At some point every researcher faces the figure problem: the schematic that would make the paper, the graphical abstract the journal keeps hinting at, the pathway diagram the collaborators cannot stop redrawing in email. The internet's answer is usually "just use BioRender" — but BioRender is one option of at least seven, each with real trade-offs in cost, speed, and output quality.
This is the comparison without affiliate links: seven routes to scientific illustrations, what each is genuinely good at, where each hurts, and how to choose.
1. Vector editors (Illustrator, Affinity Designer, Inkscape)
Strengths: total control, industry-standard output, infinitely reusable components, vector export for any journal spec. Illustrator's scientific plugin ecosystem even adds lab-flavored icon libraries. Weaknesses: steep learning curve; every figure starts from a blank canvas; the good software costs money (Inkscape, to be fair, is free and capable); time cost per figure is the highest of any option here. Typical cost: Inkscape free; Affinity a one-off moderate fee; Illustrator by subscription. Wins when: you make figures constantly, need pixel-perfect control, or your figures must match an exact lab style across years.
2. Presentation software (PowerPoint, Keynote)
Strengths: you already know it; alignment guides are genuinely good; group members can edit your files; fast for simple block diagrams. An enormous fraction of published schematics quietly started here. Weaknesses: raster or awkward vector export; limited precision typography; figures that outgrow the tool start looking like it; version chaos across collaborators. Typical cost: usually already paid for. Wins when: the figure is simple, the deadline is tomorrow, and polish expectations are moderate.
3. Icon-library figure builders (BioRender, Mind the Graph and similar)
Strengths: purpose-built for biology and medicine — thousands of domain-accurate icons, drag-drop templates, fast learning curve, consistent output. For immunology or molecular biology schematics, nothing stock gets you to credible faster. Weaknesses: subscription pricing with feature tiers that frustrate (export limits, watermarks on free tiers); icon style is recognizable, so thousands of papers share a visual language; export resolution gated by plan; weaker outside life sciences — a physics or materials schematic has thinner icon coverage. Typical cost: annual subscription for meaningful use; check current tiers before committing. Wins when: you need a credible life-sciences schematic this week with minimal drawing effort.
4. General AI image generators
Strengths: unmatched speed from idea to visual; strong for concept art, cover-style imagery, and early brainstorming; improving fast at text-in-image. Weaknesses: scientific accuracy is not guaranteed anywhere — invented labels, impossible geometry, and confident nonsense are standard output; style is hard to pin across a manuscript; text handling still unreliable for precise terminology; licensing of outputs varies by tool; journals increasingly require disclosure. Typical cost: monthly subscription or per-image credits. Wins when: you need inspiration and visual exploration, not a submission-ready asset — yet.
5. Scientific AI figure tools (SciScroll's generator and peers)
Strengths: the general-generator experience tuned for research: understands scientific prompt vocabulary, biases toward diagram-style output, and connects figures to a researcher context — for instance generating directly into a platform where the figures can be shared and discovered, rather than into a downloads folder. Iteration speed on schematics and graphical-abstract drafts is the headline win. Weaknesses: the category is young — output still needs the same human verification pass as any AI work; label text typically needs final correction; less granular control than a vector editor once the draft exists. Typical cost: free tiers with paid plans for heavy use. Wins when: you need a credible scientific draft in minutes and plan to refine it — the draft-plus-editor workflow, where AI proposes and you dispose. Our AI scientific figure generation guide walks that workflow in detail.
6. Commissioning a professional illustrator
Strengths: journal-cover quality, a distinctive visual identity no template matches, and a collaborator who translates your napkin sketch into publication graphics while you do science. Weaknesses: cost per figure is real money; turnaround measured in days to weeks; revisions cost communication rounds; you cannot endlessly tweak the source file unless the illustrator hands it over. Typical cost: varies widely by complexity and region; graphical abstracts and full-figure commissions are usually quoted individually. Wins when: the stakes justify it — cover art, grant applications, review figure of the year, a figure destined for the lab's public face.
7. Code-drawn figures (Matplotlib schematics, TikZ, D3)
Strengths: perfect reproducibility — regenerate the figure from the script when data changes; version-controllable; typographically consistent with your document (TikZ especially); zero manual redo across hundreds of iterations. Weaknesses: effort goes into code instead of composition; the learning curve is a curve; organically "designed" schematics (cells, animals, machines) are painful or impossible; aesthetics lag deliberate design tools. Typical cost: free tooling; your time is the cost. Wins when: the figure is data-driven or must regenerate automatically — every figure that updates with every new dataset belongs here.
How to choose
Three questions settle most cases:
- Is the figure data? Then it is code-drawn, full stop — a plot must regenerate from real measurements.
- Is it a one-off schematic in life sciences with a deadline? Icon-library builders or a scientific AI tool for the draft, refined in a vector editor or presentation software.
- Is it the public face of the work — abstract, cover, grant hero figure? Professional illustrator, or AI-draft-plus-serious-human-refinement if budget is tight.
Most working scientists end up with a portfolio approach: code for plots, an AI or icon tool for schematics, and an editor for final polish — keeping one consistent color and typography system across all of it. That consistency, more than the tool, is what makes figures look professionally made; the template thinking in our graphical abstract guide applies to every route above.
Key takeaways
- There is no best tool — there is a best tool per figure type: data plots from code, schematics from icon libraries or AI drafts, hero figures from professionals.
- The draft-plus-refinement workflow (AI or template first, human polish second) is currently the best speed-to-quality trade for most researchers.
- Whatever the route, one manuscript-wide color and typography system is what makes the output look designed.
Frequently asked questions
Is BioRender free? It has a free tier, but meaningful use — full icon library, high-resolution export, publication licensing — sits behind paid plans whose terms and limits change periodically. Check the current pricing page before building your workflow around it, and compare against the alternatives on this list at the same usage level.
Are AI-generated figures allowed in journals? Increasingly yes, with disclosure: most major publishers expect AI assistance to be acknowledged and hold authors responsible for accuracy. What is not acceptable is AI imagery presented as data. Always check your specific journal's current AI policy — they are evolving quickly.
What is the fastest option for a graphical abstract tonight? A scientific AI generator for the composition draft, then thirty disciplined minutes of refinement: fix labels, apply your color system, verify every scientific relationship. For the structure to refine toward, use the template in our graphical abstract guide.
Which tool should a PhD student learn first? The one that matches their figures: wet-lab biologists get enormous value from icon-library tools; anyone with plots should learn a code-plotting library; and an AI figure generator plus a free vector editor covers most schematic needs at zero cost. All of these skills compound across a career.
Ready to make figures that look as good as your science? Browse real researcher-made examples on the SciScroll discover feed, try the AI generator in the Studio, and start free — plans scale when your lab does.