Handsum offers a new LQIP format to improve on BlurHash's quality-per-byte

The new Handsum file format claims superior image placeholders by using a DCT-based approach, but its adoption depends entirely on an ecosystem that doesn't yet exist.

The Answer Up Front

Handsum is for web developers obsessed with image loading performance who want visually superior placeholders and are willing to accept the costs of a new, unsupported format. If you prioritize stability, broad library support, and minimal JavaScript overhead, you should stick with BlurHash. The bottom line is that Handsum presents a technically compelling alternative for low-quality image placeholders, but its practical utility is severely limited by the classic chicken-and-egg problem of format adoption. It solves the image quality problem while creating a new ecosystem problem.

Methodology

This is a v0 review based exclusively on the creator's technical blog post announcing the format. Update cadence: this review will be updated if and when independent benchmarks or widespread adoption change the calculus.

• Tool: Handsum File Format
• Version: As described in the July 2026 blog post.
• Source: Nigel Tao, "Handsum: An LQIP Image File Format", published at https://nigeltao.github.io/blog/2026/handsum.html on July 11, 2026.

This review covers the technical design, stated goals, and the creator's own comparison examples as presented in the source article. What is not covered is any independent performance testing, the real-world bundle size impact of its JavaScript decoder, or its behavior on a wide corpus of images beyond the curated examples. All performance and quality comparisons are the creator's claims and have not been independently verified.

What It Does

Handsum is a file format, specified with a .hsm extension, for creating low-quality image placeholders (LQIPs). Its goal is to provide a better-looking preview than alternatives like BlurHash at a comparable or smaller file size.

A file format, not a string

Unlike BlurHash, which produces a short ASCII string, Handsum is a binary file format. The creator, Nigel Tao, argues this structured approach allows for more sophisticated compression and features. The format is built around a Discrete Cosine Transform (DCT), the same mathematical foundation used in JPEG compression. This allows it to store image information more efficiently, particularly for photographic content.

Better compression and quality

The central claim is that Handsum provides better visual fidelity per byte. The announcement post includes several side-by-side comparisons against BlurHash. In these examples, Handsum appears to avoid the color banding and blocky artifacts that can affect BlurHash, especially in images with subtle gradients or detailed textures. The DCT approach is better suited for representing the low-frequency color and light information that makes for a good placeholder.

Designed for the web

Handsum is intended to be decoded client-side via JavaScript or, potentially, WebAssembly. The creator provides a reference decoder. The workflow is similar to other LQIP techniques: generate the .hsm file at build time, embed it or fetch it with the initial page load, and render it to a canvas until the full-resolution image arrives.

What's Interesting / What's Not

The technical approach is sound. Using DCT is a proven method for image compression, and applying it to the specific problem of placeholders is a logical step. The visual evidence presented in the blog post, while curated, is compelling. Handsum placeholders look smoother and more representative of the final image than their BlurHash counterparts. For teams fighting for every millisecond of perceived performance and every point on Core Web Vitals, a higher-quality placeholder can make a meaningful difference to the user experience.

The primary obstacle is not technical merit but ecosystem inertia. A new image format is a heavy lift. BlurHash is successful not just because it works well enough, but because it has dozens of mature libraries across every major language and framework. You can generate and render BlurHashes in Ruby, Python, Go, Rust, Swift, and JavaScript with a single command. Handsum, by contrast, launches with a reference decoder and nothing else. Adopting it means taking on the maintenance burden of that decoder and integrating it into your stack manually. This also means adding more JavaScript to your initial page load, which can counteract the very performance benefits you're seeking.

Pricing

Handsum is an open-source file format and the reference implementation is provided under a permissive license. It is free to use.

Pricing snapshot taken July 11, 2026.

Verdict

Handsum is a technically impressive piece of engineering that produces visually superior image placeholders compared to BlurHash in its author's own examples. If you are a performance-focused frontend developer, and your project's success hinges on the absolute best perceived load time, and you are willing to absorb the integration and bundle-size cost of a new JS dependency, then Handsum is worth investigating. For most teams, however, the value proposition is not yet clear. The vast ecosystem of tools supporting BlurHash makes it the pragmatic, stable choice. Handsum is a better tool without a workshop.

What We'd Test Next

A v2 review would require independent benchmarking. First, we would measure the file size and visual quality of Handsum versus BlurHash and other techniques across a large, diverse image set (photographs, vector art, screenshots). Second, we would measure the performance of the JavaScript decoder itself. We need to know its gzipped bundle size, its parse time, and the CPU cost to decode and render an image on a low-to-mid-range mobile device. An LQIP that blocks the main thread is a performance regression, not an improvement.

The investor read

Handsum is a feature, not a company. The market for image formats is defined by massive inertia and the requirement of native browser or large library support to succeed. While technically interesting, Handsum is unlikely to displace established placeholder techniques like BlurHash on its own. It signals continued developer focus on Core Web Vitals and perceived performance, a market that is very real. The investment opportunity isn't in a new format itself, but in higher-level image optimization CDNs and services (e.g., Cloudinary, Imgix) that could eventually adopt Handsum as one of many optimization techniques. For now, this is a passion project and a signal of where developer attention is, not an investable asset.

Pull quote: “The primary obstacle is not technical merit but ecosystem inertia.”