From a Methods Paragraph to a Figure Prompt: A Worked Example
Build a microfluidics diagram from a structured brief, distinguish flow from causality, and revise one relationship without redrawing the whole figure.
“Draw a professional scientific diagram of our experiment” leaves almost every scientific decision to the image model. A more useful prompt identifies the objects, the relationships between them, and what the reader should learn. This walkthrough turns a short methods description into a diagram brief you can actually revise.
Our example is illustrative, not a report of experimental results: two solutions enter a microfluidic chip, mix in a serpentine channel, and leave through one outlet. A microscope observes a region after the mixing section. The intended figure explains the apparatus; it does not claim that the mixture is homogeneous or that a treatment worked.
1. Extract the structure before choosing a style
Suppose your methods paragraph says:
Solution A and solution B were supplied through separate inlets. Their streams joined at a Y junction and passed through a serpentine channel. Images were acquired downstream of the mixing region, and the outlet fluid was collected in a tube.
Convert it into a small relationship table:
| Object | Connection | Meaning in the drawing |
|---|---|---|
| Inlet A | Enters the left branch of the Y junction | Fluid path |
| Inlet B | Enters the right branch of the Y junction | Fluid path |
| Y junction | Leads into the serpentine channel | Physical connection |
| Observation region | Lies after the serpentine section | Measurement location |
| Outlet | Connects to a collection tube | Fluid path |
| Microscope | Views the observation region | Observation, not fluid flow |
That last distinction matters. Drawing the microscope into the same chain as the fluid can suggest that liquid travels through the microscope. Represent observation with a dashed sight line, and reserve solid arrows for the flow direction.
2. Use this first-draft prompt
Paste the following into the figure workspace, adapting the geometry to your actual apparatus:
Create an explanatory apparatus diagram, viewed from above.
Show a microfluidic chip with two separate inlets at the left. Label
those inlets exactly "Solution A" and "Solution B". Join their channels
at a Y junction. Continue through one serpentine mixing channel, then
one straight observation section, and finally one outlet at the right.
Connect the outlet to a collection tube labelled "Collection".
Use solid arrows inside channels to indicate left-to-right fluid flow.
Place a microscope objective above the observation section. Show its
viewing direction with a dashed line; do not connect it to the fluid path.
Label the observation section exactly "Imaging region".
Use blue for Solution A and orange for Solution B. After the junction,
show both streams schematically without implying measured mixing quality.
Keep the chip background white, with dark channel outlines. Use a wide
horizontal composition and place labels outside the channels.
Do not invent channel dimensions, flow rates, scale bars, measurements,
results, extra inlets, or additional components.
The color choice communicates the identity of the two inputs. It does not encode concentration. If concentration matters, give the actual values separately and explain the encoding in the caption.
3. Inspect topology before typography
Do the channels connect in the right order? Are there exactly two inlets and one outlet? Does the objective view the correct section? A beautifully typeset diagram with three inlets is still the wrong apparatus.
Write corrections as discrete operations:
| Problem in the draft | Useful revision request |
|---|---|
| Microscope is drawn before the mixer | “Move only the microscope objective and its dashed viewing line to the straight section after the serpentine channel. Keep all channel geometry and labels.” |
| Outlet arrow points toward the chip | “Reverse only the arrow in the outlet so it points toward the collection tube.” |
| A third inlet was added | “Remove the unlabelled third inlet and its connecting channel. Preserve the two labelled inlets and their Y junction.” |
| Labels overlap the channel | “Move the two inlet labels above their respective branches. Do not rename them or change flow direction.” |
Avoid “make the science more accurate.” It gives the model no specific edit to carry out and no invariant to preserve.
4. Decide which information belongs in the caption
A useful figure and caption divide the work. The drawing should carry spatial relationships and reading order. The caption can state the experimental conditions, abbreviations and interpretation.
For this example, a draft caption could read:
Schematic of the two-inlet microfluidic setup. Solutions A and B meet at a Y junction, pass through a serpentine section and are imaged downstream before collection. Solid arrows show fluid flow; the dashed line indicates the microscope viewing direction. The drawing is not to scale.
Replace that caption with your actual apparatus description. Add verified flow rates or dimensions only when they are needed to understand or reproduce the experiment. A decorative scale bar is particularly misleading on a schematic that was never drawn to scale.
5. Finish at the intended reading size
If the output is a raster image, text is part of the pixels. Renaming “Imaging region” may require regeneration or a manual overlay. If you need editable labels, inspect the exported file in the SVG editor: can you select the label separately, and does editing it leave the channel unchanged? An SVG extension alone does not establish that.
Finally, place the diagram at its intended manuscript width, rather than judging only a full-screen preview. Check that the dashed observation line remains distinct from solid flow arrows and that both inlet labels are readable. Keep the prompt and the accepted version together so a coauthor can understand what the diagram was meant to show.
The practical sequence is objects → relationships → first draft → local corrections → caption → final-size inspection. Style comes after the scientific structure is clear.