Tutorial

Five Export Formats, One BPMN: How to Pick the Right Artefact for Every Audience

Every process diagram eventually leaves the platform. The question is which format. PNG for slides, SVG for designers, PDF for executives, BPMN 2.0 XML for workflow engines, JSON for custom tooling. Picking wrong means the downstream consumer cannot do what they need, which is usually worse than not exporting at all.

April 14, 20266 min read

Which format for which audience

Every BPMN eventually leaves the platform. Sometimes it lands in a PowerPoint deck, sometimes in a compliance binder, sometimes in a workflow engine, sometimes in a custom analytics pipeline. The five export formats LucidFlow produces are not interchangeable: each is engineered for a specific downstream consumer, and the mistake analysts make is exporting the format they know rather than the format the consumer needs.

PNG: raster image, widely supported, lowest-friction for inclusion in slides and documents. The right format for 80 percent of non-technical sharing: email attachments, PowerPoint slides, Confluence pages, Word documents. Weakness: not infinitely scalable, looks pixelated on high-DPI displays if exported at low resolution.
SVG: vector image, infinitely scalable, editable in Illustrator or Figma. The right format for designers who will re-style the diagram for brand consistency, and for print production where resolution independence matters. Weakness: less universally supported in business tooling than PNG; some older versions of Word and PowerPoint handle SVG imperfectly.
PDF: multi-page document with embedded diagram, metadata, and legend. The right format for executive readouts, compliance archives, and any consumer who wants a self-contained artefact that preserves formatting across any PDF reader. Weakness: not editable, and the embedded diagram is raster rather than vector in most renderers.
BPMN 2.0 XML: the canonical, executable format. The right format for importing into Camunda, Flowable, jBPM, or any other workflow engine that supports the BPMN standard. Also the right format for interoperability with other BPMN tools: Visio, Bizagi, Signavio can all read BPMN 2.0 XML. Weakness: not human-readable; you cannot meaningfully open it without a BPMN viewer.
JSON: the raw data of the diagram, with every node, edge, KPI, and metadata field exposed. The right format for custom analytics pipelines, integration with internal reporting tools, and ETL into business intelligence platforms (Power BI, Tableau, Looker). Weakness: requires the consumer to write code to do anything with it: this is not a format you send to a VP.

What each format preserves, and what it loses

An important detail most users miss: different export formats preserve different subsets of the diagram's data. PNG and SVG are visual formats, so they preserve what the canvas shows: nodes, edges, labels, colours, swimlanes, heatmap overlay if active. They do not preserve the KPI data behind the visuals; a PNG of the cost dashboard has the numbers as pixels but not as data.

PDF sits in the middle: it embeds the rendered diagram as a PNG inside a single-page jspdf document: there is no metadata layer, just the rasterised picture. BPMN 2.0 XML preserves only the canonical structural elements (events, tasks, gateways, sequence flows, pools, lanes). It does NOT include the per-task KPI fields (estimatedDuration, estimatedCost, frequency), the heatmap scoring, or any other LucidFlow-specific data: those are stripped from the XML output entirely. If you export a process to BPMN XML, send it to a workflow engine, then re-import it into LucidFlow, the structure round-trips but every cost and frequency value comes back blank. JSON is the only export format that preserves the full payload: structure, KPIs, metadata, the lot.

The hidden details that matter more than most users realise

Four details of the export pipeline affect output quality and are worth understanding even if you never manually adjust them.

Viewport fitting

The export pipeline runs a pass called 'withFittedViewport' that recalculates the visible area of the canvas so the exported image shows the whole diagram without empty white space and without cropping. This matters because the on-screen view you see is usually zoomed-in on a specific part of the process, but the export is supposed to show the whole thing. If you have ever exported a diagram from an older tool and found that half of it was cut off, viewport fitting is the feature that prevents that specific failure.

Legend rendering

If the diagram is currently showing a heatmap (Cost, Duration, or Impact mode), the export includes a legend explaining the colour scale. The legend is rendered directly into the exported image rather than drawn as a separate layer, so it survives the export process and is visible to anyone who opens the file later. Without this, a recipient of a heatmap PNG would see coloured nodes with no explanation of what the colours mean, which is worse than exporting the plain view.

