Can GordenSuperPPTSkills End the AI‑PPT Struggle? From Visual Drafts to Editable Slides

Core Goal

Generate high‑quality visual PPT and then restore it to a fully editable PPTX.

AI‑generated PPT looks good but is hard to edit; manual re‑layout is slow.

Common Issues with Existing AI PPT Tools

Good visual effect but not editable.

Editable but plain layout.

Complex pages require costly manual tweaks.

Project Approach

Leverage a large model to create complex, high‑density, visually luxurious slides.

Apply visual recognition, layer decomposition, and coordinate reconstruction to split images into editable structures.

Output both image‑based PPT and editable PPTX.

Skill Breakdown

1. GordenImagePPTGen

Generates image‑based PPT. Input: topic/content. Output: per‑page PNG + image PPTX. Suitable when you only need a striking visual and do not care about deep editing.

2. GordenImage2PPTX

Converts image PPT to editable PPTX. Input: image. Output: editable PPTX. Decomposes each slide into background, framework, icon, and text layers, then reassembles them in PowerPoint so text becomes real text boxes and icons become movable objects.

3. GordenSuperPPTSkill

Chains the first two skills to deliver a complete workflow: from topic to image PPT, to editable PPTX, plus intermediate artifacts.

Technical Workflow

Step 1: Generate Image PPT

Uses GPT’s image generation to produce page‑by‑page visual drafts with high information density, complex layouts, luxurious style, and real text instead of placeholders.

Step 2: Reverse‑engineer Page Structure

Background layer – remove noise and rebuild a clean background.

Framework layer – extract containers, cards, connectors, chart structures.

Icon layer – isolate decorative icons and artistic elements.

Text layer – recognize text content, position, font size, weight, alignment.

Step 3: Rebuild PPTX

Scripts place the extracted layers back into PowerPoint according to coordinates, yielding a truly editable file where structure, elements, and text are modifiable while preserving the original visual fidelity.

What Problems It Solves

Pain point 1

AI‑generated content is beautiful but not editable; the project retains design quality and brings results back into an editable state.

Pain point 2

Manual recreation of complex layouts is time‑consuming; the tool automates “layer + position + assembly”.

Pain point 3

Content teams need both efficiency and deliverable files; AI handles creative layout while humans perform final tweaks.

Installation

pip3 install python-pptx pillow numpy

git clone https://github.com/GordenSun/GordenSuperPPTSkills
cd GordenSuperPPTSkills
cp -R GordenImagePPTGen "${CODEX_HOME:-$HOME/.codex}/skills/GordenImagePPTGen"
cp -R GordenImage2PPTX "${CODEX_HOME:-$HOME/.codex}/skills/GordenImage2PPTX"
cp -R GordenSuperPPTSkill "${CODEX_HOME:-$HOME/.codex}/skills/GordenSuperPPTSkill"

You can also copy the whole repository into the Codex skills directory.

Highlights

Not a demo but a complete workflow with strong engineering sense.

Separates visual quality from editability, tackling realism before editability.

Modular skill design enables independent use and reuse.

Extensive constraints on layers, coordinates, and QA ensure stable delivery.

Ideal for high‑quality office content that requires both aesthetics and editability.

Limitations

Currently limited to Codex environment because it depends on GPT’s image capabilities.

Converting each image consumes significant model quota; batch processing may be costly.

Quality depends on slide complexity; very fancy layouts may not be perfectly restored.

Final Takeaway

How can AI‑generated high‑quality visual drafts become collaborative, editable, deliverable PPT files?

GordenSuperPPTSkills combines AI image generation with structural reconstruction, offering a practical solution for creating beautiful yet editable presentations.

Reference

GitHub repository: https://github.com/GordenSun/GordenSuperPPTSkills