Packaging Weaving Drafts Into PDF Patterns
Packaging Weaving Drafts Into Digital Products
Given draft.wif (or a scanned image of a historical draft):
- Parse/transcribe the draft into four grids: threading, tie-up, treadling, drawdown.
- Compute structure metadata: shafts, treadles, repeat width/height, structure name (e.g. "Point Twill," "Summer & Winter").
- Generate a drawdown image (black/white or colored per warp/weft colors).
- Calculate yarn requirements (warp length, weft length, total yardage, sett, width in reed).
- Assemble into a PDF with: cover page, materials list, threading/tie-up/treadling diagram, drawdown, weaving instructions, finishing notes.
- Output
pattern-packet.pdfready for sale (Etsy/Ravelry/etc.).
Never ship just the four grids and call it done — see "Common Pitfalls" below.
Progress:
- [ ] Parse source (WIF fields OR transcribe scanned image into grid data)
- [ ] Validate draft (shaft count matches threading, treadle count matches tie-up, no orphan ends)
- [ ] Identify structure name and historical/technique notes
- [ ] Generate drawdown image (render actual interlacement, not just the grids)
- [ ] Calculate yarn estimates (warp yardage, weft yardage, sett, shrinkage %, loom waste)
- [ ] Write weaving instructions in narrative prose (not just grids)
- [ ] Write finishing instructions (wet finishing, hemming, pressing)
- [ ] Assemble cover page + materials list + size/gauge info
- [ ] Lay out as print-ready PDF (consistent margins, legible grid scale, page breaks that don't split a repeat)
- [ ] Proof: can a weaver with no other context execute this draft from the PDF alone?
Parsing a WIF file
Extract: [THREADING], [TIEUP], [TREADLING], [WARP], [WEFT], [WARP COLORS], [WEFT COLORS], [COLOR TABLE]. If [WARP]/[WEFT] thread counts are missing, infer from threading/treadling length and flag for user confirmation — do not silently assume.
Parsing a scanned historical draft
Historical drafts (especially pre-1900 American/Scandinavian drafts) often use non-standard notation:
- Numbers-in-boxes threading rather than X-marks
- Treadling written as a tied sequence ("1,2,1,2,3,4" rather than a grid)
- Walking/point threading implied by brackets, not drawn out Transcribe conservatively — if a shaft number is ambiguous, note it in a "transcription notes" section rather than guessing silently.
Example 1:
Input: rosepath_4shaft.wif — 4 shafts, 4 treadles, 20-end threading repeat, cotton warp noted in file.
Output: 6-page PDF — cover (title, photo placeholder, skill level: Intermediate), materials (8/2 cotton warp, sett 20 epi, 2 skeins @ 1lb estimated from a 2-yard, 18" wide warp with 10% loom waste + 15% shrinkage), threading/tie-up/treadling diagram scaled to one page without splitting the repeat, rendered drawdown image, narrative weaving instructions ("Thread the rosepath point following the 20-end repeat 9 times for a 24-inch width..."), finishing instructions (wet-finish by machine wash warm, press damp).
Example 2 (comparison/teardown use case): Input: Same WIF file run through a generic AI pattern generator vs. a hand-laid packet from an experienced weaver. Output: Two-column teardown table (see below) showing the agent output stopped at the four raw grids with no yarn math, while the hand-laid packet included sett rationale, loom waste buffer, and a drafted note on substituting 8/2 tencel for the cotton warp.
Use this format when asked to audit or compare an AI-generated pattern packet against a professionally hand-laid one.
| Element | Typical Agent Default | Hand-Laid Professional |
|---|---|---|
| Threading grid | Rendered literally from WIF data, unscaled to page width | Scaled to repeat, labeled with shaft numbers on left, repeat count annotated ("repeat 9x") |
| Tie-up | Included as raw grid | Included + annotated with treadling system (direct tie-up vs. skeleton) |
| Treadling key | Columns = treadle numbers, no footer | Includes a plain-English treadling order to prevent off-by-one errors |
| Drawdown | Often omitted or low-res auto-render with no color accuracy | Rendered at actual warp/weft color, used as a visual proof step |
| Yarn estimates | Frequently omitted, or a generic "2 skeins" with no math shown | Full calc: EPI × width × (length + loom waste) ÷ yards/lb, shown with shrinkage % |
| Sett rationale | Rarely stated | States why this sett for this structure/yarn (balance vs weft-faced) |
| Finishing instructions | Often missing entirely | Always present: wet-finish method, hems, pressing |
| Skill-level framing | Missing | Stated explicitly (beginner/intermediate/advanced) with why |
| Substitution notes | Absent | Common yarn substitutions with resulting sett/yardage changes |
| Page layout | Grids auto-exported, can split mid-repeat across pages | Repeat units kept whole per page, legible grid scale for photocopying |
Diagnosis pattern: AI agents default to treating a draft as "data to transcribe" — they reproduce the four grids faithfully but stop there. Hand-laid packets treat a draft as "instructions for a human with a loom," adding the judgment calls (yarn math, sett rationale, finishing) that only come from having actually woven the structure.
- Always show yarn math, not just a final number — buyers and weavers trust shown work.
- Keep a full repeat unit on one page; never split threading/treadling repeats across a page break.
- State skill level and why (number of shafts, tie-up complexity, pick-up requirements).
- Include a rendered drawdown image, not just the three grids — most buyers visualize color interaction before weaving.
- Flag any transcription ambiguity from scanned sources explicitly rather than silently resolving it.
- Include finishing instructions every time; a pattern without wet-finishing guidance is incomplete for sale.
- Shipping only the four raw grids with no narrative instructions — unusable for anyone outside the software's native format.
- Omitting yarn yardage math or guessing a round number without showing the calculation.
- Silently "fixing" ambiguous shaft numbers from scanned historical drafts instead of flagging them.
- Letting PDF export tools auto-paginate and split a threading repeat mid-page.
- Forgetting to state sett and its rationale, leaving the weaver unable to adapt for substitute yarns.
- Treating the drawdown as optional — it's often the single image that sells the pattern.