Registro de cambios

Sci-Vis is ready for real research data

Sci-Vis now provides a complete data-to-figure workflow: upload one data file, generate a publication-ready figure, preview the result, and continue refining the same figure through conversation. The original dataset, generated versions, thumbnails, and download links stay with the conversation.

• Upload one CSV, TSV, TXT, XLSX, or XLS file up to 5 MB; Excel imports the first worksheet
• Choose chart types, journal style, color palette, figure size, multi-panel layout, and panel labels in Advanced Options
• Open any generated image at full size and continue editing from a selected historical version
• Progressive PNG preview appears as soon as it is ready, while the remaining export files finish in the background
• Automatic recovery, output-quality checks, and clearer failure guidance make long-running generations more reliable
• A richer guest demo now shows an initial single-cell figure followed by a publication-style revision

46 supported chart types

• Basic and trends: bar, line, scatter, dot, time series, waterfall, stacked area, and radar charts
• Distributions and statistics: box plot, violin plot, histogram, density plot, raincloud plot, beeswarm plot, Q–Q plot, ogive, ECDF, and ridgeline plot
• Relationships and models: regression plot, error-bar plot, correlation matrix, Bland–Altman plot, funnel plot, standard curve, bubble chart, hexbin plot, and paired dot plot
• Biomedical research: volcano plot, Kaplan–Meier curve, forest plot, ROC curve, Manhattan plot, MA plot, enrichment plot, dose–response curve, growth curve, enzyme-kinetics curve, precision–recall curve, calibration curve, and confusion matrix
• Flows and structures: heatmap, dendrogram, UpSet plot, Sankey diagram, stem-and-leaf plot, and calendar heatmap

Image and publication exports

Every completed figure supports PNG, PDF, EPS, and TIFF downloads. PNG becomes available first for preview and immediate download; the publication formats become active as soon as processing finishes.

VSDX (Visio) conversion — Beta

You can now rebuild a diagram image as a fully editable Microsoft Visio (VSDX) file. Unlike SVG/PPTX conversion, which makes the text editable, VSDX conversion turns the entire diagram into native objects: shapes you can move, connectors that stay attached, and labels you can rewrite.

• Two entry points: a new VSDX tab in the Image Converter, and a new Convert to Visio option (last in the menu) on generated images in the AI Drawing editor
• Scope guidance built in: works best on flowcharts, architecture and block diagrams; not recommended for photos, curve charts, or artistic illustrations — the UI says so up front
• Async conversion: takes about 1 minute; results appear with a preview and a .vsdx download
• 20 credits, charged only when the conversion succeeds
• Powered by LayerBack — for programmatic VSDX conversion, use the LayerBack API directly
• Available in all 11 interface languages

Convert menu polish

The convert popover in the AI Drawing editor was tightened up: shorter one-line descriptions per option and no more awkward line wraps in CJK languages.

Public Demo Mode

We added a logged-out Demo Mode for SciDraw AI so new visitors can preview the actual editor before creating an account.

This release covers both AI Drawing and Scientific Data Visualization:

• AI Drawing demo conversation: Visitors now land on a ready-made demo conversation showing text-to-image, sketch-to-image, and image-editing examples instead of an empty workspace
• Localized demo assets: Demo prompts and preview images are matched to the selected language, so the example figure and the prompt feel consistent
• Scientific Data Visualization demo: Logged-out users can explore a sample dataset, chart selections, and generated visualization results before uploading their own data
• Task-first entry points: Demo CTAs now point users toward concrete tasks such as mechanism figures, workflow diagrams, graphical abstracts, figure editing, and data visualization
• Cleaner sign-in handoff: When users are ready to generate, save history, continue editing, or use their own data, the sign-in prompt keeps them in the same product context

Consistent Workspace Preview

The demo is not a separate marketing page. It reuses the same workspace layout as the signed-in product, so visitors can understand how SciDraw AI works before committing to an account.

The signed-in experience remains focused on the user's own conversations, data, credits, and saved history.

Editable PPTX Export

We added a new PPTX conversion workflow for generated AI Drawing images, making it easier to continue editing scientific figures directly in Microsoft PowerPoint.

