SciFig

Agent Details

SciFig is an AI scientific figure generator built for researchers who are not illustrators. Instead of assembling a figure icon by icon in an illustration tool, you describe what you need in plain language and SciFig generates it in publication style. It takes five kinds of input, not just text. Sketch a pathway on paper and photograph it — SciFig redraws it cleanly. Point it at a reference figure and it re-renders that layout in your own style. Drop in a PDF and it extracts a figure from inside the paper and regenerates it as an editable vector. Upload a lab photo and it becomes a labeled schematic. Already have a figure? The enhancer upscales it, removes the background, or rewords any label without touching the rest. Output is the part that actually matters for a paper. Figures come back as editable SVG — not a flat raster — and can be refined on a built-in vector canvas. Export to 8K PNG, editable SVG, or a PowerPoint deck in which every label is a native, retypeable text box, which is what most labs actually need for lab meetings, posters and slides. No watermark on any plan, including the free one. There is a free tier with daily credits, credit packs from $50, and Starter/Plus/Pro subscriptions. The interface is available in 8 languages, so non-native-English authors can describe a figure in their own language and still get publication-standard output.

Key Features

• Five ways in — a prompt, a hand-drawn sketch, a reference figure, a PDF, or a lab photo
• PDF figure extraction — pull a figure out of a paper and regenerate it as an editable vector
• Editable output — figures arrive as editable SVG, not a flattened raster
• Text-editable PPTX — labels export as native PowerPoint text boxes you can retype in PowerPoint
• Built-in vector canvas — refine, relabel, recolor and regroup after generation
• Figure enhancer — upscale, remove background, or reword any label on a figure you already have
• 8K PNG export — poster-scale raster when you need it

Use Cases

• 论文作者：Figure 1 的机制图 / 流程图，投稿前一周还能改标签
• 系统综述作者：PRISMA 流程图（站内该页搜索量最高之一）
• 会议海报：AAIC / EHA 等大会的 poster 图，8K 导出直接送印
• Graphical abstract：期刊要求的图形摘要，一句话生成再精修