--- title: "Henry Jin (@videoall) | Hashnode" description: "AI Video Generation. Read the latest articles by Henry Jin on Hashnode." url: "https://hashnode.com/@videoall" --- @videoall # Henry Jin @videoall · Joined September 2026 AI Video Generation ### [Why I stopped asking one AI video model for 30 seconds in a single shot](https://videoall.hashnode.dev/why-i-stopped-asking-one-ai-video-model-for-30-seconds-in-a-single-shot?utm_source=hashnode&utm_medium=feed) Most AI video models top out near 10 seconds. How I split a 45-second demo into six anchored shots, and what still breaks. 4 min read · [videoall.hashnode.dev](https://videoall.hashnode.dev/why-i-stopped-asking-one-ai-video-model-for-30-seconds-in-a-single-shot?utm_source=hashnode&utm_medium=feed) [4d ago](https://hashnode.com/posts/why-i-stopped-asking-one-ai-video-model-for-30-seconds-in-a-single-shot/6aefebf9d25976eb3d70c768) ### [Text-to-Video vs Image-to-Video: How I Pick the Right Input Before Writing a Prompt](https://videoall.hashnode.dev/text-to-video-vs-image-to-video-how-i-pick-the-right-input-before-writing-a-prompt?utm_source=hashnode&utm_medium=feed) A hands-on guide to choosing between text-to-video and image-to-video, with real prompts, render parameters, and a checklist for consistent AI clips. 9 min read · [videoall.hashnode.dev](https://videoall.hashnode.dev/text-to-video-vs-image-to-video-how-i-pick-the-right-input-before-writing-a-prompt?utm_source=hashnode&utm_medium=feed) [Oct 29](https://hashnode.com/posts/text-to-video-vs-image-to-video-how-i-pick-the-right-input-before-writing-a-prompt/6ae2bfe3d19a03a1f3abd909) ### [What Telegram's compression really does to AI video (and the settings I use)](https://videoall.hashnode.dev/what-telegram-s-compression-really-does-to-ai-video-and-the-settings-i-use?utm_source=hashnode&utm_medium=feed) Why AI video looks mushy on Telegram, and the export settings I use to fix it — tested on real channel posts over a month. 4 min read · [videoall.hashnode.dev](https://videoall.hashnode.dev/what-telegram-s-compression-really-does-to-ai-video-and-the-settings-i-use?utm_source=hashnode&utm_medium=feed) [Oct 28](https://hashnode.com/posts/what-telegram-s-compression-really-does-to-ai-video-and-the-settings-i-use/6ae203d3a2cb567af4c24785) ### [Prompt Engineering an Anime Opening: Deconstructing the Shot List Behind an OP](https://videoall.hashnode.dev/prompt-engineering-an-anime-opening-deconstructing-the-shot-list-behind-an-op?utm_source=hashnode&utm_medium=feed) An anime opening isn't one shot - it's a compressed sequence of hooks, reveals, and rhythm changes. When you move that sequence to an AI video model, your prompt stops being a sentence and starts bein 3 min read · [videoall.hashnode.dev](https://videoall.hashnode.dev/prompt-engineering-an-anime-opening-deconstructing-the-shot-list-behind-an-op?utm_source=hashnode&utm_medium=feed) [Oct 25](https://hashnode.com/posts/prompt-engineering-an-anime-opening-deconstructing-the-shot-list-behind-an-op/6addc967d00c9b16a1b91478) ### [Deconstructing a GPT Image 2.5 Reference-Led Restyle Prompt, Clause by Clause](https://videoall.hashnode.dev/deconstructing-a-gpt-image-2-5-reference-led-restyle-prompt-clause-by-clause?utm_source=hashnode&utm_medium=feed) Reference-led editing is the workflow where a model takes an uploaded image as a reference and regenerates it with changes described in a prompt. The hard part of writing the prompt is that it has to 5 min read · [videoall.hashnode.dev](https://videoall.hashnode.dev/deconstructing-a-gpt-image-2-5-reference-led-restyle-prompt-clause-by-clause?utm_source=hashnode&utm_medium=feed) [Oct 12](https://hashnode.com/posts/deconstructing-a-gpt-image-2-5-reference-led-restyle-prompt-clause-by-clause/6acc6046ca94148640c7dd5a)