For most of art history, the still image and the moving image have lived in separate rooms. Painters painted, animators animated, and the people who did both were rare enough to be remembered for it. Over the past year that wall has become noticeably thinner. A growing number of painters, illustrators and printmakers are taking their own finished works and turning them into short moving pieces, using image-to-video software that animates a single frame.

The results range from gimmicky to genuinely interesting, and the debate around them is more nuanced than the usual argument about artificial intelligence and art. This is not, for the most part, about machines making pictures. It is about artists deciding what happens when their own picture starts to move.

Why artists are doing it

The first reason is unromantic: attention. Exhibition announcements, gallery social accounts and online portfolios now live on platforms that reward motion. A still image of a painting scrolls past; three seconds of the same painting with light slowly shifting across the canvas stops the thumb. Several gallerists describe the animated versions as the digital equivalent of a good invitation card, a way of getting people through the door, not a substitute for the work.

The second reason is more interesting. Some artists find that animating a piece shows them something about it. A landscape painter who animated her own coastal studies described watching the generated clip as being like seeing the painting through someone else's eyes: the software chose what should move, and its choices revealed which elements of the composition were carrying weight and which were merely decorative.

The third reason is commercial. Illustrators working for publishers, game studios and advertising clients are increasingly asked for motion versions of still assets. Learning to animate a single illustration convincingly used to require a separate skill set. Now an illustrator can deliver a short motion study alongside the still and let the client decide whether a full animation budget is justified.

How the process actually works

The tools take a single image and generate a short clip, typically five to ten seconds, guided by a written description of the motion. The artist supplies the picture; the software infers depth, lighting and plausible movement, and produces a sequence of frames that begins from the original.

In practice, the craft lies in restraint. The most effective animated paintings move very little: a drift of smoke, a slow change of light, water that breathes rather than crashes. Artists who ask for dramatic motion tend to get footage that looks like a stock video wearing their painting as a costume. Those who ask for something small, specific and in keeping with the original mood get pieces that feel like an extension of the work.

Most artists generate many variations and keep few. A ratio of ten attempts to one keeper is commonly reported, and the editing, choosing which of the ten respects the painting, is where the artist's judgement comes back in.

The authorship question

This is where the conversation becomes difficult, and it should. When a painter animates her own canvas, who made the moving piece?

There are roughly three positions among practitioners. The first holds that the animation is simply documentation, the same category as a well-lit photograph of the work, and claims no artistic status for it. The second treats it as a new, collaborative object: the composition and intent are the artist's, the motion is the software's, and both should be credited. The third, held mostly by artists who have refused to use the tools, argues that handing motion decisions to a model trained on other people's work compromises the piece, however small the movement.

A practical consensus seems to be forming around disclosure. Artists who use these tools increasingly label the moving versions as such, keep the original still as the primary work, and do not offer the animations as editioned artworks. Several galleries have adopted the same line: animated versions may promote a show, but they are not sold as part of it.

What it does badly

Honesty about the limitations matters, because they shape where the technique is worth using.

Texture. Visible brushwork, impasto and the grain of paper are exactly what these models smooth away when they start moving things. A heavily textured oil painting often comes back looking like a digital print of itself. Flatter styles, such as gouache, ink and digital illustration, survive better.

Figures. People in paintings rarely animate well. Faces drift, hands multiply, and the stillness that made a portrait powerful is replaced by an uneasy fidget.

Style drift. Over a longer clip, the model tends to pull the image towards its own idea of what pictures look like. The final frame can be noticeably less like the painting than the first.

The practical side

Most artists start with consumer apps, which is reasonable for experiments. Studios and illustration agencies that produce motion versions of client work at volume tend to move to tools that call the underlying models directly, because they need consistent settings, file naming and the ability to batch a whole series. Pricing for video generation is normally per second of output rather than per image, so a series of twenty animated works with ten attempts each adds up faster than people expect. For those comparing options, image-to-video fidelity matters more than anything else, since the whole point is to respect the source; the published rates and capabilities of the Kling AI API on APIMart and other current video models give a realistic sense of what a series costs before committing.

What the better examples have in common

Looking across the animated works that have drawn genuine praise rather than polite tolerance, a few patterns stand out.

They are short. Most of the successful pieces last between four and eight seconds and loop cleanly, which keeps the viewer's attention on the composition rather than on the motion. Once a clip runs longer, the eye starts looking for the moment the illusion breaks.

They respect the original palette. Artists who reject any variation where the colour temperature drifts, even slightly, end up with pieces that feel like the painting rather than an interpretation of it. This often means discarding technically impressive clips because the shadows have turned a shade too blue.

They move what the painting already implied. A sky painted with visible wind in the clouds can drift; a still life arranged around a single candle can flicker. When the motion grows out of something the artist already put on the canvas, the animation reads as an extension of intent. When it introduces movement the painting never suggested, it reads as an effect.

And they are presented alongside the original. Galleries that show the still and the moving version side by side, rather than replacing one with the other, report that viewers spend longer with both. The comparison itself becomes part of the experience, a small lesson in how the eye reads a composition.

A reasonable place to land

The animated painting is not going to replace the painting, and the artists using these tools well are the first to say so. What it offers is a new kind of companion piece: a way to promote, to study and occasionally to extend a still work, with clear limits on what it can do and an obligation to be transparent about how it was made.

Whether that is an addition to an artist's practice or a distraction from it is a question each artist will answer differently. What has changed is that the question is now available to anyone with a finished canvas and an afternoon, rather than only to those who already knew how to animate.