What DPI does Nature Communications require?

300 DPI minimum for photographs and bitmapped images, supplied as TIFF. Line art should be vector instead; if vector is impossible, 1,200 DPI.

What is the Nature Communications figure width?

88 mm for single column figures and 180 mm for double column, corresponding to at least 1,040 and 2,080 pixels of figure content respectively.

How many figures can I include?

Up to 10 display items in total, counting figures and tables together, depending on word count.

Does Nature Communications accept PowerPoint figures?

Yes, Word and PowerPoint files are accepted, but any placed images must also be supplied as separate TIFFs prepared to the photographic standard.

What font should figure labels use?

A clear sans serif such as Helvetica, with the same typeface and approximately the same size used across all figures in the paper.

Do I need final-quality figures for the first submission?

No. Initial submissions are format flexible and may combine text and figures in one file up to 30 MB. High resolution files are requested after acceptance.

Nature Communications Formatting Requirements 2026

Name: SciFig
Author: SciFig

Nature Communications requires vector line art, 300 DPI photographs, and figures built to 88 mm or 180 mm.

You have the figures. What you do not have is confidence that they will survive production without a "please resubmit at higher resolution" email. Nature Communications publishes its figure specifications in full, and most of the delays we see trace back to three of them: line art that was saved as a bitmap, figures built at the wrong column width, and text that stops being legible after reduction.

This page collects what the journal actually requires, in the order you need it while preparing files.

Nature Communications states its scientific figure requirements precisely, which is helpful — provided you build to them from the start rather than retrofitting.

Nature Communications figure specifications at a glance

Parameter	Requirement
Line art, graphs, charts, schematics	Vector only — EPS, AI, or PDF, exported directly from the source application
Line art that cannot be vector	1,200 DPI
Photographs and bitmapped images	TIFF, minimum 300 DPI
Single-column width	88 mm (≥ 1,040 pixels)
Double-column width	180 mm (≥ 2,080 pixels)
Target page size	210 × 276 mm
Colour mode (photographs)	RGB
Figure lettering	Sans-serif, e.g. Helvetica — same typeface and approximate size across every figure in the paper
Thinnest line	No smaller than 1 point
Maximum display items	10 (figures and/or tables), depending on word count
Chemical structures	ChemDraw template, submitted as .cdx at 100%

Everything above is stated in the journal's own submission guidance. The rest of this page explains what each one means in practice.

How wide should a Nature Communications figure be?

Figures must be built to one of two column widths — the journal states that "one- or two-column format figures are required," so an in-between width will be resized by production, usually not in your favour.

Single column: 88 mm, which at 300 DPI means at least 1,040 pixels of actual figure content
Double column: 180 mm, which means at least 2,080 pixels

The pixel counts exclude peripheral white space. A 2,080-pixel canvas with 300 pixels of margin on each side is a 1,480-pixel figure, and it will look soft in print.

Build to the final page size from the start. The journal asks you to prepare figures "to fit the pdf page size (210 × 276 mm)" — designing at final size is the only reliable way to check the thing everyone gets wrong, which is whether the labels are still readable after reduction.

Build at 88 mm or 180 mm from the start

Designing at final page size is the only reliable way to check that labels are still legible after reduction.

Explore the Tool

Does Nature Communications require vector figures?

The same line chart as clean vector paths versus degraded pixel blocks, showing why vector line art is required (illustration generated with SciFig)

For line art, effectively yes. The requirement is explicit: all line art, graphs, charts and schematics should be supplied in vector format — EPS, AI, or PDF — and "should be saved or exported as such directly from the application in which they were made."

The journal then closes the usual escape routes. Vector files "should not be flattened, compressed, converted or saved as bitmaps, jpegs or other non-vector file types." Exporting your plot to PNG and placing it in Illustrator does not make it a vector figure.

If you genuinely cannot supply vector — a rendering, a simulation output, a screenshot of an instrument display — the fallback is 1,200 DPI, prepared at page size. That is four times the resolution required for photographs, which tells you how much the journal would rather have the vector file.

Adobe Illustrator is the stated preference, but Word and PowerPoint files are accepted. If you submit either, supply any placed images as separate TIFFs prepared to the photographic standard below.

What resolution do photographs need?

TIFF at a minimum of 300 DPI, sized as close to final page size as possible, with colour photographs in RGB colour mode.

One specific warning in the guidance is worth repeating because it still happens: do not scan laser printouts of figures. The dot pattern on a laser print creates a moiré pattern when scanned, and no amount of resolution fixes it.

The journal is direct about consequences — figures that miss these standards "will not reproduce well and may delay publication until we receive high-resolution images."

What font and line weight should figures use?