Key capabilities in this release:

• Editable PowerPoint text labels: Detected labels are exported as native PowerPoint text boxes, so users can rename, translate, recolor, and move labels without regenerating the figure
• High-quality figure background: The converted slide keeps the scientific illustration as a full-slide background while preserving editable OCR text on top
• Original-image preview in chat: PPTX conversion results appear in the conversation like other generated assets, with a visible PPTX badge and a normal image preview
• Correct PPTX download behavior: Clicking download on a PPTX result downloads the .pptx file, not the preview image
• Credit-aware conversion: PPTX conversion uses the same account credit flow as SVG conversion, with insufficient-credit handling built into the UI

Unified Convert Menu

The old "Convert to SVG" action is now a broader Convert menu.

Users can choose between three conversion types:

• Editable PPTX: Creates a PowerPoint file with editable text labels
• Extract Text Only SVG: Converts detected text into editable SVG text layers
• Convert All Elements SVG: Runs the full SVG vectorization workflow for layered vector output

The menu now uses clear format indicators (PPTX and SVG) so users can quickly tell which file type each option produces.

Why It Matters

Many researchers finish figures in PowerPoint because collaborators, labs, and supervisors already work there. This release gives SciDraw AI a more practical editing path: users can generate a figure with AI, convert it to PPTX, and make last-mile label edits in a familiar slide editor.

For workflows that need deeper vector editing, SVG export remains available alongside the new PPTX option.

New public Media Kit

We added a public Media Kit page so partners, writers, and academic creators have a single place to grab everything they need to introduce or recommend SciDraw AI.

What is included in this release:

• Single-click asset bundle: Logos (light & dark), Open Graph image, campaign visuals, and customer figure examples packaged in one downloadable ZIP
• Ready-to-use product copy: Twelve copy snippets — one-liner, tagline, short and long social posts, email pitch, and hashtags — each with a copy-to-clipboard button
• Feature briefs for partners: Six short feature descriptions sized for a single bullet point in articles, posts, or partner pages
• Eleven languages: Full translations across English, Simplified and Traditional Chinese, Japanese, Korean, German, French, Spanish, Brazilian Portuguese, Italian, and Russian
• Linked from the Academic Support Program: Researchers preparing institutional or community track applications can now grab the official assets and copy directly from the application page

The Media Kit is reachable from the footer Resources column and at /media-kit.

New Education Discount Workflow

We added a dedicated Education Discount workflow for students and educators who want to start using SciDraw AI at a lower cost.

Key capabilities in this release:

• Student and educator eligibility: Students can apply for a 40% discount, while teachers and educators can apply for a 20% discount
• Dedicated application page: Eligible users can submit their role, real name, institution, contact email, and proof documents from the education discount page
• Admin review flow: Admins can review applications, approve eligible users, reject incomplete submissions, and inspect duplicate-risk signals
• Automatic checkout discount: Once approved, the discount is applied automatically during checkout through the configured Stripe coupon
• One-time first-purchase support: Education discounts are only available before a user's first successful purchase and remain valid for 12 months after approval

This gives students, teachers, and academic staff a clearer path to access SciDraw AI while keeping review and checkout behavior controlled from the admin side.

New Figure Legend Tool

We added a new Figure Legend tool under Image Tools to help researchers draft figure legends directly from uploaded figures.

Key capabilities in this release:

• Legend drafting from figures: Upload a scientific figure and generate a concise, manuscript-style first draft of the legend
• Multilingual output: The generated legend can be written in the current page language or another supported output language
• Copy as text or LaTeX: Copy the result either as plain text or as a LaTeX caption for manuscript workflows
• Broader source support: The tool now accepts PNG, JPG, WebP, TIFF, SVG, and PDF files
• Preview support improvements: PDF and TIFF inputs are normalized for processing, and TIFF preview is now supported inside the tool

This feature is designed as a drafting assistant rather than a final submission substitute. Users should still review methods, abbreviations, sample information, and statistics before submission.

Video Tutorials

We published three video tutorials on YouTube to help new users get started with SciDraw AI's core features. Each tutorial is about 2 minutes long with bilingual narration and subtitles.

• AI Drawing Tutorial: Demonstrates all three drawing modes — Prompt (text to image), Sketch (sketch to image), and Replicate (image editing) — along with multi-round conversation editing
• Scientific Visualization Tutorial: Walks through uploading CSV data, configuring chart types and journal styles, AI-powered Python code generation, multi-format export, and iterative chart refinement
• Figure Checker & Image Converter Tutorial: Shows the full preflight workflow — checking DPI, format, grayscale readability, and colorblind safety, then fixing issues with the format converter (TIFF upscaling, SVG vectorization)

Blog Posts in 11 Languages

Each tutorial now has a companion blog post with the YouTube video embedded. Blog content is available in English, Chinese (Simplified & Traditional), German, Spanish, French, Italian, Japanese, Korean, Brazilian Portuguese, and Russian.

