PNG To Editable PowerPoint

Problem Being Solved

This skill targets a concrete workflow: you already have a finished PNG page, but you want it to become an editable PowerPoint slide. Typical inputs are architecture diagrams, process flows, report pages, infographics, or annotated diagrams with module boxes. If the page is inserted as one whole image, later edits to text, icons, spacing, or layout become cumbersome. The hard part is not “create a slide,” but decomposing the visual structure into editable native PowerPoint objects, while not forcing complex photos, screenshots, or realistic objects into vector shapes.

How It Works And Where It Fits

The core strategy is to rebuild the page instead of using the source image as the only slide object. Text becomes editable text boxes; module boxes and cards become rectangles or rounded rectangles; arrows, connectors, and flowchart elements become editable shapes and connectors; simple icons become native shapes; complex photo-like assets are embedded as cropped picture objects. In practice, the skill inspects the image size and visual structure, then creates a JSON layout based on layout_spec.example.json. It uses pixel coordinates, and the script maps pixels to PowerPoint EMUs at 96 DPI. Complex raster assets are handled in the assets section, while editable structure is reconstructed in elements, then the final PPT is generated.

It fits best when the input is a single image, the output is a single slide, editability matters, and the layout is mostly text plus shapes. Be aware that photographic content remains an image object, quality depends on accurate layout detection, and you should still check object counts, aspect ratio, and rendered output for clipped text or missing areas.