Back Office · office.temerarii.xyz
One asset, all the way in — composition, the wireframe + storyboard, the output format stack, and the template, all read from the SAME content-index record. The expected output matches what /media surfaces for this post.
post longform-W40-Thukind longformweek W40date 2026-10-08campaign longform-youtubepillar multimediabeat asset videoduration 186.7sground whitescenes 7

Checklist the per-video bar — engine/sim

100.0/100
plain languagevo coverageno dead airuniquenesscaption fitcompletenesscleanliness
quantitative quality · weights learn from your reviews (engine.sim.memory review longform-W40-Thu good|bad)
✓ all static checks pass — one-focal/scene · tier-by-beat · one-track caption · colorway · cast+shape correct · no banned/fabricated. (audio + visual tiers verify on the rendered finals — Phase 2)

Composition comp · template family · expected output

composition LongFormChaptersfamily / template LongFormChapters
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 7 scenes · 186.7s · comp_id + rendered still + tier + the script

#Layer (comp_id · still · tier)BeatTimecodeMotionLogoAudioVO / on-screen / caption
1s1
matches intent
shared field
signature-3d
open0–26.3sspatial-parallaxicon·color♪ bed_in
Defining who you are includes a question most brands answer with stock photos. What does your world look like when you draw it. Today is illustration. Stock images make you look like everyone who bought the same stock. Custom illustration makes you unmistakable. By the end you will know how to build a small, consistent illustration system you can extend forever, using code and AI without it looking like generic AI slop.
on-screen: Draw your own world
expected on screen: white ground · octa hero in the shared Signal Field · Lumen leads · node-graph · spatial-parallax · icon·color logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual node-graphshape octaground whitetreatment colormotion spatial-parallaxpower summoninstrument summon→Draw your own world
2s2
matches intent
ChecklistCard
template
teach26.3–52.900000000000006skinetic-buildicon·wireframe♪ node_lock
Step one. An illustration system is a short rule list, not endless options. Choose your line weight, your two or three colors, and one shape language, say everything is built from circles and straight lines. Write those rules down. The whole point is that a stranger could draw a new icon next week and it still looks like yours. Constraints are what make a set of drawings feel like one hand made them.
on-screen: Pick a few rules, not a style
expected on screen: white ground · a ChecklistCard panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id ChecklistCardvisual node-graphshape octaground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Pick a few rules, not a style menucurate items, nodes
3s3
matches intent
NumberedList
template
teach52.9–79.2skinetic-buildicon·wireframe♪ node_lock
Step two. For clean, repeatable shapes, draw them as vector SVG, even with code. A circle, a line, a path are just a few lines of text. You can generate a whole icon set from a small script that stamps out the same stroke width and colors every time. This kills the drift you get when a human eyeballs each one. The machine guarantees the line weight is identical across fifty icons.
on-screen: Build simple shapes in code
expected on screen: white ground · a NumberedList panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id NumberedListvisual node-graphshape octaground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Build simple shapes in codecurate items, nodes
4s4
matches intent
ProcessFlow
template
teach79.2–108.30000000000001skinetic-buildicon·wireframe♪ node_lock
Step three. When you do want richer art, use an image model like Flux through a tool such as Replicate, but do not let it run wild. Lock the prompt to your color words and your shape language, generate several, then keep only the ones that obey your rules. The model is a fast intern, not the art director. You are the filter. That filtering is the difference between a custom world and the same blurry look everyone else ships.
on-screen: Use AI for texture, then control it
expected on screen: white ground · a ProcessFlow panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id ProcessFlowvisual node-graphshape octaground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Use AI for texture, then control itcurate nodes, steps
5s5
matches intent
CheatSheet
template
teach108.3–135.6skinetic-buildicon·wireframe♪ node_lock
Step four. If your world has recurring figures, lock them. Save a reference sheet, the same proportions and colors, and make every new image start from it. Consistency is what turns a pile of pictures into a cast people recognize. We did this with our own set of characters. They show up across the calendar built from the same rules, so they read as one family instead of strangers who happen to share a palette.
on-screen: Reuse characters, do not reinvent
expected on screen: white ground · a CheatSheet panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id CheatSheetvisual node-graphshape octaground whitetreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Reuse characters, do not reinventcurate nodes, points, uses
6s6
matches intent
ComparisonTable
template
proof135.6–159.4sreceipts-counticon·wireframe♪ node_lock
Honest proof. The shapes and figures across this campaign come from a small rule set and code, not from a stock library. Same line logic, same colors, same hands every week. When we need a new piece, we extend the rules instead of starting over. No claim of awards, just a world that stays one world because the rules, not a mood, decide what belongs.
on-screen: Our shapes come from rules
expected on screen: white ground · a ComparisonTable panel over a dimmed Signal Field · Lumen leads · receipts · receipts-count · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id ComparisonTablevisual receiptsshape octaground whitetreatment wireframemotion receipts-countpower receiptsinstrument spotlight→Our shapes come from rulescurate colA, colB, rows, statsLabels
7s7
matches intent
shared field
signature-3d
resolve159.4–186.70000000000002scoalescenceicon·color♪ bed_out
So illustration is a short rule list, simple shapes you can code, AI used as a filtered intern, and characters you reuse instead of reinvent. Do that and you get a world nobody can copy because it is yours by construction. Our whole illustration set is on display at office dot temerarii dot xyz. Pick your three colors and one shape language tonight. The rest is just stamping out the world you already decided on.
on-screen: See the set at office.temerarii.xyz
expected on screen: white ground · octa hero in the shared Signal Field · Lumen leads · coalescence · coalescence · icon·color logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual coalescenceshape octaground whitetreatment colormotion coalescencepower coalescenceinstrument coalescence→See the set at office.temerarii.xyz

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)

youtubeHow to build a custom illustration system with AI without the slop Stock images make you look like everyone who bought the same stock. Custom illustration makes you unmistakable. This video shows how to build a small, consistent illustration system you can extend forever, using code and AI without it looking like generic AI slop. The method: Pick a few rules, not a style. Choose your line weight, two or three colors, and one shape language, like everything built from circles and straight lines. Write the rules down. The test is whether a stranger could draw a new icon next week and it still looks like yours. Constraints are what make a set of drawings feel like one hand made them. Build simple shapes in code. A circle, a line, a path are a few lines of SVG text. Generate a whole icon set from a small script that stamps out the same stroke width and colors every time. The machine guarantees the line weight is identical across fifty icons, which kills the drift you get when a human eyeballs each one. Use AI for the richer art, then control it. Generate a base with an image model like Flux through Replicate, but lock the prompt to your color words and shape language. Generate several, keep only the ones that obey the rules. The model is a fast intern, you are the filter. Reuse characters, do not reinvent them. Save a reference sheet, same proportions and colors, and make every new image start from it. That is what turns a pile of pictures into a cast people recognize. The shapes and figures across our campaign come from a small rule set and code, not a stock library. See the whole set at office.temerarii.xyz. Keywords: illustration system, AI illustration, Flux, Replicate, icon design, brand characters, SVG.

Cross-links every lens is a view on this one record