Generative AI Review

A thumbnail of the comparison… raw full version below Click for an enormous zoomable version: large 12560 x 3760px raw link For your viewing pleasure, side-by-side small multiples of all the OpenAI image models across all the quality levels.

Here are some of my prototype games made real with GenAI.

JUMP STRAIGHT TO THE SIDE BY SIDE COMPARISONS Here are screencap vids of the vibecoding from Claude Fable and versus Kimi K3.

Here are screencap vids of the vibecoding from ChatGPT 5.5, Opus 4.8, Claude Fable and now Zhipu.AI GLM5.2.

Here are screencap vids of the vibecoding from ChatGPT 5.5, Opus 4.8 and now Claude Fable.

Jump to ChatGPT 5.5 2026 Jump to Claude Opus 4.8 2026 Since my last Egyptology vibe coding it has been an AI Decade (one human year).

Here are my standardised images.

Table of Contents Overview The Source Image The Qwen Resulting Separations/Layers The POH Separations Created Manually Overview Qwen-Image-Layered is an AI tool which is given an image and splits out layers from that image.

I thought I’d post a piccy from a presentation I gave at PyData London in December 2025 @terry-lurie Giving overview of quality settings.

Overview Today I am going to take a handwritten poem and use generative AI tools to read the handwriting into computer text.