wtf is naqal?

naqal is a small collection of prompts for reverse engineering the visual style of any video and then recreating that style yourself. The author started after watching a striking product launch video and decided to break it down shot by shot: the pacing, lighting, color grading, sound design, and typography. The repository holds a written workflow rather than code. One prompt walks a frame by frame analysis of a reference video, capturing its visual language and structure into a description. A second prompt uses that description to generate still images in the same style. A third prompt turns those generated images into a video clip. Put together, the three prompts let you take any video you admire, extract the pattern behind it, and produce your own variant of it using image and video generation tools. The method is not tied to any single AI model. You can use whatever image generator or video generator you already have access to, such as Midjourney, Flux, Kling, Runway, Luma, or Pika, and the quality of the result depends on that tool rather than on naqal itself. What stays the same across tools is the four step process: study a reference, encode what makes it distinctive, generate new assets, and assemble them into a variant. The files included are image-to-video.md for turning images into clips, video-analysis.md for the full frame by frame breakdown, a lighter video-analysis-lite.md version for quicker runs, and a References folder with example analyses and prompt sets from videos already broken down. Usage is straightforward: pick a video, run the analysis prompts on it, generate matching images, feed them into the video prompt, and repeat until the result feels right. This is a very new project with a single star and no license information given.