Why Do Museum Images Look Low-Resolution? (And How to Get the Real File)
That 800-pixel JPEG isn't the real image — it's a thumbnail of a scan that may run to 40 megapixels. Here's where the full file lives and how to reach it.
Quick answers
- Why is the museum image on the webpage so small?
- Collection pages serve small derivative copies so that pages load quickly. The underlying scan is usually many times larger and is delivered separately through a download link or an IIIF image server, so the displayed image is a preview rather than the archival file.
- What is IIIF and why does it matter for downloads?
- IIIF is the International Image Interoperability Framework, a standard museums use to serve images as zoomable tiles. Because a IIIF endpoint can render any region at any size on request, it makes full-resolution regions of a scan available rather than only a fixed thumbnail.
- Can I just upscale a small museum image?
- Upscaling invents detail that was never captured, and at print size the result looks soft and artificial. Since the original high-resolution file is usually available at no cost, finding it is almost always faster than trying to repair a small one.
- How do I find the largest available version of a museum image?
- Look for an explicit download control on the object record, check whether the museum offers a IIIF manifest for the object, and prefer any option labelled original, archival, or maximum size. On The Met, the API field primaryImage is the full-size file while primaryImageSmall is the preview.
You find the painting. You right-click, save, open it — and it's 843 pixels wide. Useless for anything but a slide.
The scan behind it is probably thirty to forty times that size. Museums photograph works at archival resolution; what a collection page shows you is a derivative built for page-load speed. The full file is almost always free and almost always one layer away. Here's where it hides.
Why the page serves you a small one
Not obstruction — plumbing. A museum with a million objects cannot ship 60-megabyte TIFFs to every browser that loads a search grid, so the pipeline generates derivatives: a thumbnail for grids, a mid-size for object pages, and a full version served on request. Three practical consequences:
- What you see is never the archival file.
- The download route is separate from the display route.
- On museums without an open-access programme, the full version may be deliberately withheld — a rights decision wearing a technical costume.
Route one: the download control
The obvious route, and it works where open access does. On an open-access object page there is a download or share control near the image that links to the full file. Two habits:
- Take the largest option offered. "Large" on a museum site often still isn't the original — look for "original", "archival", or "maximum".
- If no control exists, that usually means restricted rights rather than a missing file. Check the rights statement before you look for workarounds — How to Tell If an Artwork Is in the Public Domain.
Route two: IIIF, the standard most people have never heard of
IIIF — the International Image Interoperability Framework — is how the cultural-heritage world serves images, and it changes what "the image" even means. Instead of a fixed JPEG, a IIIF server renders any region at any size on demand, which is what makes smooth deep-zoom possible.
The request pattern is readable by design:
- `{server}/{identifier}/{region}/{size}/{rotation}/{quality}.{format}`
So `.../full/max/0/default.jpg` asks for the whole image at the largest size the server will give (IIIF Image API 3.0; the 2.x spelling is `/full/full/0/default.jpg`). And because region is a parameter, you can request one quadrant of a forty-megapixel scan at full pixel density — a crop taken at native resolution, not a crop of a downscale.
An object's manifest — a JSON file listing its images and metadata — is the entry point, and museums that support IIIF usually link it from the object page. This is also the machinery under the deep-zoom viewer you've used without thinking about it.
Route three: the API
Where there's a public API, the full file is usually a named field. The Met's object endpoint is the clearest example: `primaryImage` is the full-size file and `primaryImageSmall` is the preview, with `additionalImages` holding alternate views — versos, details, frames — that the web page may not surface at all.
The Smithsonian's API works similarly through a free api.data.gov key, which is also what enables it as a source inside Musist. See How to Explore the Smithsonian Open Access Collection.
How big is big enough?
Work backwards from the output:
- Screen and social: anything over 2,000 px on the long edge.
- Book cover or magazine page: 3,000–4,000 px.
- A3 poster at 300 dpi: about 3,500 × 4,900 px.
- Large wall print at 150 dpi: a 24 × 36 inch piece needs roughly 3,600 × 5,400 px.
The Rijksmuseum routinely clears all of these with room to spare; the Met's Open Access files are large; the Smithsonian varies by unit. Downloading Public-Domain Images from the Rijksmuseum has the specifics.
Don't upscale
An AI upscaler invents detail that the camera never recorded. At thumbnail size it's convincing; at print size it reads as smeared, plasticky, and subtly wrong — and on a painting it destroys exactly the thing you wanted, the surface. Since the real file costs nothing, ten minutes of looking beats any amount of reconstruction.
The shortcut
Musist was built around this problem. Every object page renders the source's IIIF tiles as deep-zoom, so you can inspect a work at full magnification in the browser before deciding it's the one — and the Download image button appears only when the work is public domain and a full-resolution file genuinely exists behind it.
That last part is the useful bit: the button's presence is a one-glance answer to "is there a real file here?" across The Met, the Rijksmuseum, and the Smithsonian at once. Browse by source, or between related works and check resolution as you go.
- download
- how to
- resolution
- iiif