by SciFigure Pro (LX AI)
SciFigure Pro helps researchers prepare publication-oriented figures — a deterministic quality checklist and caption guidance so you catch dpi, legend, and honesty issues before submission.
Decision-support only — confirm before you publish or act.
When you first open the application you see a blank workspace. Click the “New Project” button in the top toolbar. A dialog asks for a project name and a folder where all assets will be stored. Choose a name that reflects the manuscript you are preparing, for example “SmithEtAl_2025_Fig3”. Press Create.
The workspace splits into three panels: the file explorer on the left, the canvas in the centre, and the inspector on the right. Use the file explorer to drag and drop your raw image files (TIFF, PNG, PDF, or SVG) into the project folder. The explorer shows each file with a thumbnail and its current resolution.
After importing, select a file in the explorer to load it onto the canvas. The canvas displays the figure at 100 % scale. You can zoom with the mouse wheel or pinch gesture on a trackpad. The inspector shows metadata extracted from the file: pixel dimensions, colour space, embedded fonts, and compression settings.
SciFigure Pro runs a fixed set of checks each time you press the “Validate” button. The checks are deterministic, meaning they produce the same result given the same input. This removes variability that can come from manual inspection.
The tool reads the pixel dimensions and the physical size you set in the inspector. It calculates dots per inch (DPI) using the formula DPI = pixel count / size in inches. For most journals the minimum is 300 DPI for line art and 150 DPI for halftone images. If the calculated DPI falls below the threshold the item appears in red with a suggestion to increase pixel count or reduce physical size.
Each legend entry is matched to a graphical element in the figure. The tool checks that every symbol, line style, or colour used in the plot appears exactly once in the legend and that the legend contains no extra items. Mismatches are highlighted and you can edit the legend directly from the inspector.
SciFigure Pro simulates how the figure looks under deuteranopia, protanopia, and tritanopia. It computes a contrast ratio for each pair of adjacent colours. If any ratio falls below 4.5 : 1 the pair is flagged. You can swap colours using the built‑in palette or import a custom colour‑blind‑safe set.
For vector formats (PDF, SVG) the tool scans the file for font names. It checks whether each font is embedded or outlined. If a font is only referenced the validator warns you to embed the font before export. For raster formats this check is skipped.
Based on the target journal’s author guidelines you can set a required format (TIFF with LZW compression, PDF/X‑1a, etc.). The validator compares the actual file properties against the rule set and reports any deviation.
A good caption does three things: describes what is shown, explains how it was produced, and states the key takeaway. SciFigure Pro offers a structured template that you fill in step by step.
Click the “Caption” button in the inspector. A modal window opens with three fields: Description, Methods, and Interpretation. Start with the Description field. Write a single sentence that names the main object, the condition, and the observable outcome. For example “Fluorescence intensity of GFP‑tagged protein X in HeLa cells treated with 10 µM drug Y for 6 h.”
Below each field the tool shows a checklist of items that many journals require: scale bar, staining details, antibody catalogue numbers, exposure time, and statistical test used. As you type, matching keywords are ticked off automatically. Missing items appear in orange and you can add them directly.
Select the target journal from a dropdown list. The tool loads the journal’s caption style guide (character limit, punctuation rules, required abbreviations). It then displays a live character count and warns if you exceed the limit. You can switch journals at any time and the caption reforms to match the new rules.
SciFigure Pro includes an optional language model that proposes caption improvements and highlights potential ambiguities. The model is hosted locally; no data leaves your machine.
When you press the “Suggest” button the current caption text is tokenized and fed to a fine‑tuned transformer. The model has been trained on a corpus of published figure captions from biomedical journals. It returns a list of alternative phrasings with a confidence score for each.
Scores range from 0.0 to 1.0. A score above 0.8 indicates high confidence that the suggestion aligns with typical journal style. Scores between 0.5 and 0.8 are tentative and should be reviewed. Scores below 0.5 are shown for completeness but are unlikely to be useful.
You accept a suggestion by clicking the check‑mark next to it. The caption field updates instantly. If you wish to ignore a suggestion, click the “X” button. The model does not learn from your choices; each session starts from the same baseline weights.
For safety information about the model’s limitations see the OWASP LLM Top 10 document: https://owasp.org/www-project-top-10-for-large-language-model-applications.
Many labs keep figure source files in Git repositories. SciFigure Pro can add a pre‑commit hook that runs the validator automatically.
In the project settings toggle “Enable Git integration”. The tool writes a script to .git/hooks/pre‑commit that calls the command‑line interface (CLI) of SciFigure Pro. The CLI exits with a non‑zero status if any check fails, which aborts the commit.
If you use a continuous‑integration service (GitHub Actions, GitLab CI, or Bitbucket Pipelines) you can add a step that runs the CLI on every push. The step publishes a summary artifact that lists any DPI, legend, or font issues. Reviewers can view the artifact before approving a merge request.
When a commit passes the hook, SciFigure Pro creates a JSON report that records the state of each checklist item. The report is stored alongside the commit in a hidden folder .sciFigure/reports. You can diff reports between commits to see whether a change improved or worsened a specific metric.
For guidance on managing AI risk in automated pipelines consult the NIST AI Risk Management Framework: https://www.nist.gov/itl/ai-risk-management-framework.
It is important to understand the limits of the tool so you do not rely on it for tasks it cannot perform.
SciFigure Pro helps you meet technical requirements, but it does not guarantee that a journal will accept your figure. Editorial decisions depend on scientific novelty, data quality, and many other factors.
The tool checks format and readability. It does not evaluate whether the experimental controls are adequate, whether the statistical analysis is correct, or whether the interpretation is supported by the data.
If the underlying image contains mislabeled bands, incorrect fluorescence channels, or artifacts from acquisition, the validator will not correct those problems. You must return to the original data source and regenerate the figure.
The caption assistance is limited to the figure itself. You still need to compose the surrounding text, discuss results, and cite references.
I keep a written checklist next to every release that touches this topic. The list is boring on purpose: name the owner, name the evidence artifact, and name the rollback. If a step cannot produce a file, a ticket, or a log line, it is not a step — it is a wish.
| Signal | What to record | Stop-ship? |
|---|---|---|
| Unexpected write or send | Tool name, args, identity | Yes until gated |
| Missing confirmation on destructive action | Call path and bypass | Yes |
| Untrusted text treated as instruction | Source surface and resulting action | Yes for agent reach |
| Scanner clean, judgment open | Deferred items + owner | Only if core path safe |
scifigure-pro helps with the measurable first pass. Architecture choices, legal sign-off, and customer-facing claims still need a human owner. Treat every checklist row as decision support, not a certificate.
When a prospect asks whether the product "covers everything," I answer with the boundary: automated checks catch known patterns; novel chains and policy judgment still need review. That honesty keeps support load down and keeps the blog aligned with what the product actually does.
If you need a second opinion, pair this write-up with the references below and with your own staging reproduction. Do not skip the reproduction step just because a model wrote a confident summary.
SciFigure Pro helps researchers prepare publication-oriented figures — a deterministic quality checklist and caption guidance so you catch dpi, legend, and honesty issues before submission.
Decision-support only — confirm before you publish or act.
Want to try the product? SciFigure Pro
Generated 2026-09-17 · primary sources only · no fabricated traffic metrics