Three requirements here are easy to satisfy and commonly missed:

Typeface. Clear sans-serif, Helvetica given as the example. The requirement that catches people is consistency: "the same typeface in approximately the same font size should be used for all figures in a paper." Figures made across a two-year project by three lab members will not satisfy this by accident.

Line weight. The thinnest lines in the final figure must be no smaller than 1 point. Default hairlines in most plotting libraries are thinner than this and disappear in print.

Legibility after reduction. The stated test is that labelling must be "of sufficient size and contrast to be readable, even after appropriate reduction." Multi-panel figures are held to the same rule — each panel must be sized so the whole figure can be reduced by the same amount and still show essential detail.

Panel labels are lower-case bold a, b, c, in the same type size as other figure text. Figure text is lower case with only the first letter capitalised.

What Nature Communications does not want to see

A clean flat bar chart beside a cluttered 3D chart with a truncated baseline marked undesirable, styles the journal rejects (illustration generated with SciFig)

The guidance names specific things to avoid, and these are stylistic rejections rather than technical ones:

Excessive boxing around figure elements
Unnecessary colour
"Spurious decorative effects" — three-dimensional 'skyscraper' histograms are called out by name
Highly pixelated computer drawings
Truncated vertical axes on histograms that exaggerate small differences

Two more conventions: use scale bars rather than magnification factors, with the bar length defined in the legend rather than printed on the bar itself; and use visual cues in legends rather than verbal explanations.

There is also a rule you will not find at most journals: Nature Communications no longer publishes or peer-reviews manuscripts using the Lena image. Authors who include it are asked to substitute it or justify its scientific necessity before review.

Can you use AI to generate a Nature Communications figure?

There is no blanket ban — and no blanket permission either. Nature Communications is a Springer Nature journal, and Springer Nature now governs AI use through a risk-assessment framework rather than a list of forbidden tools:

"Use of AI is governed according to risk level, not by tool type. The same framework applies across all roles: authors, reviewers and editors."

If you have seen the sentence "Springer Nature journals are unable to permit its use for publication" quoted in guidance about AI-generated images — including in earlier versions of this page — that wording is no longer on the publisher's policy page. Neither are the three exceptions that used to accompany it. The page carries no version number and no effective date, which is exactly why the operative habit is to read the current policy rather than any summary of it, this one included.

What the framework actually says about images. Use is sorted into three levels. The only image-specific entry in the Red — Not permitted tier is:

"Creating photorealistic images (deepfakes)"

The other Red entries that can catch a figure are about substance rather than medium: "Fabricating data, citations or results," and presenting AI-generated conclusions "as human-derived." A schematic of a signalling pathway is neither photorealistic nor fabricated data, so it does not land in Red by virtue of having been drawn with software.

That is not the same as being cleared. Two obligations apply to everything, at every level:

"Human accountability is non-transferable — Accountability for scholarly content, evaluation, and editorial decisions cannot be delegated to AI systems."

and, for authors specifically, "Disclose AI use for trust and transparency." Green-tier use is described as "likely to be reversible and verifiable," which is a useful test to apply to your own figure: can you show where every element came from, and change any of them?

Where to put the disclosure. The Portfolio-level policy states the principle; the journal-level guidance still gives the mechanics, and it is more specific: use of a large language model "should be properly documented in the Methods section (and if a Methods section is not available, in a suitable alternative part) of the manuscript." Methods is the safe default for a figure disclosure too. If your figure sits close to the line — it interprets, emphasises or frames results rather than just rendering them — the framework puts that in Amber, "permitted with human oversight, verification, and transparency through disclosure," which means the disclosure is not optional.

Policies differ sharply between publishers, so a figure prepared for one journal may need a different declaration at another: eLife permits AI-generated figures with a sentence in Materials and Methods. If your manuscript may be transferred or resubmitted elsewhere, plan for the strictest policy in your target set — the cross-publisher picture is in the guide to AI figure policies across journals.

What this means for using SciFig

Stated plainly, because getting this wrong costs you a submission.

The framework asks two things of any figure you did not draw by hand: that you remain accountable for it, and that the work be reversible and verifiable. A flattened image a model produced in one shot satisfies neither — you cannot show how any element got there. A vector canvas you edited element by element does: every object is separable, checkable and yours to change, and the editable PPTX or layered SVG output is the evidence. That is the difference the policy is actually pointing at, and it is why "who checked it" matters more here than "what tool opened it."

Three practical boundaries for this journal:

Disclose it. If a generative step contributed to a manuscript figure, say so in Methods. The current policy asks for disclosure even in its lowest-risk tier; under-declaring is the failure mode that costs you a correction, not over-declaring.
You still have to check every element. Accountability is non-transferable, and no export step transfers it. A pathway diagram with a mislabelled arrow is your error regardless of what drew the arrow — review it as you would a figure a co-author sent you.
Photorealistic imagery is out. That entry sits in Red without qualification, and it is the one line here that has no "with disclosure" escape.

