image-to-editable-ppt

Convert raster images into element-level editable PowerPoint slides with native text and cropped picture objects.

2|Updated Jul 27, 2026
One-click install
npx skills add https://github.com/Arinin77/ppt-editable-workflow --skill image-to-editable-ppt-arinin77
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: image-to-editable-ppt
Source: https://github.com/Arinin77/ppt-editable-workflow/tree/main/skills/image-to-editable-ppt
Command: npx skills add https://github.com/Arinin77/ppt-editable-workflow --skill image-to-editable-ppt-arinin77

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires Pillow, and includes scripts (resource) and references (resource) components.

What problem does it solve? Screenshots, storyboard slides, diagrams, and posters are flat bitmaps that cannot be edited in PowerPoint. This Skill decomposes a raster image into separate PPT elements—native editable text and simple shapes plus crop-safe picture objects for icons, figures, and charts—so every element can be moved, resized, or replaced independently. ## Core Features & Use Cases - Element-Level Decomposition: Rebuilds ordinary text, numbers, simple lines, and boxes as native PPT objects while cropping stylized icons, logos, charts, tables, and photos as named picture objects with source and display bounding boxes. - Manifest-Driven Build: Uses a JSON conversion manifest with pixel coordinates and a Python script (build_editable_pptx.py) that writes a one-slide OpenXML PPTX per input image. - Mandatory QA Gate: Requires rendered preview comparison against the source image, with a blocking-defect checklist covering clipped crops, OCR errors, broad screenshot crops, and layout drift, plus retry until defects are resolved. - Use Case: Convert an imagegen storyboard slide or a UI mockup screenshot into a one-slide PPTX where the title is editable text, the robot icon is a padded crop named model_brain, and the chart is a complete figure object—verified against the original render. ## Quick Start Use the image-to-editable-ppt skill to convert this screenshot into an element-level editable PowerPoint slide with native text and named picture crops.

Frequently Asked Questions about image-to-editable-ppt

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert a screenshot into an editable PowerPoint slide?▼

Inventory every visible element, rebuild ordinary text and simple boxes as native PPT objects, and crop icons, figures, and charts as named picture objects. Then run build_editable_pptx.py with a JSON manifest to emit a one-slide PPTX and compare the rendered preview against the source image.

How to convert imagegen storyboard slides to editable PPTX?▼

Keep ordinary text and simple geometry native, crop only meaningful visual assets as named objects like model_brain or road_rocket, and record each crop with source_bbox, display_bbox, role, and padding. Avoid dense anonymous segmentation that produces hundreds of tiny fragments.

What Python dependencies does the PPTX build script need?▼

The build_editable_pptx.py script uses only the Python standard library plus Pillow for image cropping and saving. It writes the OpenXML PPTX package directly with zipfile, so no python-pptx installation is required.

Should charts and tables be cropped or rebuilt as native PPT objects?▼

Crop charts, plots, and tables as complete independent picture objects by default, including axes, legends, labels, and colorbars. Rebuild them natively only when the user explicitly requests chart or table editability.

Why do cropped icons look clipped in the generated slide?▼

Clipping happens when crop bounds are too tight for shadows, glows, antialiased strokes, or circular outlines. Add 4-12 px or more of safety padding via source_bbox while keeping the display bbox unchanged, then rebuild and re-check the preview.

What are the limitations of image-to-editable-PPT conversion?▼

The output is one slide per input image, and only text, numbers, simple lines, and simple boxes become native objects; complex visuals remain picture crops. Pixel-accurate positioning is prioritized over exact font matching, and OCR-heavy stylized text may need to stay as a crop.