# Tutorials > Step-by-step workflows with practice inputs. Illustrated version with screenshots: https://figeditor.com/tutorials ## Text → a scientific figure Practise with a deliberately small mechanism: three stages and a limited label set. A precise brief is easier to verify than “draw a professional figure”. Open: https://figeditor.com/generate-figure Practice input: Create a left-to-right conceptual diagram: ligand binds a cell-surface receptor → intracellular signalling → gene expression in a nucleus. Use three separated panels and arrows between stages. Labels: Ligand, Receptor, Signal, Nucleus. Flat shapes, white background, blue and orange accents. Do not add quantitative data or claim a specific molecular pathway. 1. **Write a checkable brief** — Open Generate figure and paste the practice input. Replace the subject only after this small example works. State the direction of arrows and the exact labels. 2. **Set the output deliberately** — Choose a landscape aspect ratio for the three stages. Review the selected model, requested resolution and the credit amount shown beside the generation control. Sign in with Google when prompted. 3. **Generate once, then inspect** — Submit only when the brief is ready. Open the resulting project and compare every arrow, compartment and label with your brief. Treat plausible-looking biology as unverified until you check it. 4. **Ask for one correction at a time** — Use the project chat to describe a concrete correction, such as moving the nucleus into the right panel while keeping the other two panels unchanged. Check the displayed cost before sending another AI request. Expected result: A saved project with an image you can reopen, compare and export. The model may still misspell a label or draw an incorrect arrow. ## Sketch → a clean figure draft Use your own simple layout sketch to explain spatial relationships. The upload is a reference for generation, not a promise that every mark becomes an editable vector. Open: https://figeditor.com/sketch-to-figure Practice input: Use the attached sketch as a layout reference. Keep three boxes in one row and two arrows pointing right. Replace the handwritten labels with Sample preparation, Measurement, Analysis. Use a clean white background and consistent line weight. Do not invent instruments or results. 1. **Prepare a readable sketch** — Draw three boxes and two arrows on plain paper. Photograph from directly above in even light; crop away the desk and personal notes. 2. **Attach the image** — Open the sketch workflow and use the image attachment control. Wait until the reference preview appears before submitting. The tutorial screenshot is an interface example, not a required input file. 3. **Explain what must stay** — Paste the practice input, choose the output ratio and review the shown cost. Name each box and specify which arrows must keep their direction. 4. **Compare before exporting** — Compare the generated draft with the original sketch side by side. Check ordering, connections and labels; ask for a targeted correction if the model changed a relationship. Expected result: A polished draft that follows the intended layout. Use the original sketch as a checklist for relationships, not as evidence of scientific correctness. ## Reference → your own visual composition Learn from layout and visual hierarchy while supplying your own content. Use a reference you own or have permission to upload. Open: https://figeditor.com/img-to-figure Practice input: Use the reference only for its two-column composition and restrained blue palette. My topic is a comparison of workflow A and workflow B. Each column has Input → Process → Output. Use those six labels only. Do not copy the reference logo, title, experimental data or decorative artwork. 1. **Choose a usable reference** — Use your own diagram or an appropriately licensed image. If starting from the inspiration gallery, inspect the example before choosing Use this; review any attached reference in the workspace. 2. **Separate style from facts** — Attach the reference and paste the practice input. Explicitly say which layout features to borrow and which content must be replaced. 3. **Review the generated interpretation** — After generation, check whether both workflows have the correct three stages. Remove copied logos, unintended captions and invented results through a targeted revision. 4. **Keep a provenance record** — Save the reference source, licence and your brief outside the project. A generated image does not automatically remove rights in uploaded material. Expected result: A new draft using your own labels and relationships, with the reference serving only the stated visual purpose. ## Paper → a reviewed academic poster Prepare a short structured summary before using a full paper. This makes missing sections and invented numbers easier to catch. Open: https://figeditor.com/scientific-poster Practice input: Title: A pilot workflow for water-sample preparation Background: Compare two preparation workflows. Methods: Filter samples, prepare aliquots and record processing time. Results: This is a design example; no experimental results are supplied. Conclusion: Show the planned comparison, without claiming improved performance. References: Add only references present in the source paper. 1. **Provide selectable text** — Paste the practice summary into the poster text area, or upload a supported PDF/DOCX. Check the extracted text; a scanned page may need OCR outside this workflow first. 2. **Open the template preview** — Click a template card to inspect its full preview. Choose a layout that fits the number of sections and the reading direction; close the preview to return to the form. 3. **Generate and inspect the actual output** — Review settings and displayed cost before generating. Read the result’s actual pixel dimensions and compare the poster title, methods, results and references against the source. 