None of this reaches the figures you make outside a manuscript — grant applications, conference posters, slide decks, theses, lab websites. A journal's AI policy governs what that journal publishes; it says nothing about the rest of your year, which for most researchers is the larger share of the figures they make.

How many figures can a Nature Communications paper have?

Up to 10 display items — figures and/or tables combined — with the exact allowance depending on word count. A limited number of uncaptioned molecular structure graphics and numbered equations may be included on top of that.

This is a real constraint, and it is stricter than some comparable journals. eLife, by contrast, sets no limit on the number of figures or figure supplements — if your study genuinely needs fifteen panels of primary data, the two journals will treat that very differently.

The allowance is explicitly tied to length: papers under 2,000 words should carry no more than four figures or tables. Nature Communications asks that the number of display items be "commensurate with the word length," so the figure budget and the word budget are one decision, not two — see the word limits below.

What are the word limits for a Nature Communications paper?

None of them are hard caps. Nature Communications states that it "does not have strict formatting requirements" and offers these as best practice — the phrasing in the author guidelines is "as a general guide, we suggest." Editors do not bounce a manuscript for running 5,200 words. What the numbers really tell you is the shape of paper the journal expects.

Element	Suggested limit
Main text	5,000 words (excluding abstract, Methods, references, figure legends)
Title	15 words maximum
Abstract	200 words, and no references
Methods	typically under 3,000 words
Each figure legend	350 words
References	should not exceed 70

Two of these matter more than the rest.

The abstract limit is the one to take literally. 200 words with no references is a tight budget for a general-audience summary that also has to carry your main result. It is also the field most likely to be checked mechanically at submission.

The main-text figure is where the word count meets your figure plan. Because display items have to be "commensurate with the word length," a 5,000-word Article supports the full ten display items, while a short 2,000-word paper is capped at four. If you are planning a multi-panel mechanism figure, that panel count is competing with your text budget — decide both together rather than discovering the conflict at submission.

What this means in practice: a paper near the 5,000-word guide with 8–10 display items reads as a normal full Article and needs no justification. Going substantially longer is allowed, but the editor will expect the length to be earned — more data, not more discussion.

Do figures need to be formatted this way at first submission?

No, and this saves real time. Nature Communications states that it "is flexible with regard to the format of initial submissions," does not require a particular structure or format at first submission, and lets you combine text and figures into a single Word, TeX/LaTeX or PDF file up to 30 MB.

Full-resolution figure files are requested after acceptance — the acceptance letter specifies what is needed and in which format.

The practical reading: prepare figures well enough to be judged at review, and do the production-grade export once the paper is accepted. What you should not do is design at the wrong column width at the start, because rebuilding a figure at 88 mm when it was composed at an arbitrary size means re-laying-out every label.

Figure problems that delay publication

From the requirements above, the failures that cost the most time:

Line art delivered as PNG or JPEG. Requires rebuilding from source, not just re-exporting.
Figure built at an arbitrary width. Labels that looked fine at 240 mm become unreadable at 88 mm.
Mixed typefaces across figures. Each figure looks fine alone; the set looks inconsistent.
Hairlines below 1 point. Invisible in print, and the plotting library's default.
Placed images inside Word or PowerPoint files without the separate TIFFs the journal asks for.

Four of these five are decided when the figure is first created, not when it is exported. If you are rebuilding figures to hit these specifications — or building them for the first time — generating them as editable vector output from the start removes the resolution and rebuild problems entirely, because there is no fixed pixel grid to fall below and every label stays editable.

Fix these before production does

Vector line art, 1 pt minimum line weight, and one consistent typeface across every figure in the paper.

Try Free

Where to go next

"When do these requirements actually bind?" — the first submission is format-flexible; these specifications apply at final submission. Nature Communications submission guide sets out which rules apply at which stage.
"How long until I hear anything?" — a median of 9 days to the first editorial decision, 242 days to acceptance. Nature Communications review timeline.
"Is it worth it?" — impact factor 18.1, five-year 18.9, plus the five other metrics the publisher reports. Nature Communications impact factor.

Frequently asked questions

Sources

Every specification here was read from the journal's own author pages, quoted for Articles — the only research format Nature Communications publishes. Reviewed 2026.

Nature Communications — How to submit — figure formats, resolution, column widths, lettering and line weight
Nature Communications — Article content type — word limits, display item allowance, abstract and title limits
Nature Communications — Open Access Fees and Funding — article processing charge
Nature Communications — Guide to authors — editorial scope and submission policy

See our editorial standards for how these guides are researched and corrected.