--- title: "eden zhang (@edenzhang) | Hashnode" description: "eden zhang (@edenzhang) writes on Hashnode. Read their latest articles and follow their work." url: "https://hashnode.com/@edenzhang" --- @edenzhang # eden zhang @edenzhang · Hefei, China · Joined August 2026 ### [Canvas as a Codec: Transcoding WebP and AVIF to PNG Right in the Browser](https://omnipic-studio.hashnode.dev/canvas-as-a-codec-transcoding-webp-and-avif-to-png-right-in-the-browser?utm_source=hashnode&utm_medium=feed) Every designer knows the pain. You browse a beautiful portfolio, right-click a stunning image, and your file lands as photo.avif — which your asset manager, your client's PowerPoint, and half the web 6 min read · [omnipic-studio.hashnode.dev](https://omnipic-studio.hashnode.dev/canvas-as-a-codec-transcoding-webp-and-avif-to-png-right-in-the-browser?utm_source=hashnode&utm_medium=feed) [Oct 26](https://hashnode.com/posts/canvas-as-a-codec-transcoding-webp-and-avif-to-png-right-in-the-browser/6adebeb595c18396145270de) ### [CORS, CSP, and Referrers: The Hidden Network Layer Between You and Every Image on the Web](https://omnipic-studio.hashnode.dev/cors-csp-and-referrers-the-hidden-network-layer-between-you-and-every-image-on-the-web?utm_source=hashnode&utm_medium=feed) If you've ever built a "download all images" tool — a bookmarklet, a script, a browser extension — you know the sequence: walk the DOM, collect URLs, fetch, save. And then you hit the wall. The URLs w 6 min read · [omnipic-studio.hashnode.dev](https://omnipic-studio.hashnode.dev/cors-csp-and-referrers-the-hidden-network-layer-between-you-and-every-image-on-the-web?utm_source=hashnode&utm_medium=feed) [Oct 25](https://hashnode.com/posts/cors-csp-and-referrers-the-hidden-network-layer-between-you-and-every-image-on-the-web/6add656f4578e9414191bc41) ### [Beyond \`\`: How Crawlers Discover Every Real Image on a Modern Web Page](https://omnipic-studio.hashnode.dev/beyond-img-src-how-crawlers-discover-every-real-image-on-a-modern-web-page?utm_source=hashnode&utm_medium=feed) This post digs into how contemporary web pages hide images — lazy loading, responsive \`srcset\`, data-attribute placeholders — and how a crawler (browser extension or otherwise) can still enumerate eve 5 min read · [omnipic-studio.hashnode.dev](https://omnipic-studio.hashnode.dev/beyond-img-src-how-crawlers-discover-every-real-image-on-a-modern-web-page?utm_source=hashnode&utm_medium=feed) [Oct 19](https://hashnode.com/posts/beyond-img-src-how-crawlers-discover-every-real-image-on-a-modern-web-page/6ad5820a29e19db93dbe9b26) ### [How Local Image Libraries Really Work: File Layout, Metadata Sidecars, and the Export Pipeline](https://omnipic-studio.hashnode.dev/how-local-image-libraries-really-work-file-layout-metadata-sidecars-and-the-export-pipeline?utm_source=hashnode&utm_medium=feed) If you collect images professionally — for design, ecommerce, or content — browser bookmarks stop working after a few hundred items. Tabs get closed, CDN links rot, thumbnails expire, and a "saved" im 4 min read · [omnipic-studio.hashnode.dev](https://omnipic-studio.hashnode.dev/how-local-image-libraries-really-work-file-layout-metadata-sidecars-and-the-export-pipeline?utm_source=hashnode&utm_medium=feed) [Oct 18](https://hashnode.com/posts/how-local-image-libraries-really-work-file-layout-metadata-sidecars-and-the-export-pipeline/6ad429e68bdbbc8157d2b1d1) ### [How Reverse Image Search Actually Works: 1024-Dimensional Embeddings and Cosine Similarity](https://omnipic-studio.hashnode.dev/how-reverse-image-search-actually-works-1024-dimensional-embeddings-and-cosine-similarity?utm_source=hashnode&utm_medium=feed) If you've ever clicked an image and asked "find me more like this," you used a reverse image search. Under the hood, it has nothing to do with filenames or keywords. An image has no natural text token 5 min read · [omnipic-studio.hashnode.dev](https://omnipic-studio.hashnode.dev/how-reverse-image-search-actually-works-1024-dimensional-embeddings-and-cosine-similarity?utm_source=hashnode&utm_medium=feed) [Oct 16](https://hashnode.com/posts/how-reverse-image-search-actually-works-1024-dimensional-embeddings-and-cosine-similarity/6ad18960ac35882edafef651)