Every researcher who has submitted a manuscript knows the frustration: you spend hours perfecting a figure, export it as PNG, upload it to the journal portal — and get a desk rejection email saying your figures don't meet format requirements. The resolution is too low, the format is wrong, or the editor asks you to change one label and you realize you can't because the file isn't editable.
In 2026, this problem has gotten worse, not better. AI image generators have made it trivially easy to create beautiful scientific illustrations. But most of them export only flat raster images — PNG or JPG files where every element is baked into a single layer of pixels. You cannot select a text label. You cannot move an arrow. You cannot change a color without regenerating the entire figure.
This is why editable exports — specifically SVG and PPTX — are not a nice-to-have feature. They are the difference between a figure you can use once and a figure you can use throughout your entire publication lifecycle.
What "Editable" Actually Means for Scientific Figures
An editable scientific figure is one where individual components remain separate, selectable objects:
- Text labels can be selected, retyped, and reformatted
- Arrows and connectors can be moved, resized, and recolored
- Shapes and icons can be repositioned without affecting other elements
- Colors can be changed globally (e.g., switching from a color scheme to grayscale)
A flat PNG has none of these properties. Once exported, a PNG is just a grid of colored pixels — the AI's "understanding" of your figure's structure is lost.
SVG: The Format Journals Actually Want
SVG (Scalable Vector Graphics) is an XML-based vector format. Every shape, line, and text element is defined mathematically rather than as pixels. This means:
- Infinite resolution — SVG files look sharp at any zoom level, from a phone screen to a conference poster
- Tiny file sizes — a complex diagram might be 50 KB as SVG versus 5 MB as high-resolution PNG
- Editability — open in Illustrator, Inkscape (free), or any SVG editor and modify any element
- Easy conversion — SVG converts losslessly to EPS, PDF, and other vector formats journals accept
Which journals require or prefer vector formats?
| Journal / Publisher | Vector format accepted | Notes |
|---|---|---|
| Nature Portfolio | EPS, PDF, SVG | "We strongly prefer vector formats for line art and diagrams" |
| Science (AAAS) | EPS, PDF | Vector required for all non-photographic figures |
| Cell Press | AI, EPS, PDF | "All text must be editable" — this rules out flat raster |
| PLOS ONE | EPS, TIFF | EPS preferred for diagrams |
| Elsevier | EPS, PDF, SVG | SVG explicitly listed as acceptable |
| Springer Nature | EPS, PDF, SVG | "Vector graphics are preferred for line drawings" |
The pattern is clear: every major publisher either requires or strongly prefers vector formats for diagrams and schematics. Flat PNG/JPG is acceptable only for photographs and microscopy images.
PPTX: The Collaboration Format Nobody Talks About
While SVG is the right format for final journal submission, PPTX (PowerPoint) is the right format for the 90% of the figure's lifecycle that happens before submission:
- Lab meetings — presenting draft figures to your PI and getting feedback
- Co-author review — sending figures to collaborators who don't have Illustrator
- Conference presentations — using figures directly in slide decks
- Revision rounds — quickly modifying figures based on reviewer comments
When FigPad exports a PPTX file, every element in your scientific figure becomes a native PowerPoint object. Your co-author can open it in PowerPoint, double-click a text label, change it, and save — no design software required.
This solves a real collaboration problem. In a typical multi-author paper, the person creating the figures and the person reviewing them are different people. If figures exist only as flat images or Illustrator files, the review cycle becomes:
- Creator exports PNG → sends to co-author
- Co-author annotates feedback on the PNG ("move this label up, change this color")
- Creator opens the source file → makes changes → re-exports → sends again
- Repeat 3-5 times per figure
With editable PPTX, step 2 and 3 merge: the co-author makes the change directly and sends back the modified file.
The Real Cost of Non-Editable Figures
Here's a scenario every researcher has experienced:
You submit a manuscript with 7 figures. Reviewer 2 asks you to "add error bars to Figure 3B and change the color scheme in Figure 5 to be colorblind-friendly." If your figures are flat PNGs generated by an AI tool:
- Figure 3B: You need to regenerate the entire figure with a modified prompt, hoping the AI produces something close to the original but with error bars. It probably won't match exactly, so you might need to regenerate Figure 3A as well for consistency.
- Figure 5: You need to regenerate with a different color specification. Again, the rest of the figure will change unpredictably.
Total time: 2-4 hours of regeneration, comparison, and quality checking.
If your figures were exported as editable SVG:
- Figure 3B: Open in Inkscape, add error bar elements, save. 10 minutes.
- Figure 5: Open in Inkscape, select all red elements, change to blue. Select all green elements, change to orange. 5 minutes.
Total time: 15 minutes.
Over the course of 2-3 revision rounds with 7 figures each, editable exports save roughly 10-15 hours per manuscript.
How to Get Editable Exports from AI-Generated Figures
Not all AI figure tools offer editable exports. Here's the current landscape:
| Tool | SVG Export | PPTX Export | Editable? |
|---|---|---|---|
| FigPad | Yes | Yes | Fully layered and editable |
| Midjourney | No | No | Flat raster only |
| DALL-E | No | No | Flat raster only |
| Stable Diffusion | No | No | Flat raster only |
| BioRender | Yes (paid) | No | SVG is editable; no PPTX |
| Canva | SVG (paid) | PPTX | Partially editable |
FigPad's approach is to run a vectorization step after AI generation. The raw AI output (a raster image) is processed through a vector conversion pipeline that identifies structural elements — text, shapes, arrows, connectors — and converts them into separate SVG layers. The PPTX export then maps these SVG elements to native PowerPoint objects.
This is different from simply "saving as SVG." A naive raster-to-SVG conversion (like using an auto-trace tool) produces a single complex path that looks like the original but isn't actually editable in any meaningful way. FigPad's pipeline produces semantically meaningful layers — a text label is an actual text element, not a traced outline of letter shapes.
Practical Workflow: From AI Generation to Journal Submission
Here's the workflow that minimizes rework and maximizes flexibility:
- Generate your figure in FigPad using a text prompt
- Export as SVG — this is your editable master file
- Export as PPTX — send this to co-authors for review and feedback
- Iterate using the SVG in Illustrator/Inkscape based on co-author feedback
- Final export from Illustrator as EPS or PDF (the formats most journals prefer for final submission)
- Keep the SVG and PPTX for revision rounds — when reviewers ask for changes, you edit the SVG directly instead of regenerating
The key insight: treat the AI generation as Step 1, not the final step. The AI gives you a structured starting point. The editable exports give you complete control over the finishing process.
Key Takeaways
-
Most AI image generators export only flat PNG/JPG files that cannot be edited after creation — this makes revision rounds extremely time-consuming.
-
Major journals (Nature, Science, Cell, Elsevier) require or strongly prefer vector formats (EPS, PDF, SVG) for diagrams and schematics.
-
Editable SVG exports let you modify individual text labels, arrows, and shapes in Illustrator or Inkscape — saving 10-15 hours per manuscript across revision rounds.
-
Editable PPTX exports solve the collaboration problem by letting co-authors modify figures directly in PowerPoint without specialized design software.
-
The optimal workflow is: generate with AI → export SVG (editable master) and PPTX (collaboration) → fine-tune in Illustrator → submit EPS/PDF to journal.
Dr. Anna Kovács is a research fellow in neuroscience at the Max Planck Institute for Brain Research. She serves as a figure preparation consultant for three research groups and has guided over 40 manuscripts through the journal submission process.
👉 Try FigPad free — export editable SVG and PPTX from AI-generated figures