These posts serve as both a written guide for each feature and an SEO entry point to drive organic traffic from search engines to the platform and YouTube channel.

🔎 Figure Checker Launch

We added a new Figure Checker workflow for researchers who need to review figures before journal submission.

• Submission-focused preflight: Upload one figure and get a report built around common submission checks instead of raw file metadata alone
• Effective DPI by final width: Evaluate whether the same image still passes at single-column, double-column, or custom journal widths
• Format and transparency warnings: Catch risky exports like JPEG artifacts or alpha-channel issues before upload
• Accessibility review: Preview grayscale readability and common color-vision deficiency simulations in the same report

🧪 Built for Real Submission Moments

Figure Checker is designed for the moment when a journal or submission portal starts asking questions like:

• Is this figure really 300 DPI at the final layout size?
• Should I submit TIFF, PNG, or JPEG?
• Will this still pass at single-column or double-column width?
• Are labels, contrasts, and colors still readable in grayscale or colorblind conditions?

🌍 Public Landing Page + SEO Expansion

We also expanded the Figure Checker page into a stronger SEO landing page:

• Added richer FAQ content around journal figure requirements, 300 DPI, file format choices, and layout width
• Added new content blocks aimed at the real search intent researchers have right before submission
• Localized the page content across the existing supported locales for better discovery

💡 Why It Matters

Researchers often discover figure requirements late in the submission flow. This release gives SciDraw AI a dedicated preflight step so users can catch technical and readability issues before the journal system rejects the upload.

🌐 Image Text Language Selection

Control the language of text elements in your generated scientific illustrations:

• Multi-Language Support: Choose from 10+ languages including English, Chinese, Japanese, Korean, German, French, Spanish, and more
• Smart Defaults: Automatically selects your current page locale as the default language
• Sorted by Preference: Your current locale appears first, followed by English, then other languages
• Flag Indicators: Easy-to-identify language options with country flags

🎯 Use Cases

Perfect for researchers publishing internationally:

• Generate figures with Chinese labels for domestic journal submissions
• Create English diagrams for international publications
• Switch between languages without regenerating entire illustrations
• Maintain consistent terminology across multi-language publications

🔧 How It Works

1. Select your preferred language from the "Image Text Language" dropdown
2. Enter your prompt in any language
3. AI generates illustrations with text labels in your selected language
4. All text elements (axis labels, legends, annotations) follow the language setting

💡 Tips

• For bilingual publications, generate the same figure twice with different language settings
• Technical terms and scientific nomenclature remain consistent across languages
• Brand names and abbreviations (DNA, RNA, ATP, etc.) are preserved regardless of language selection

🖼️ Multi-Format Image Export

Export your AI-generated scientific illustrations in various professional formats:

• TIFF Export: Lossless compression, ideal for high-quality print and journal submissions
• EPS Export: Encapsulated PostScript format for professional publishing workflows
• PDF Export: Portable document format with embedded fonts and vector elements
• PNG/JPEG: Web-optimized formats with customizable quality settings

📐 Custom DPI Settings

Precise control over image resolution for different use cases:

• 300 DPI: Standard print quality for most journal submissions
• 600 DPI: High-resolution for detailed scientific diagrams
• Custom DPI: Set any resolution from 72 to 1200 DPI
• Auto-calculate dimensions: Preview pixel dimensions based on DPI selection

🎨 CMYK Color Mode

Professional color mode conversion for print publishing:

• RGB to CMYK: Automatic color space conversion for print-ready output
• Color Profile Support: Embed industry-standard ICC profiles
• Preview Mode: See how colors will appear in print before export

📊 Enhanced Batch Export

Export multiple images efficiently:

• Batch Format Conversion: Convert all conversation images at once
• Consistent Settings: Apply same DPI and format to multiple exports
• Download History: Track and re-download previous exports

💡 Journal Submission Ready

Optimized for academic publishing requirements:

• Meet strict format requirements for Nature, Science, Cell, and other top journals
• Automatic file naming with figure numbers
• Metadata preservation for archival purposes

📊 Credit Usage

• Format Conversion: 2 credits per conversion
• High DPI Export (600+): 3 credits per export
• Batch Export: Credits calculated per image

