# Skill Tree — Learn to Draw

An **open-ended dependency map**, not a schedule. Do a node when its
prerequisites hold; pick your own path. Sessions stay inside ~3 hr/week.

**Goal (top of the tree):** line-dominant drawing with *selective*, form-describing
hatching — on both an observed natural form (from life) and an invented product
(from imagination).

## How to read this
- **Branches:** `[O]` observational (see it) · `[C]` constructive (build it) ·
  `[•]` both / shared.
- **`needs:`** lists prerequisite nodes. Don't start a node until they hold.
- **Priority tags** reflect this learner (ME, CAD-fluent — see `CLAUDE.md`
  §Learner profile): weak channels first.
  - `★ INVEST` — genuine gap, not helped by CAD. Spend real time here.
  - `» FAST` — you already own the concept from CAD; reps are for freehand
    *execution*, not comprehension.
- Each tier carries a **have / lack** line calibrating effort.

---

## T0 · Instrument control  `[•]`  ★ INVEST — root of everything
*Have from CAD: nothing — this is pure motor skill. Lack: all of it. This is the
real beginner zone and the bottleneck for every tier above.*

- **Grip & motion source** — draw from the shoulder/elbow, not the wrist.
- **Straight lines** — ghosted, single confident pass, no re-tracing. `needs: grip`
- **Arcs & curves** — same ghosting, smooth. `needs: grip`
- **Ellipses** — ghosted, drawn *through*, minor axis honest. `needs: arcs`
- **Line-weight control** — vary pressure/lead for a deliberate heavy-vs-light
  line; the foundation of "line-dominant." `needs: straight lines`

**Exit:** confident committed lines and passable freehand ellipses without a
straightedge. → unlocks T1 and T2.

## T1 · Perception (observational)  `[O]`  ★ INVEST — weakest channel
*Have from CAD: little — CAD is 100% constructive. Lack: seeing by eye. This is
the anti-CAD-brain tier: suspend the urge to construct and measure.*

- **Contour (blind → modified)** — draw by the sense of touch, eye on subject.
  `needs: T0 grip`
- **Comparative proportion & measuring** — sight-size ratios, plumb/angle
  by eye (not by ruler). `needs: T0 grip`
- **Negative space** — draw the gaps to correct the object. `needs: contour`
- **Gesture** — capture the whole/movement fast before detail. `needs: T0 grip`

**Exit:** proportions judged by eye land close; a contour reads as observed, not
invented. → feeds T3.

## T2 · Construction (constructive)  `[C]`  » FAST — ride your CAD skill
*Have from CAD: the whole mental model. Lack: only the freehand execution and
line convergence. Frame every node in CAD terms; don't re-learn 3D.*

- **1pt / 2pt perspective** — vanishing points, horizon, convergence by hand.
  `needs: T0 straight lines`
- **The box in space (250-box challenge)** — freehand convergence reps; the
  discipline, not the concept. `needs: perspective`
- **Primitives from any angle** — box, cylinder, sphere in perspective
  (= sketch-plane + extrude/revolve). `needs: box, T0 ellipses`
- **Ellipses-on-forms** — cylinders, wheels, bores; degree shift with viewing
  angle. `needs: T0 ellipses + perspective`
- **Form intersection / boolean** — combining primitives, drawing the seam
  (= union/subtract). `needs: primitives`
- **Symmetry across perspective** — mirror a form convincingly for product
  design. `needs: primitives`

**Exit:** invent a simple multi-part object in correct perspective, freehand,
no ruler. → feeds T3.

## T3 · Surface & structure  `[•]`  » FAST (CAD head start)
*Have from CAD: reading surface curvature and light — zebra, curvature comb,
shaded previews. Lack: putting it down by hand.*

- **Cross-contour lines** — lines wrapping the surface to describe curvature
  (= surface isolines by hand). `needs: T1 contour + T2 primitives`
- **Planar analysis** — break a form into flat planes; the bridge to hatching.
  `needs: T2 primitives`
- **Form shadow vs cast shadow** — locate the terminator, core shadow, and cast
  shadow (a shaded model gives you a head start on *where*). `needs: planar analysis`

**Exit:** show a form's volume with cross-contours and correctly place its core
and cast shadow. → unlocks T4.

## T4 · Hatching / tone (convergence)  `[•]`
*Have from CAD: intuition for where shadow falls. Lack: the manual mark systems
and — crucially — the restraint to hatch* selectively.
*needs: T3 form/shadow + T0 line-weight*

- **Single hatch → cross-hatch** — even, controlled parallel sets.
- **Form-following (curved) hatch** — hatching that rides the cross-contours.
  `needs: T3 cross-contour`
- **Selective placement** — hatch *only* core + cast shadow and accents, not
  everywhere. This is the "selective" in the goal — the hardest restraint.
- **Value via density** — light/mid/dark by spacing, not pressure alone.
- **Texture** — bark, stone, matte vs. gloss metal via mark character.

**Exit:** a form reads as solid and lit from a few well-placed hatch passes, with
most of the paper left as line and white.

## T5 · Integration (the goal)  `[•]`
*needs: T4*

- **Natural form, line + selective hatch, from life** `[O]` — closes the
  observational track.
- **Invented product, constructed → rendered, from imagination** `[C]` —
  closes the constructive track; your stated end goal.
- **Line-weight hierarchy** — heavy contour, lighter interior; the polish that
  makes it "line-dominant." `needs: T0 line-weight`
- **Memory-drawing loop** — draw a studied subject from memory; test retention.

---

## Cross-cutting: CAD-brain failure watchlist
Recurring symptoms to catch in error analysis regardless of tier (see
`CLAUDE.md` Operating principle 9):
- Ruler-straight / mechanically stiff lines → not drawing from the shoulder.
- Tentative or re-traced strokes → not committing; ghost then commit once.
- Over-construction — every drawing buried in scaffolding → trust the eye (T1).
- Drawing as dimensioned projection, not felt gesture → run a gesture/contour
  warm-up (T1) before constructive work.
