Chitrashala: a museum-grade open archive for Raja Ravi Varma

I went looking for a good place to look at Raja Ravi Varma's paintings. I could not find one.

His most important images are scattered. A high scan on one museum site. A washed-out thumbnail somewhere else. The same goddess reproduced ten different ways, rarely catalogued, almost never built for close looking.

So I built the place I wanted. chitrashala.vayuai.ai

Chitrashala is Sanskrit for picture gallery, a hall of images. That is what this is: 395 of Varma's paintings and Ravi Varma Press oleographs in high resolution, each one catalogued with its subject, its place in myth, and why it matters. Everything is public domain. Everything is free to explore and to download.

Who Raja Ravi Varma was, and why this matters

Raja Ravi Varma lived from 1848 to 1906.

He did something that sounds small and was not: he took European academic oil technique, the kind taught in the salons, and turned it on Indian subjects. Gods, goddesses, queens, and scenes from the epics, painted with the modelling, depth, and drapery of Western academic art.

That fusion was the achievement. But it is not why most people have seen his work.

The reason is the Ravi Varma Press. He set up a lithographic press and mass-produced his paintings as oleographs, cheap color lithographs that could hang in an ordinary home. Those prints traveled further than any original could.

For more than a century, when people in India pictured Lakshmi or Saraswati or Rama, they were often picturing a Ravi Varma composition.

His paintings became the popular face of those figures. Calendars, shrine prints, matchbox labels, all of it traces back to his compositions. Few painters anywhere have shaped a culture's mental image of its own gods that directly.

His work is in the public domain. It belongs to everyone. It just was not easy for anyone to actually use.

What I built

One open, fast, quiet place to see the work properly.

Not a slideshow. Not a board of low-resolution rephotographs. A catalogue, where each work is more than a filename and you can lean all the way in until you are looking at the grain of the scan.

Three things had to be true for it to be worth doing.

It had to be open: public domain throughout, free to view, free to download, with the data reusable by anyone.

It had to be close: deep zoom, so you can study brushwork and the texture of an oleograph, not just admire a thumbnail.

It had to be catalogued: every work carrying its subject, its myth, and its significance, so the archive teaches and does not just display.

How the sourcing works

The scans come from open cultural sources.

Principally Wikimedia Commons and the Wellcome Collection, both of which hold high-resolution public-domain reproductions of Varma's paintings and Ravi Varma Press oleographs.

Gathering the files is the easy half. A bare scan is a picture with no memory: no subject, no story, no reason to stop on it.

So each work is enriched with catalogue context. What it depicts. Where it sits in the myth. Why it is significant. That layer is what turns a folder of images into something closer to a gallery wall with labels.

Attributions follow the source institutions. Where a museum holds the original or made the scan, that is where the credit points.

The image pipeline

This is the part that makes the front end feel calm.

A source scan is one large file, often very large. You cannot ship that to a browser and ask it to feel fast. So everything passes through a pipeline built on the sharp image library, which derives exactly the sizes and formats the gallery needs from each original.

The masonry grid loads small, sharp thumbnails. Open a work and you get a gallery-sized image. Zoom in and the viewer serves a high-resolution image up to 2560 pixels so you can go to pixel level on the scan.

There is a second path, already built. Tiled deep zoom, where the viewer loads only the tiles for the region you are actually looking at, the way a maps app loads only the streets on your screen. It means you could zoom into a very large scan and the browser would never download the whole thing, only the patch under your cursor.

That path is staged, not live. It switches on once the media moves to object storage (Cloudflare R2). I will come back to that in the honest-limits section, because I would rather tell you where the boundary is than imply the whole thing is finished.

The gallery

The front end is React and Tailwind, in the VayuAI house style: warm-paper light theme, Fraunces for display and Inter for text, lucide line icons, no casual emoji. Museum-quiet. The design gets out of the way of the paintings.

What is in it:

• a masonry gallery that lets the work breathe at different aspect ratios instead of forcing everything into one grid;
• full-text search across the catalogue, so you can find a work by subject or figure, not just by guessing a filename;
• curated collections, groupings that make the archive wanderable instead of a flat wall of 395 tiles;
• related works, so one image leads to its neighbors in subject and series;
• an artist timeline, placing the work across his life from 1848 to 1906;
• a deep-zoom viewer that takes you to pixel level on the scan.

The whole thing is built for one gesture: open something, look closely, follow it sideways to the next thing.

Open by design

The catalogue is not locked inside one site.

There is a JSON API. The same data that drives the gallery, subjects, myth, significance, the catalogue structure, is available as structured data so others can build on it. A teacher, a researcher, another developer, anyone who wants the catalogue can take it.

This matters to me as a principle. Public-domain heritage that lives inside one closed product is only half free. If the work belongs to everyone, the data describing it should be reusable by everyone too.

Open is not a license badge in the footer. It is whether someone else can actually pick the thing up and carry it somewhere new.

An honest note on the limits

I do not think a useful archive needs to pretend it is the final word. Credibility comes from saying plainly what it is and what it is not.

It is a curated 395-work archive, not the complete catalogue raisonne. Varma was prolific, and the press produced more oleographs than any single collection holds. This is a strong, representative selection, gathered and catalogued with care. It is not exhaustive.

Scan quality varies with the source. These are reproductions from open collections, made at different times, at different resolutions, by different institutions. Some are superb. Some are merely good. I did not fabricate detail that the source scan does not contain.

The tiled deep zoom is staged behind the single-image zoom until the media moves to object storage. Today you get a genuine high-resolution zoom up to 2560 pixels, which is enough to study most of these works closely. The tile-by-tile path is built and waiting on the storage move, not vaporware and not yet flipped on.

Attributions follow the source institutions. I credit where the scans come from. If a museum holds the original and made the reproduction, the trail points there.

That is the boundary. I would rather you know exactly where it is.

Where it goes next

The clearest next step is the storage move.

Once the media lives in object storage (Cloudflare R2), the tiled deep zoom switches on, and close looking gets cheaper and faster for everyone, especially on the very largest scans.

After that, the obvious direction is more. More works as better public-domain scans surface. Deeper catalogue context on each piece. Richer collections that follow a single goddess or a single epic across the oleographs and the originals.

But the shape stays the same. This is not trying to become a marketplace or a feed. It is trying to be the place I could not find: a quiet, fast, open hall where Raja Ravi Varma's work can be seen properly and taken freely.

Preservation and access are the whole point. Making public-domain heritage genuinely usable, not just technically free.

The work already belongs to everyone. My job was only to make it possible to actually see.

Open Chitrashala →