireframe logo · caption bottom-left spec (the prompt): comp_id RankListvisual receiptsshape knotground whitetreatment wireframemotion receipts-countpower receiptsinstrument spotlight→Lock it once, summon it anywherecurate rows, statsLabels | |||||||
| 7 | ![]() matches intent shared field signature-3d | resolve | 152.1–157.1s | coalescence | icon·3d-extrude | ♪ bed_out | Staff training: teams that actually execute on AI on-screen: Chapter 6 |
expected on screen: white ground · knot hero in the shared Signal Field · Magister leads · coalescence · coalescence · icon·3d-extrude logo · caption bottom-left spec (the prompt): comp_id shared Signal Fieldvisual coalescenceshape knotground whitetreatment 3d-extrudemotion coalescencepower coalescenceinstrument coalescence→Chapter 6 | |||||||