Scientific Reports Figure and Formatting Requirements (2026): DPI, Width, Fonts, File Types

Scientific Reports publishes more papers than any other journal in the world, which means more people are hunting for its figure specifications than for almost any other set. It also means most of what you find is wrong — advice written for Nature and relabelled, or numbers invented to fill a table.

This page separates two things that are usually mixed together: what Scientific Reports itself publishes in its submission guidelines, and what the Nature Portfolio final-artwork guide adds. Where the journal states nothing, this page says so rather than filling the gap with a number that sounds right.

Scientific Reports figure requirements at a glance

What resolution does Scientific Reports require?

The submission guidelines do not give a DPI for ordinary figures. The only resolution figure anywhere on that page applies to chemical structures, which must be exported as a 300 dpi RGB TIFF.

This is not an oversight. The guidelines push you towards vector formats for anything that is not a photograph, and a vector file has no resolution to specify — it renders at whatever size the typesetter needs. The journal's own phrasing is that figures which "do not meet these standards will not reproduce well and may delay publication until we receive high-resolution images."

For the panels that genuinely are bitmaps — micrographs, gels, photographs — the Nature Portfolio guide to preparing final artwork is explicit: place bitmapped images at 300 dpi, or at their native resolution if they were captured below that. Upsampling a 150 dpi micrograph to 300 dpi in Photoshop adds pixels but no detail, and the image-integrity policy is the reason not to try.

So the practical answer:

• Plots, diagrams, schematics: export as vector. DPI does not apply.
• Photographic panels: 300 dpi at the size the panel will actually print.
• Never upsample to hit a number.

How wide should a Scientific Reports figure be?

Scientific Reports states no width. This is the single most commonly invented specification on pages about this journal, so it is worth being precise about what is and is not on the record.

The Nature Portfolio final-artwork guide gives, for Articles and comparable formats:

Those numbers come from the guide that covers Nature Portfolio journals at the final-artwork stage, not from the Scientific Reports submission page. Treat 180 mm as the maximum design width and you will not be wrong in a way that costs you anything.

What the journal does state is the constraint that actually matters, and it is about reduction rather than width: labelling must be "of sufficient size and contrast to be legible, even after appropriate reduction", and each panel of a multipart figure should be "sized so that the whole figure can be reduced by the same amount and reproduced at the smallest size at which essential details are visible."

The test that follows from this is simple and worth running: print the figure at 88 mm wide. If any label, axis number or scale bar is unreadable, the figure fails — regardless of what width you designed it at.

What is the minimum font size in a Scientific Reports figure?

No point size appears in the submission guidelines. What they require instead:

• A clear sans-serif typeface, Helvetica given as the example.
• The same typeface at the same size across every figure in the paper.
• A 'symbols' font for Greek letters.
• Lettering large enough to survive reduction.

The Nature Portfolio artwork guide is the one that gives numbers: text should be sans-serif, preferably Helvetica or Arial, with a minimum of 5 pt and a maximum of 7 pt.

Two things surprise people about that range. The first is how small 5 pt is — it is a final-artwork size, measured after the figure has been reduced to column width, not the size you type in your plotting library. The second is that there is a maximum. Oversized axis labels are as much a house-style violation as tiny ones, and a figure that arrives with 14 pt text gets redrawn.

The rule that keeps you out of trouble: build the figure at final print size from the start, so the point size in your editor is the point size on the page.

Which file formats does Scientific Reports accept?

The submission guidelines split this by content type:

• Line art, graphs, charts and schematics — vector format, "such as EPS or AI", saved or exported directly from the application in which it was made.
• Photographic and bitmap images — "a bitmap image format such as tiff, jpg or psd". If you save TIFFs, select the compression option to avoid very large files.
• Not accepted — Word or PowerPoint files with placed images.
• Chemical structures — produced in ChemDraw or similar, exported to a 300 dpi RGB TIFF.

Two upload rules cause more resubmissions than the formats do:

1. Each complete figure is a separate file. A multi-part figure must be prepared and arranged as a single image file including all sub-parts — do not upload panel a, panel b and panel c individually.
2. The stage matters. At initial submission you may embed figures in the manuscript file (a single file up to 3 MB, Word preferred, LaTeX or PDF also accepted). At revision you must upload all figures as separate files, and PDF is no longer accepted for the article text itself.

