Files

T

soroush.asadi 7ed2ccc414 feat(remotion): Instagram channel-promo template + taste system + design-quality kit

The reference-round workflow, run end to end for a real template:

Taste system (how we learn the user's taste, persisted):
- references/TASTE_PROFILE.md (living design contract) + references/README.md (the
  daily loop) + a "reference round" stage in docs/TEMPLATE_BRIEF.md (provide refs or
  I suggest+mock directions).

Design-quality before/after:
- HeroDemo — the fix recipe vs the faint default: layered-depth background, a proper
  big video type scale, and a bold composed focal object. (Backgrounds were naked,
  text too small, scenes had no objects.)
- YaldaSofreh3D + IGPromoDirections + IGProfileMock — reference-match proofs
  (low-poly 3D, 3 IG-promo style directions, the realistic IG-light page).

Instagram channel-promo template (the deliverable — a flexible 5-scene FlexStory):
- igkit + 5 blocks: IGIntro, IGProfile (realistic IG-light profile, scales to all
  aspects), IGFeed (post grid), IGStats (animated count-up), IGFollowCTA (Follow taps
  to "Following").
- FlexStory gains a `finish` toggle so the IG-light scenes render clean (no brand
  grade). INSTAGRAM_PROMO preset + 3 aspect comps in Root.

Verified: a still of every scene at 9:16 renders clean; full preview MP4 rendering.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-25 09:16:31 +03:30

4.2 KiB

Raw Blame History

Taste Profile — character & illustration

The living design contract. References go in, recurring design DNA comes out, and every build consults this file. The vaguer the input, the weaker the steer; the more concrete (actual files, exact palettes, "I like X because Y"), the sharper.

How it persists: this doc + the feedback-character-design-taste memory + the flat-artist skill. It compounds — every approve/reject sharpens it.

Reference board

2026-06-24 — first character drop (4 refs)

R1 — 3D clay / toy characters ("ANONYMOUZ", Blender + PNG files) Rounded soft-clay 3D, minimal or featureless faces (sunglasses, simple noses), strongly diverse + inclusive (skin tones, ages, a wheelchair user, body types), everyday activities (camera, gym, skate, gaming, shopping), bright casual clothing. The popular Gumroad/UI8 "3D casual character" look.

R2 — bold flat vector (winter / sports pack) Chunky confident shapes, saturated primaries (red / blue / yellow / navy) on light-blue panels, dynamic athletic poses, minimal faces (often hidden by goggles/helmets), thick forms. Alegria-adjacent but bolder and more saturated.

R3 — modern thick-outline cartoon (sunglasses kid, sunset) Thick black outlines, flat cel fills, warm gradient backgrounds (orange→pink), chill/cool mood, simple expressive features, light environmental detail (wires/poles). Contemporary "cool" cartoon.

R4 — NFT / streetwear bold cartoon (bearded, teal skin, surreal companions) Thick black outlines, hyper-saturated flat fills, surreal/edgy characters with attitude, solid color backgrounds. Doodles / Cool-Cats-adjacent.

Core taste signal (the through-line)

Across all four, consistently:

Bold & saturated — strong, confident color. Not muted, pastel, or washed-out.
Thick outlines (in the 2D refs) — clean, heavy black line.
Minimal / stylized faces — features simplified, abstracted, or hidden (shades, goggles).
Clean confident shapes — rounded, deliberate forms (3D clay or flat).
Attitude & personality — cool, expressive, characterful. Not corporate-bland.
Modern / on-trend — every ref is a current style (3D clay, bold flat, NFT cartoon).
Diverse & inclusive — especially R1.

Anti-patterns (what to AVOID — the look that failed before)

Muted / pastel / desaturated palettes.
Soft, thin, timid line work; vague forms.
Over-detailed, "trying to be realistic" faces → reads amateur ("the 5-year-old").
Corporate flat-design blandness with no personality.

Sourcing strategy (honest path to this quality)

These references are purchasable / artist asset packs, not styles I can reliably hand-redraw at this bar. The reliable path:

Vendor the actual packs the user owns. R1 even ships PNG files → those cutouts drop straight into the 2.5D engine as character layers (fastest win). Honor the license firewall (docs/ASSET_LIBRARY.md, public/illustrations/assets.json).
Build motion / composition / templates around vendored characters — that's our strength. Don't redraw characters from scratch (that's the hand-code ceiling).
For any character I do author, match the through-line: thick outline + saturated fill + minimal face + confident shape.

Feasibility per style (in our engine)

Style	Path
R1 3D clay	Blender for the source (it literally ships `.blend`); PNG cutouts vendor directly into 2.5D today. Three.js can approach soft-clay (rounded geo + soft mat) but the polished look = Blender.
R2 bold flat	Matchable as SVG — thick shapes + saturated fills; best vendored as a pack for consistency/volume.
R3 thick-outline cartoon	Matchable as SVG (heavy stroke + flat cel + gradient bg). Volume → vendor.
R4 NFT bold cartoon	Matchable as SVG; surreal characters are hard to author at scale → vendor.

Status / open questions

Which style is the primary direction (or several as tiers)?
Does the user own the files for these packs (to vendor), or are they inspiration only?
Drop pack files into references/characters/ → I vendor + ledger them.

4.2 KiB Raw Blame History