Watermark application

Every PNG, SVG, and PDF export carries a 'LucidFlow.ai' brand mark in the bottom-right corner of the rendered diagram. The watermark is applied unconditionally: there is no tier-based check that toggles it on or off, and the shared export utilities do not take a plan parameter. Teams that specifically need unbranded exports for embedding in customer-facing artefacts should reach out: that's a feature request, not a tier upgrade. The BPMN 2.0 XML and JSON exports do not include the visual watermark since they're not images.

PDF is dynamically loaded

The PDF exporter uses jspdf and fflate libraries that are too large to ship in the main bundle, so they are loaded dynamically only when the user actually clicks Export PDF. The first PDF export of a session may have a slight delay (300 to 800 milliseconds) while the libraries load; subsequent PDF exports are instant. This is invisible to users but worth knowing because it is a common question from devs who inspect the bundle.

Frequently asked questions

Why export BPMN 2.0 XML if I am not using a workflow engine?

Three reasons even non-engine users find it useful. First, portability: a BPMN 2.0 XML file can be opened in Visio, Bizagi, Signavio, and most other BPMN tools, which matters if your organisation uses multiple process tools and wants the diagrams to be tool-independent. Second, future-proofing: if you eventually move to a workflow engine, the XML is already ready without rework. Third, interoperability with AI analysis: other AI-native platforms that accept BPMN 2.0 XML as input can ingest your LucidFlow diagrams directly. For pure documentation use where none of these apply, BPMN XML is overkill; PNG or PDF is sufficient.

What resolution does the PNG export use?

PNG and PDF both render at 2x pixel ratio: twice the on-screen canvas resolution, which produces crisp results on standard and retina displays and is adequate for slides, documents, and most print. The pixel ratio is hard-coded in the shared export utilities; there is no UI or settings panel to bump it to 3x or 4x. For use cases that need arbitrarily high resolution (large-format print, detailed inspection at extreme zoom), SVG is the right format, it is resolution-independent by construction, so it will render cleanly at any size without the raster trade-off.

Is the exported BPMN 2.0 XML genuinely standards-compliant?

Yes for the canonical structure: events, tasks (mapped to userTask, serviceTask, scriptTask, sendTask, receiveTask, manualTask, businessRuleTask, subProcess depending on the task type), gateways (exclusive, parallel, inclusive, eventBased, complex), sequence flows, intermediate events, data objects, and text annotations all serialise into standards-compliant tags. What is NOT in the XML: LucidFlow's per-task KPI fields (estimatedDuration, estimatedCost, frequency), the heatmap scoring, the ESSII analysis results: none of these are wrapped in `bpmn:extensionElements`; they are simply not emitted. The exporter intentionally produces a clean, portable BPMN file that any standards-compliant parser will accept, at the cost of losing LucidFlow's analytical metadata. Use the JSON export if you need to preserve that data alongside the structure.

Why does the JSON export include fields I do not recognise?

The JSON is the raw internal representation of the diagram, which includes fields that are only meaningful to LucidFlow's own rendering and analysis engines: layout hints, heatmap cache data, collaboration state references, and a few internal flags. These fields are stable (they do not change meaning across platform versions) but they are not documented for external consumers because they are not part of a supported API. For stable integration, pick the specific fields you need (nodes, edges, KPIs, metadata) and ignore the rest. The LucidFlow documentation includes a JSON schema specifying which fields are considered public and which are internal.

Can I batch-export all processes in my workspace at once?

Not today. The export module runs entirely client-side: each button click triggers a browser anchor click and dispatches a single download, so there is no batch-export action and no REST export endpoint to script against. If you need to archive every process, the practical workflow is to open each session and export it individually. Bulk export (including a ZIP bundle from the dashboard) is a recognised feature request but has not shipped as of April 2026.

Document to BPMN: How AI Reads a Transcript and Produces a Compliant Diagram in Under Two Minutes Swimlane Diagrams: How the Layout Makes Accountability Visible Before Anyone Reads the Text From Meeting Transcript to BPMN: A Worked Example of the Document-to-Diagram Flow

Ready to Build Your AI Transformation Plan?

Upload any process document and co-build an AI transformation plan with real tool recommendations and ROI projections — in minutes, not weeks.

Try LucidFlow Free