There is one wrinkle worth knowing if you look at both documents. The Nature Portfolio artwork guide, describing the application used to assemble the final layout, recommends AI, EPS, PDF or PPT and says it does not accept BMP, GIF, GIMP, JPG, PNG, Tex or TIFF for that layout file. That is not a contradiction of the submission page — it is about the container file you build the figure in, not the photographic panels you place inside it. Photographs still go in as bitmaps; the text, axes, arrows and scale bars laid over them should stay editable vector.

RGB or CMYK — and why this one is a trap

Scientific Reports accepts either. The sentence that follows is the one to read twice:

Images can be supplied as RGB or CMYK (note: we will not convert image colour modes).

The journal will not fix your colour mode. So whichever you choose, every panel in every figure has to agree, and the colours you approve on screen are the colours that go to press.

The Nature Portfolio artwork guide sets the default by content type: original research content in RGB, and other content such as Perspectives, Progress Articles and Review Articles in CMYK. Scientific Reports publishes original research, so RGB is the sensible default unless you have a specific reason otherwise.

The practical failure mode is mixing: a vector plot exported as RGB placed next to a micrograph that came off the microscope in CMYK. Both are individually acceptable and the combination still looks wrong in print.

How many figures can a Scientific Reports paper have?

Display items are limited to 8, and that limit counts figures and tables together — not 8 of each.

Beyond the hard cap the journal asks for proportion: "to enable typesetting of papers, we advise making the number of display items commensurate with your overall word length. So, for Articles of 2,000 words or less, we suggest including no more than 4 figures/tables."

Also relevant to how you package them:

• Tables have a maximum size of one page.
• Schemes should not be used — present a sequence of chemical reactions or experimental procedures as a figure with a caption instead.
• A limited number of uncaptioned molecular structure graphics and numbered equations may be included and do not consume the display-item budget in the same way.

Word, title, abstract and legend limits

Scientific Reports does not impose strict limits in most cases, but it publishes strong recommendations, and figure legends are the one place where the number is firm.

Two details that catch people out. The abstract must be unstructured — no Background/Methods/Results subheadings, which is the opposite of what many clinical journals want, so a resubmission from one of those needs rewriting. And footnotes should not be used at all.

A data availability statement is mandatory.

Does Scientific Reports accept graphical abstracts?

No. The guidelines state it in one line: Scientific Reports does not support graphical abstracts.

This matters because it is often the first thing prepared when a manuscript is moved from another journal. There is nowhere to upload it, and no editor will see it. If you have one already, the useful move is to strip it for parts — a graphical abstract is usually a cleaner, better-composed summary of the study than Figure 1, and Figure 1 is what carries that job here.

Going the other way, to a publisher that does want one, the size and content rules change per publisher and none of them accepts a reused paper figure. Our graphical abstract requirements guide sets out what Elsevier, Cell Press, ACS and MDPI each specify.

The house-style rules that get figures sent back

These are the specifications people skip, and they are the ones an art editor notices immediately. All are from the submission guidelines.

Lettering and panels

• Panels are labelled with a lower-case bold letter — a, b, c — in the same type size as the rest of the figure. Not (A), not capitals.
• Lettering is lower-case, with only the first letter of each label capitalised.

Numbers and units

• One space between the number and the unit.
• SI nomenclature — ms, not msec.
• Thousands separated by commas: 1,000.
• Unusual units or abbreviations spelled out or defined in the legend.

Scale and axes

• Use scale bars, not magnification factors, and put the length on the bar itself rather than in the legend.
• Never truncate the vertical axis of a histogram to exaggerate a small difference.
• The thinnest line in the final figure must be no smaller than 1 point.

Composition

• White background.
• No excessive boxing, unnecessary colour, or "spurious decorative effects (such as three-dimensional 'skyscraper' histograms)".
• No highly pixelated computer drawings.
• A figure should not contain more than one panel unless the parts are logically connected.
• In legends, use visual cues rather than verbal explanations — do not write "open red triangles" when the figure can show them.

Attribution

• State the source of any image. If you or a co-author drew it, say so in the acknowledgements.
• For software, state the name, version number and URL.
• You are responsible for permission to publish any copyrighted figure, and images downloaded from the internet cannot be published without appropriate permission.

