Export a reproducibility package

Create a local ZIP containing a readable methods/evidence report and a portable workflow with its explanatory notes. Review the exact contents before sharing it with a collaborator or including it in an analysis archive.

New in 0.15.0a3

This export requires VIPP 0.15.0a3 or newer. It does not calculate the graph, upload anything, or include raw images, output images, tables, meshes, thumbnails or previews.

Choose the right evidence

Start here	What the package describes
Main toolbar gear → Workflow actions → Export reproducibility package…	The current workflow recipe, captured when the review window opens. This is not evidence that the graph was run or its outputs verified.
Batch workflow → Run & results → run report card → Export reproducibility package… (also Export package… in the finished-run footer)	The recorded run's archived workflow, settings and available item evidence—not the graph as edited afterward. The action is enabled after a recorded run, while no run is active.

Both batch actions open the same review window; neither immediately saves a ZIP. The package keeps the workflow's explanations, settings, software versions and run records, but not images, intermediate results or saved outputs. The results toolbar keeps run-recovery and file-status actions, not package export.

For batch work, keep failed, partial, skipped and cancelled outcomes visible; the package does not turn them into successful results. If the archived evidence is incomplete or incompatible, preparation fails rather than substituting the current graph. Keep the original manifest archives and sidecars separately.

Prepare, review and export

1. Open the export action for the evidence you want. Initial preparation runs in the background, leaving the interface responsive.
2. Review the Report title and optional Notes. Folder locations are hidden automatically. Select Anonymise filenames if filenames should also be replaced; it is off by default.
3. After changing any of those fields, select Prepare report again. The old preview and sharing confirmation are invalidated; export waits for the new report.
4. Read Report, inspect the file inventory and exact text in Package contents, and check Privacy & omissions for removed information, substitutions and limitations.
5. Check I reviewed the report and included files for sharing. Then select Export package… and choose a new .zip filename. Existing files are not overwritten. The ZIP contains the prepared bytes you reviewed; export does not regenerate the report from later graph or file changes.

Closing during preparation discards the pending preview. Once ZIP writing has started, Close becomes available when writing finishes. A cancelled destination dialog writes nothing.

In the preview, an official release or setup-guide link opens your browser only when you select it. The workflow.json link explains how to export, extract and open the workflow in VIPP; it does not start an analysis. The preview does not fetch remote images or other resources.

Repeat the analysis in VIPP

1. Extract the ZIP and open report.html. Start with How to repeat the analysis; its workflow.json link refers to the file beside the report.
2. Review the recorded VIPP version and follow the report's official release and installation-guide links, when available. Run versions and the version that exported the package are separate records. A development or locally modified build may have no matching public installer; ask the author for that build's setup information rather than assuming the latest release is equivalent.
3. In VIPP, choose Open workflow and select the extracted workflow.json. For a recorded workflow with verification information, choose Reproduce original run or Use workflow on new data when prompted, review the version status, then select Continue to batch setup. If the package contains batch settings, that file carries them together with the graph and restores Batch workflow → Setup. Review any restoration warning before continuing. Opening the workflow or viewing an Image Source does not change its saved settings; input verification starts when you select Check batch.
4. In Setup, choose the intended input folder for each collection source and a new output folder. Select any fixed reference images separately in their Image Source inspectors; the restored batch settings are retained. Review filename patterns, source pairing, image axes, calibration, overrides and output settings.
5. Select Check batch, review Items & outputs and Overrides, then select Run when the checked plan is correct. Checking does not calculate images or save processing outputs. Compare the new results with the separately supplied reference results and validation evidence.

For a graph-only recipe, reconnect each Image Source in the inspector and review any save destinations before calculating. The package does not include the images or results you need for either route. Ordinary GUI reuse does not require editing JSON or copying workflow/configuration hashes.

Reproduce or use new data

• Reproduce original run: use the original inputs. Check batch compares them with the recorded hashes and image selections. The summary shows how many need attention; the Checks column identifies affected items. Changed, missing or unexpected inputs block Run until resolved.
• Use workflow on new data: no match to the original inputs is required. VIPP keeps its normal input checks. Saving this reused workflow does not keep the original reproduction requirement; export a new recorded batch to create a new reference.