🔄 SVG Smart Conversion

Transform your AI-generated scientific illustrations into fully editable vector graphics:

• Text Extraction Mode: Extract all text elements as editable SVG text layers while preserving the original image as background
• Full Vectorization Mode: Convert the entire image including graphics into pure vector format for unlimited scalability
• One-Click Export: Download high-quality SVG files ready for professional editing

✏️ Professional Editing Compatibility

Generated SVG files work seamlessly with popular design and presentation tools:

• Microsoft PowerPoint: Import SVG directly, edit text, modify colors, resize without quality loss
• Adobe Illustrator: Full vector editing with layer preservation
• Figma / Sketch: Perfect for design workflows and collaboration
• Inkscape: Free open-source alternative with complete SVG support
• Google Slides: Easy import for online presentations

🎯 Key Features

• Smart OCR Recognition: AI-powered text detection using Google Cloud Vision for accurate text extraction
• Automatic Color Matching: Extracted text retains original colors from the source image
• Background Cleanup: AI automatically removes original text from the background using Recraft inpainting
• Font Size Preservation: Text elements maintain proportional sizes matching the original layout
• Scalable Output: Vector graphics can be resized to any dimension without pixelation

💡 Use Cases

Perfect for researchers and academics who need to:

• Edit figure labels for journal submission requirements
• Translate text to different languages for international publications
• Customize colors to match journal style guidelines
• Create presentation slides with editable scientific diagrams
• Modify diagrams for different contexts without regenerating

📊 Credit Usage

• Text Extraction: 5 credits per conversion
• Full Vectorization: 10 credits per conversion

SVG conversion is now available for all generated images in your conversation history!

🎨 Enhanced Image Quality

We've upgraded to Google's latest Gemini 3 Pro Image model to bring you even better results:

• Improved Image Quality: More detailed and accurate scientific illustrations
• Better Understanding: Enhanced interpretation of complex scientific concepts
• Consistent Output: More reliable generation across all modes (Text-to-Image, Sketch-to-Image, and Image Editing)
• Same Performance: Maintains 1K resolution with fast generation speeds

✨ What's Improved

The new model provides:

• More accurate representation of scientific concepts and terminology
• Better handling of complex diagrams and multi-element compositions
• Enhanced style consistency when using professional templates
• Improved text rendering in generated images

🎨 AI Generation Modes

We're thrilled to announce the official launch of SciDraw AI, the AI-powered scientific illustration platform trusted by 10,000+ researchers worldwide:

• Text-to-Image Mode: Describe your scientific concept in words, AI automatically generates publication-ready illustrations
• Sketch-to-Image Mode: Upload hand-drawn sketches, AI transforms them into high-quality professional figures
• Image Editing Mode: Upload reference images to replicate styles or continuously optimize through multi-round conversations

🎯 Professional Templates & Styles

• 7+ Professional Templates: TOC graphical abstract, journal cover, experimental workflow, mechanism illustration, signaling pathway, system architecture, comparison analysis
• 6+ Visual Styles: Flat illustration, isometric 3D, photorealistic, schematic diagram, artistic abstract, scientific illustration
• Multi-Aspect Ratios: Supports 1:1, 16:9, 9:16, 4:3, 3:4 for various publication requirements

⚙️ Core Features

• Multi-Round Optimization: Supports 3-5 rounds of continuous editing with conversation history
• Smart Prompt Fusion: Proprietary technology automatically combines user input with template and style libraries
• Complete History Management: All conversations and generated images permanently saved with smart auto-archiving
• High-Resolution Output: Up to 4K quality with Cloudflare R2 global CDN acceleration
• Credit System: Flexible pay-as-you-go pricing, new users get 20 free credits

🌍 International Support

• 7 Languages: Chinese, English, Japanese, Korean, German, French, Spanish
• Multi-Currency Payments: Stripe integration supporting 135+ countries
• Global CDN: Fast image delivery worldwide

🔒 Technical Stack

• Built with Next.js 15 and React 19 for optimal performance
• Google Gemini AI for intelligent image generation
• Drizzle ORM for type-safe database operations
• Radix UI for accessible component design
• Tailwind CSS for modern, responsive styling
• Cloudflare R2 for reliable image storage

🎁 Launch Promotion

• New users receive 20 free credits (4 images)
• Save 90% drawing time compared to traditional methods
• Reduce 95% costs compared to hiring designers