That last one is worth reading as a warning rather than a formality. "We cannot publish images downloaded from the internet without appropriate permission" covers a great deal of what ends up in methods schematics.

Does Scientific Reports allow AI-generated figures?

The submission guidelines address the authorship side directly: large language models such as ChatGPT do not satisfy the authorship criteria, because authorship carries accountability that cannot be applied to a model. Use of an LLM "should be properly documented in the Methods section (and if a Methods section is not available, in a suitable alternative part) of the manuscript."

The guidelines then point to a dedicated AI section in the editorial policies, which is where the current position on generated images lives.

Two things follow. First, the rule that governs a figure is the digital images integrity and standards policy, and it applies to what was done to the image, not to which tool produced it — adjusting, splicing or enhancing a panel in a way that misrepresents the data is the violation, whether it happened in Photoshop or a model. Second, this is the fastest-moving part of the guidelines. The figure specifications on this page have been stable for years; the AI policy has not. Check it at submission, not from memory.

For what it is worth, the honest line to take with a schematic: a generated illustration of a mechanism or workflow is a drawing, and drawings have always been allowed. A generated image of data is not.

A pre-submission checklist

Run this against the actual files you are about to upload, not against the figures as you remember them.

• Every plot, chart and schematic is vector (EPS or AI), exported from the source application.
• Photographic panels are TIFF, JPG or PSD at 300 dpi at final size.
• No Word or PowerPoint files containing placed images.
• Each figure is one file, with all its panels already assembled.
• Thinnest line ≥ 1 pt. Text between 5 and 7 pt at final size.
• One sans-serif typeface at one size across all figures.
• Panel labels are lower-case bold a, b, c.
• Colour mode is consistent across every panel — the journal will not convert it.
• White background, no drop shadows, bevels or 3D effects.
• Scale bars carry their length on the bar.
• Figures and tables together total 8 or fewer.
• Every legend is under 350 words and describes the error analysis.
• Software name, version and URL stated; image sources credited.
• Printed at 88 mm wide, every label is still readable.

Frequently asked questions

What DPI does Scientific Reports require for figures? The submission guidelines state no DPI for ordinary figures — the only resolution given is 300 dpi RGB TIFF for chemical structures. The Nature Portfolio artwork guide says bitmapped images should be placed at 300 dpi, or at native resolution if captured below that. Supply line art as vector and the question mostly stops applying.

What figure width does Scientific Reports use? Not stated by the journal. The Nature Portfolio final-artwork guide gives 88 mm single column and 180 mm double column. Design to 180 mm maximum and check legibility at 88 mm.

What is the minimum font size in a Scientific Reports figure? No point size in the submission guidelines — they require a sans-serif typeface at a consistent size that stays legible after reduction. The Nature Portfolio artwork guide gives 5 pt minimum, 7 pt maximum, preferably Helvetica or Arial.

Can I submit figures embedded in my Word document? At first submission, yes — you may combine text and figures in a single file up to 3 MB. At revision, no: all figures must be uploaded as separate files, and PDF is not accepted for the article text.

How many figures am I allowed? Eight display items in total, counting figures and tables together. Four is suggested for Articles of 2,000 words or less.

Does Scientific Reports have a graphical abstract? No — the journal does not support them.

Can I reuse a figure from another paper? Only with permission, which is your responsibility to obtain, and the source must be stated. This includes figures published elsewhere and photographs taken by professional photographers.

Where to go next

The specifications above are the easy half. The hard half is that most figures fail them because of how they were made: exported once from an analysis script at screen resolution, then patched in whatever tool was open, until nothing in the file is editable any more.

If you are rebuilding rather than patching, start with our practical guide to making scientific figures, which works through the column-width, resolution and type-size decisions in the order they arise, across three publishers rather than one.

After that, the two guides worth reading next are our poster guide, which covers the same reduction-and-legibility problem at the opposite scale, and our comparison of figure tools, which is honest about which ones actually produce publication-grade vector output.

For the artwork rules this journal inherits, stated in full by the title that sets them, see our Nature impact factor and figure guide. And for a multidisciplinary journal that specifies far more than Scientific Reports does, our PNAS impact factor and figure guide is the closest comparison worth reading.