If you checked the wrong folder, choose the original inputs in Setup and select Check batch again. A successful check replaces the earlier mismatch result. Run checks for subsequent changes before processing; no hash exception is needed for corrected, matching inputs.

The compact banner keeps its actions to the right of the status text. Select Check details… for the input result, matched/needs-attention totals, any issues and the VIPP version status, including an accepted version difference. Review batch inputs opens all rows in Items & outputs; Review affected inputs shows only items needing attention and clears any search. For a version or setup problem, Review batch setup returns to Setup. These buttons only change views: they do not recheck files, approve exceptions or start Run.

The opening dialog groups the two choices and shows whether the original and installed VIPP versions match. For reproduction, a different or unknown version needs explicit acknowledgement. The nearby release and installation links help you obtain the original version; they open a web page, not an automatic download or installation. An accepted difference stays visible and is recorded; input checks still apply. For new data, the version notice does not block the analysis.

After a fully successful reproduction with matching VIPP versions, the green Run complete · Original inputs verified banner describes the completed run. It does not authorise another run: select Check batch again before running again. Changing input folders or batch settings clears the completed-run badge. Cancelled, partial or failed runs, and runs with an accepted version difference, do not receive this green completion state.

Matching hashes confirm the same input files, not identical analysis results. Software dependencies, hardware and settings also matter. Older packages, and runs without sufficient recorded input information, may lack this verification; export a fresh package from the original recorded run where available.

Check the newly discovered samples, including series within multi-image files. The portable settings do not reuse the old frozen input inventory as proof about files on your computer. Scientific overrides remain tied to their source content and logical item; changed data or selectors need review. Individual overwrite permissions from the original run are not transferred. Review any existing-output decisions afresh.

Fixed reference images need individual Image Source relinking; they are not silently converted into folder collections. Missing fixed-reference files do not discard the restored Setup, but Check batch still requires all sources to be valid before Run. Live-layer and sample inputs also need their intended data supplied separately. If VIPP cannot restore a source or its Batch Setup, stop and use the author's source-selection/setup notes to review the recipe before running. Do not substitute another series or image merely to make the check pass.

The ZIP contains no installer and does not set up an environment automatically. Use the official software links deliberately; opening or exporting the package does not install software. A matching VIPP version alone does not establish matching readers, dependencies, hardware or results.

Python is an advanced alternative

Use the supplied runners only when you need headless execution. README.md describes their setup and limitations; their --help output lists the exact arguments. For a recorded batch, use batch-runner.py with its batch configuration so batch overrides are retained. runner.py embeds the ordinary graph as an independent copy without those batch overrides. The Python route is not required to repeat the analysis through VIPP.

What is included

The inventory is specific to the selected recipe or run. Core files include:

• report.html: the self-contained report, readable offline after extracting the ZIP;
• workflow.json: the portable workflow and its explanatory notes, including portable Batch Setup when batch settings are available;
• README.md: GUI-first repeat steps, scope and limitations;
• runner.py: the generated VIPP runner for advanced headless use;
• environment.json: available run/export version summaries;
• report.json: structured report details, omissions and changes;
• SHA256SUMS.json: exact-byte checksums for the other package members.

Packages with batch settings also include batch-config.json and batch-runner.py. The standalone settings match the attachment in workflow.json; they are provided for advanced use, not a required separate GUI import. A recorded batch package additionally contains sanitized evidence views, which are not the original run receipts. Recorded source/output identities are reported without reopening the image or result files to verify them again.

The version summary is not an environment lockfile, installer or bundled Python environment. Checksums identify exported bytes; they do not authenticate an author or establish scientific correctness.

Review privacy and reuse limits

Folder removal and filename anonymisation do not guarantee anonymity. Scientific labels, node titles and explanatory notes may still identify people or samples. Workflow notes are kept, with the same folder hiding and optional filename anonymisation as other shared text. Read those notes and inspect the included files before sharing. Nothing is sent to an external service by preparation or export.

The package is not resumable run state, a complete data archive, or a promise of identical output on another computer. Follow the repeat steps above and validate the result for your data. Keep the original evidence for verified batch resume.

For a publication archive, add the separately reviewed data, results, validation and manual-decision context required by the reporting checklist.