Every institution faces the same problem in the weeks before an exhibition opens. The work is in crates, the walls are unpainted, the lighting has not been focused and the press office needs images now, because coverage decisions are made before installation, not after.The standard solution is a single loaned image from the artist's studio, reproduced flat, accompanied by a press release. It is functional and almost entirely uncompelling, which is why so much exhibition announcement content vanishes into a feed without registering.A growing number of galleries and museums have started using image-to-video generation to build preview material from the imagery they already hold. The results, used carefully, are more effective than the flat reproduction and the phrase "used carefully" carries real weight in that sentence.

What the tools do with a still artwork

Image-to-video generation takes a still image as reference and produces motion from it. In an exhibition context, the useful applications are restrained: a slow drift across a large canvas that reveals its scale, a subtle parallax that separates foreground from ground in a sculpture photograph, light shifting across a surface the way it will in the gallery.Reference fidelity is what determines whether this is respectful or ruinous. Systems designed around strong reference handling stay close to the source. A Seedance 2.5 image to video workflow accepts up to 50 multimodal reference assets in a single generation and produces native 30-second single-clip output, with sharper rendering of close-up subjects which matters enormously when the subject is a painting whose surface quality is the point.Thirty continuous seconds is roughly the length of an attentive first look at a work. That is a useful unit for a preview.

Applications that work

The scale is revealed. Reproduction destroys scale. A nine-metre canvas and a postcard look identical on a phone screen. A slow camera move that establishes the work's size in relation to a space communicates something a flat image cannot, and it is often the single most surprising thing about seeing a work in person.
The surface study. Impasto, gilding, glaze, weave, corrosion. Material qualities are what reproduction flattens most severely, and a slow move across a surface under raking light restores some of it.
The archival document. Exhibition histories, artist's letters, working drawings, installation photographs from a previous decade. These are compelling and photograph poorly. Motion gives them presence in a feed.
The atmospheric teaser. A short piece that establishes the tone of an exhibition palette, era, mood without purporting to show the hang. Explicitly interpretive, and honest about it.

The line the sector cannot cross

This requires stating plainly, because the risk is not hypothetical and the consequences are reputational.Generated motion must never alter the artwork. Not the composition, not the colour, not the surface, not the framing of the object itself. Moving a camera across a static work is documentation. Making the figures in a painting move, extending a canvas beyond its edges, or altering a palette is misrepresentation of an object the institution holds in trust and no engagement metric justifies it.

Generated content must never imply an installation that does not exist. A preview showing works arranged in a gallery configuration that was never built creates an expectation the visitor will arrive to find unmet.Labelling has to be unambiguous and consistent. Not a hashtag, not a line in a caption three taps away. Institutions are in the business of authenticity; ambiguity about what is documentation and what is generated corrodes the thing that makes them worth trusting.And the artist's position governs everything. Living artists have moral rights and, more importantly, a legitimate interest in how their work is represented. Consent should be sought explicitly, in writing, for each use not assumed from a loan agreement drafted before these tools existed. Estates warrant the same care.

Practical notes for a small press office

Source imagery quality determines everything. Use the highest-resolution conservation or documentation photography available; the generation cannot supply detail the reference never had.Ask for less motion than the tool offers. Slow, minimal, almost imperceptible movement reads as considered. Dramatic camera work reads as a screensaver and cheapens the work.

Generate at the aspect ratio each channel needs rather than cropping. Cropping an artwork for a vertical feed is its own small act of vandalism; a considered vertical composition around the work is not.Draft at lower resolution to test pacing, then produce the final at 1080p. And have someone who knows the work a curator, not only the press officer approves every clip before it goes out.

Why it is worth doing at all

The case is not that generated previews are better than seeing the exhibition. Obviously they are not. The case is that they compete more effectively for attention in the only place most potential visitors will ever encounter the announcement, and that a preview which conveys scale and surface gives someone a more accurate reason to come than a flat JPEG does.
Used with the constraints above, image-to-video tools are a documentation aid rather than a creative liberty. Used without them, they are a fast route to an apology in the arts press and the institutions that get this right will be the ones that decided the rules before they made the first clip.