4. **Revise factual problems first** — Fix invented claims and missing sections before refining colours. If a source includes numbers, verify units, denominators and figure captions individually. Expected result: A poster draft organised by the selected layout. The practice input should not be interpreted as a real experiment or a source of publishable results. ## Bitmap → SVG → a checked edit Start with a clean diagram or logo, not a noisy photograph. Tracing creates paths; editable text is a separate OCR option and still needs proofreading. Open: https://figeditor.com/svg-converter Practice input: Practice file: your own PNG with three solid-colour circles, two arrows and the labels Input, Process, Output. Use a white background and clear, dark text. 1. **Upload and choose an engine** — Open SVG converter and choose a supported bitmap under the displayed size limit. Try Fast first for flat artwork; High precision downloads a local model on first use. 2. **Decide whether labels need OCR** — Enable Keep text editable when you need actual text objects. Without that option, letters may remain traced shapes rather than text you can type over. 3. **Convert and compare** — Start conversion and keep the tab open. Move the comparison slider to inspect the original and SVG, especially thin arrows, small labels and white areas. 4. **Edit, download and reopen** — Download SVG or send it to the SVG editor. Select a path to adjust its fill or stroke. For recognised text, verify spelling before changing it. Reopen the downloaded SVG in the app that will receive it. Expected result: An SVG with vector paths; OCR-enabled results may also include text objects. Conversion history can restore recent local results on this browser. ## Panels → a journal-sized figure Assemble already checked panels. Layout controls spacing and physical dimensions; it cannot restore detail missing from a small source image. Open: https://figeditor.com/figure-layout Practice input: Practice set: four of your own panels named A–D. Choose a 2×2 grid, Nature double-column width (183 mm), uppercase panel labels and 300 DPI. 1. **Choose the physical size and grid** — Open Figure layout, select a 2×2 grid and set the journal width. The width is a starting preset; always confirm your target journal’s current instructions separately. 2. **Place panels in reading order** — Click cells to upload or drag images onto the canvas. Check each placement; excess images can appear in the Not placed tray. Drag panels to swap their positions. 3. **Inspect resolution and labels** — Set spacing and A/B/C labels, then inspect each panel’s DPI badge. A low-DPI warning means the source has too few pixels for its placed size; replace the original or reduce the panel size. 4. **Export and inspect the file** — Choose PNG, TIFF, PDF or SVG and the required DPI. Open the file after downloading: PDF page size should match the physical width; SVG embeds raster panels rather than vectorising them. Expected result: A figure with aligned panels and consistent labels. At 183 mm and 300 DPI, the full width is approximately 2161 pixels; rounding may vary by the exporter. ## Plate photo → a corrected count Start with the built-in example, then practise on your own plate. Automated detections are suggestions; the final count depends on your review. Open: https://figeditor.com/colony-counter Practice input: Practice: load Try an example. Inspect the plate edge, glare and touching colonies. Add one deliberately missed marker, remove a false marker, and compare the total before and after. 1. **Load the free prepared example** — Choose Try an example in the upload area. This prepared example does not submit a fresh paid AI request. For your own photo, use a supported format and keep the full plate in frame. 2. **Choose the analysis route deliberately** — For a new image, distinguish the rough offline pass from AI analysis. AI analysis can require sign-in and credits; review the current control and cost before running it. 3. **Correct detections manually** — Inspect marks around the rim and bright reflections. Use the displayed add/remove controls to correct missed and false marks. Check ×2 or ×3 marks on touching colonies instead of treating them as one colony automatically. 4. **Export an auditable image** — Download the marked PNG after reviewing the total. Keep the unmarked source photo, corrected export and experimental notes together. Counting alone does not supply a dilution factor or CFU/mL calculation. Expected result: A plate image with visible markers and a total that changes when you correct them. This is a reviewed image count, not a clinical or microbiological conclusion. ## Credits, saved versions and publication checks Understand what you are paying for and what an export records. Use the displayed price and account history rather than an old fixed credit figure. Open: https://figeditor.com/pricing Practice input: Practice checklist: note your balance → read the displayed operation cost → run one intended task → check the result and credit history → export the saved version. Do not buy a plan just to follow this tutorial. 1. **Read the current operation cost** — Generation, AI revisions and AI analysis can have different costs. Check the selected operation and provider state. Browser-only tools and the prepared colony example are different from paid AI calls. 2. **Check history after the result** — After a completed or failed request, inspect your balance and credit history. If the UI reports a failure, check the recorded outcome before retrying to avoid duplicate requests. 3. **Export the version you reviewed** — Open the project export controls and choose the required file format. Inspect the actual pixel size. A format change, larger canvas or SVG container does not automatically add real image detail. 4. **Read the publication use statement** — Where available in the project export dialog, download Publication use statement (PDF). It identifies the stored image version and hash under FigEditor terms. It is an unsigned terms record, not copyright registration, exclusive ownership or a grant of third-party rights. Expected result: An understandable transaction record and an exported file tied to the version you checked. Current prices and entitlements are shown on the pricing/account pages.