Wiki · Autograph / Tables / Data-driven motion graphics

Autograph Tables and CSV data-driven workflow checklist

Published 2026-07-21 · Wiki · Tables / CSV / Workflow

A practical checklist for building repeatable Autograph graphics from tabular data without allowing column changes, locale settings, encoding, missing values, or spreadsheet behavior to silently alter the result.

Why this page exists

Autograph includes a Tables workflow for data-driven graphics, and a July 2026 Creative COW tutorial demonstrates the feature with a downloadable CSV example. The useful production question is not only whether a table can drive a design, but whether the same data produces the same frames after editorial revisions, localization, spreadsheet edits, machine changes, and final render.

CSV is widely exchanged, but it does not carry reliable column types, units, uniqueness rules, or semantic meaning by itself. Treat the data file as a versioned production input rather than an informal spreadsheet attachment.

Define the table contract before design

Give every column a stable, machine-oriented name. Prefer shot_id, display_name, and value_percent over headings that are likely to be renamed for presentation.
Document the intended type of each column: text, integer, decimal, percentage, Boolean, date, duration, color, asset path, or identifier.
Document units separately. A value of 24 is ambiguous unless the project states frames, seconds, percent, pixels, or another unit.
Choose a stable row identifier that does not change when display text or sorting changes.
Define whether row order is meaningful. If it is, include an explicit order column rather than trusting spreadsheet position.
Define the meaning of an empty cell. Empty, zero, false, unavailable, and intentionally hidden are different states.

CSV preparation checklist

Export a plain CSV rather than relying on an application-specific spreadsheet file.
Keep one header row and the same number of fields in every record.
Quote fields that contain commas, line breaks, or double quotes. Escape an internal double quote with a second double quote.
Use a declared text encoding and test non-ASCII names, accents, Japanese text, symbols, and emoji if the project requires them.
Test both the presence and absence of a UTF-8 byte-order mark when the producer and receiving tools disagree. A UTF-8 BOM is an encoding signature, not a byte-order requirement, and some consumers do not expect it.
Avoid locale-dependent number formatting. Use a period as the decimal separator in the interchange file unless the complete pipeline has been tested with another convention.
Store dates in an unambiguous form such as YYYY-MM-DD when dates are data rather than display copy.
Do not mix raw numeric values with presentation strings such as 42%, $1,200, or 12 kg in the same column. Keep raw values and formatted labels separate when possible.

Autograph setup and validation

Import a minimal test table first: one normal row, one zero value, one empty value, one long text value, one non-ASCII value, and one deliberately invalid row.
Map design properties to stable column names, not to temporary spreadsheet order.
Confirm how Autograph handles missing columns, duplicate headers, duplicate identifiers, blank rows, and extra columns before the project depends on that behavior.
Test the first, middle, and last rows and any row that produces extreme layout conditions.
Confirm whether changing the CSV updates the composition automatically or requires a deliberate reload. Make that refresh step part of the render checklist.
After a data refresh, verify text, numbers, colors, visibility rules, asset references, animation timing, and responsive layout rather than checking only one visible field.
Render a small representative range and re-open it outside Autograph. A correct interactive preview is not sufficient evidence of a correct batch or final render.

Design for unstable content

Set tested minimum and maximum text lengths for every visible field.
Decide whether long text wraps, scales, truncates, scrolls, or fails validation. Do not let the host choose accidentally.
Test negative values, zero, large values, decimals, and values beyond the expected chart range.
Keep numeric geometry separate from formatted display text so localization does not change animation math.
Use explicit fallback assets and fallback text for missing references.
Do not use row number as identity when rows may be sorted, filtered, inserted, or deleted.

Versioning and handoff

Version the Autograph project, the CSV, and any referenced assets together.
Record a checksum or archive copy of the exact CSV used for an approved render.
Include the data schema or a short data dictionary with column names, types, units, allowed values, and fallback behavior.
Record the Autograph version and whether the table was refreshed immediately before rendering.
For recurring output, keep a known-good validation CSV that can be loaded before a new delivery batch.
When a column is renamed or removed, treat it as a pipeline change and re-test every dependent property.

Spreadsheet and security caution

CSV files supplied by clients, web forms, or external systems may contain cells beginning with characters that spreadsheet applications interpret as formulas. OWASP describes this as CSV or formula injection. This is primarily a risk when the file is opened in spreadsheet software, but it matters in a motion-design pipeline because artists often inspect or revise CSV files in spreadsheets before importing them.

Do not assume that quoting a field makes spreadsheet formula execution harmless.
Treat untrusted CSV as data and review it before opening it in a spreadsheet application.
Do not apply spreadsheet-oriented sanitization blindly if the file must also remain machine-readable; OWASP notes that there is no universal strategy that is safe for every spreadsheet and every downstream consumer.
Keep the original unmodified source, the reviewed production CSV, and the render-approved CSV as separate files.

Common failure patterns

Values shift into the wrong columns: an unquoted comma, line break, or quote changed the field structure.
Japanese or accented text is corrupted: the producer and reader disagree about encoding or BOM handling.
Numbers animate incorrectly after localization: formatted strings or comma decimal separators were used as numeric inputs.
A later revision changes approved graphics: row position was used as identity, or the CSV was refreshed without archiving the approved version.
Missing data becomes zero: the pipeline has no explicit distinction between blank, unavailable, and numeric zero.
Text overlaps or leaves frame: the design was tested with sample copy rather than boundary-length content.
The viewer looks correct but final output differs: cached data, unresolved asset paths, or a missing pre-render refresh changed the render state.

Minimum acceptance test

Load the known-good validation CSV.
Confirm every mapped field and fallback state.
Replace it with the production CSV and perform a deliberate refresh.
Check row count, identifiers, missing values, long text, non-ASCII text, extremes, and referenced assets.
Render representative frames for at least one ordinary row and every exception class.
Archive the project, exact CSV, assets, render preset, application version, and approval output together.

Practical rule

Keep creative formatting in Autograph and keep source values in a stable, documented table. A data-driven design is production-safe only when the table schema, refresh behavior, edge cases, and approved input file are reproducible.

Useful next pages

Sources checked

Creative COW: Maxon Autograph — Introduction to working with Tables — published 2026-07-01; checked 2026-07-21. Used to confirm the Autograph Tables workflow and downloadable CSV tutorial context.
RFC 4180: Common Format and MIME Type for Comma-Separated Values Files — checked 2026-07-21. Used for the common header, record, quoting, comma, line-break, and escaped-quote conventions.
W3C CSV on the Web: A Primer — checked 2026-07-21. Used for the limitation that CSV alone does not define column types, uniqueness, or validation rules, and for the value of separate metadata.
Unicode BOM FAQ — checked 2026-07-21. Used for UTF-8 BOM behavior and interoperability cautions.
OWASP CSV Injection — checked 2026-07-21. Used for the spreadsheet formula-injection caution and the lack of one universal sanitization method for all consumers.