This Simple Trick Fixes NotebookLM Infographics

# This Simple Trick Fixes NotebookLM Infographics **TL;DR:** NotebookLM's "detailed" infographic mode often produces garbled text; download the infographic and feed it into ChatGPT with a simple prompt to fix all errors, repeating once if needed. ## What Is NotebookLM? NotebookLM is a Google tool where you create notebooks for different topics, add your own source materials from various file types, and then ask questions based strictly on that information. It cites sources accurately and lets you click to see the context. With good source material, it drastically reduces hallucination. The right-hand "Studio" panel offers ways to transform that information for visualization and memory: quizzes, mind maps, podcasts, data tables, video overviews, and, the focus here, **infographics**. One click generates an infographic based on your sources. ## The Problem: Detailed Mode Garbles Text You can choose between simple, standard, or detailed modes. Simple and standard modes are almost always perfect. Detailed mode, marked as "beta," often introduces spelling errors and scrambled text. Examples from the video: - A simple infographic on the Nano Banana showed "altitude" written as "aftitude" and the "S" in NASA incorrect. - A more complex infographic with multiple sources had errors like "Apollo" misspelled, "anomalies" wrong, numbers mixed up, "narrative reports" mangled, and entire sections of meaningless text. ## First Attempt: Canva’s Magic Layer Canva's "Magic Layer" feature can split a flat AI-generated design into fully editable layers. It works great for simple infographics – you can click into each text box and fix it. However, for complex infographics, it often discards text it can’t parse, changes fonts, and removes elements. Not a reliable fix for the detailed mode issues. ## The Simple Trick: ChatGPT The video’s solution is surprisingly straightforward: 1. Download the generated infographic image. 2. Drag it into ChatGPT (free version works fine with the same prompt). 3. Use a simple prompt: **"Generate this exact same infographic, but fix all text errors."** ChatGPT returns a corrected version. It tends to boost contrast slightly, but the text is fixed. For the simple Nano Banana example, "aftitude" became "altitude", NASA’s S was corrected. Everything else remained identical. For the complex infographic, the result was night and day. Where the original was a mess, ChatGPT produced a clean, meaningful infographic with no errors. Example fixes: - "Apollo lunar anomalies" instead of garbled text. - "Advanced" was initially misread, but after highlighting and asking to correct that one word, ChatGPT suggested "accelerates" – and the sentence made sense: *"An enhanced federal investment accelerates the fusion of AI for cross-domain triage."* ## The Catch: Requires a Second Pass Sometimes Occasionally, fixing one part reintroduces garbled text elsewhere. The better approach: **download the first ChatGPT result and run it through ChatGPT again with the same prompt.** The model seems better at generating a full corrected infographic than at editing a single element. After the second pass, the text is fully accurate. The video tested this on multiple styles: - **Comic style:** "cooling" misspelled, "massive" S issue – fixed in one pass. - **Complex infographic with many issues:** first pass fixed most, second pass perfect. - **Kawaii style and anime style:** all errors corrected after one or two passes. - **Infographic with meaningful text boxes:** after two passes, labels were corrected and properly described. ChatGPT’s reasoning often went deep, analyzing accuracy. For example, it noticed a sequence should mention 20 side chains instead of 30, corrected a folding paradox number (10 to the 300th power), and changed "million" to "structures" where it made more sense. ## Why Not Just Generate in ChatGPT From the Start? NotebookLM is still the preferred tool because it integrates all other features (sources, notebooks, quizzes, etc.). The tip is for users who already have their notebooks set up and want to keep working inside that ecosystem. This quick fix saves the hassle of transferring everything to ChatGPT. ## Summary - NotebookLM’s **detailed** infographics often have text errors. - **Canva Magic Layer** works only for simple designs. - **ChatGPT** (free or paid) reliably fixes all text using the prompt: *"Generate this exact same infographic, but fix all text errors."* - If errors remain, **download and run the result through ChatGPT a second time**. - This preserves NotebookLM’s workflow while eliminating the garbled text issue. **Source:** [This Simple Trick Fixes NotebookLM Infographics – Futurepedia](https://www.youtube.com/watch?v=psAQ7GYLmKw)