# Good Sew digitizing guidance

Revision: 2026-09-07-sewout-1

Comparison of a 50.4 × 44.6 mm black line portrait with 19.6 × 25.6 mm pinecone and 23.4 × 25.6 mm maple PES files and their sew-outs. The owner confirmed the same fabric, stabilizer, needle and thread weight/brand; thread colors differed. This is a small observed comparison, not a universal material calibration.

Read get_design before editing and follow its digitizingGuidance. Preserve approved likeness and work in final-size millimeters. Scaling artwork requires re-digitizing and inspecting features; scaling finished PES stitches is not equivalent.

Choose a stitch structure for each feature: directional satin bands for bold curves, coherent tatami for broad areas, running lines for fine accents. Give important outlines width in the geometry; repeated passes along a hairline do not create satin sheen.

For small portraits on textured woven fabric, remove tiny sawtooth fur turns while preserving ears, eye positions, muzzle and expression. Use fewer readable tufts, a clear outline/detail line-weight hierarchy, and deliberate negative spaces. Do not trace every raster edge or convert every line indiscriminately to satin.

Treat dimensions as trial starting points for ordinary 40-weight thread: important satin bands around 1–1.5 mm or wider where the artwork allows; finer lines as single/double runs; paired-rail cycle spacing about 0.4–0.45 mm; gentle running stitches about 1.8–2.5 mm, shortened only where curvature needs it. Narrower columns and tiny highlights need a sew-out. These are not machine limits or guaranteed material settings.

Good Sew paired-rail density is full-cycle spacing (left → right → left). Nodes control direction, not density. Distribute direction changes smoothly, avoid crossed rails, spread inside-corner punctures, and split abrupt corners or overly wide columns. The 10 mm throw cap is an application limit. Legacy center-spine satin uses a different spacing convention; prefer paired rails for precise work.

For running lines, params.returnPass=true makes a whole-path double run; repeats=3 makes a bean stitch with a reversal at every segment. Short, heavily retraced lines can appear knotted or beaded. Use params.curveTolerance=0.15 as an optional starting point to follow curves; lower tolerance adds stitches. Simplify unsuitable tiny geometry rather than demanding ever shorter stitches. Zero preserves legacy sampling.

Plan support and order: modest center-walk underlay for narrow satin; broad fill before finishing detail; minimize redundant fill layers and stitch only necessary junction overlap. Fill params.underlayStyle="edge" selects light edge-only support instead of an extra sparse fill layer. Avoid converging a whole eye or nose into a tiny radial hole; use directional sections and preserve intentional highlights. Tatami spacing is the gap between single rows, unlike paired-rail full-cycle spacing. Underlay, fill spacing and pull compensation must be tuned from actual fabric results.

Reduce disconnected islands and exposed travel when artwork permits. Never stitch across open fabric just to avoid a jump. Retain tie-ins/offs at real starts and cuts. SE700 trim commands do not guarantee that internal jump threads are cut automatically.

Use generate_stitches, its object-specific digitizingReview, inspect_region, and fixed top/angled previews. Judge actual-size readability as well as magnified stitch geometry. A high local needle count or many short stitches is an inspection cue, not automatic failure: the successful pinecone had more short stitches and a higher local count than the portrait. Do not globally thin a design or increase density to chase a warning or a stitch-count target.

Export, decode the actual PES, verify dimensions, origin, color blocks, travel and full stitch extent. Keep the corrected PES origin metadata. A pretty vector picture or 3D render is not verification of sewability. Thread Lab estimates appearance; it does not simulate thread tension, fabric deformation, bobbin pickup or machine dynamics.

For each physical test, record machine, final size, fabric, stabilizer, needle, thread weight/brand, speed and settings, then compare front/back photos and change one class of variables at a time. Separate observed faults from inferred causes. Pale flecks or loops in a photo do not prove a tension diagnosis; do not prescribe a bobbin-tension adjustment from that alone. Do not claim a revision is physically validated until it has been sewn.

Sources:
- Ink/Stitch — satin columns, spacing, compensation and underlay: https://inkstitch.org/docs/stitches/satin-column/
- Ink/Stitch — running/bean stitches and curve tolerance: https://inkstitch.org/docs/stitches/running-stitch/
- Brother SE700 operation manual: https://download.brother.com/welcome/doch102582/888p10_om01en.pdf