Real video or video generated from a photo? How to choose for Extras

A video can make an Extra easier to notice and to imagine.

That doesn't mean every Extra should become a video, nor that AI-generated content is always preferable to a real clip.

Sometimes you need to show an authentic gesture. Other times it's enough to add a light movement to a good photograph. For simple, operational offers, a static image can remain the clearest solution.

The right choice doesn't start from the tool available. It starts from a question: what does the guest need to understand or perceive when looking at this Extra?

That answer decides whether to use a real video uploaded by the property owner, a video generated from a photograph, or just the Extra's image.

Three formats, three different jobs

Photograph, real video and generated video are not better or worse versions of the same content. They do different jobs.

A photograph shows a moment precisely. A real video demonstrates how an experience actually unfolds. A generated video can turn a static scene into more dynamic content without requiring a new shoot.

The mistake would be using video only because it is technically possible to produce one. The content should help the guest evaluate the offer, not distract them from it.

When to choose a real video

A real video is recommended when the Extra's value depends on something that actually happens. It is particularly useful to show:

• a person preparing or serving something;
• a treatment;
• a tasting;
• a guided experience;
• a class or sports activity;
• a decoration or setup during preparation;
• a sequence of gestures hard to represent with a single photo.

For example, a real video can show how breakfast is prepared, how a massage unfolds or which atmosphere emerges during a tasting. In these cases movement isn't just about catching the eye: it documents the experience.

The advantage of authenticity

A clip shot at the property shows real people, real spaces, actually available equipment, gestures truly performed and proportions hard to misread. This can matter when the service depends on a person's presence, a preparation or an interaction.

The real video remains the property owner's responsibility. Before using it the owner should check that quality is good enough, that the format is suitable for mobile viewing, that recognisable people have authorised the use, that no personal or sensitive data is visible and that the clip represents the service as it is actually delivered.

An authentic but confusing, overlong or badly recorded video isn't automatically more effective than a good photograph.

When to generate a video from a photo

A generated video becomes interesting when the owner already has a good photograph but no usable footage. It's a useful option especially when you want to introduce a camera push-in, a light pan, depth in the scene, a small gesture, an atmospheric movement or extra visibility in the showcase.

The starting point still needs to be a good image. AI does not automatically fix a blurry photo, an unreadable environment, a wrong framing or an ambiguously represented Extra. A generated video can add movement; it shouldn't be used to invent what's missing.

Pika 2.2 for people, gestures and actions

In StayFast, Pika 2.2 is recommended for photos containing people, gestures, treatments, ongoing activities and actions that need to develop over time. The owner can pick 5 seconds for a short movement or 10 seconds for a more complete gesture. Ten seconds is the default because it gives more room for an action to start and finish.

This doesn't mean every photo with people will be accepted or produce a natural result. The provider can apply its own moderation rules and may refuse some content. Even when the video is generated, the owner should carefully review hands, faces, movement, gesture continuity, objects used and consistency with the original image.

For a massage, for instance, the result should show a slow, believable movement without altering the people or the environment. When that fidelity is essential, a real clip may still be preferable.

Seedance 2.0 Fast for spaces, products and atmosphere

Seedance 2.0 Fast is recommended especially for rooms, terraces, set tables, products, details, wellness spaces, romantic setups, baskets, bottles or in-room amenities. Available durations are 5, 7, 10 and 15 seconds, with 7 seconds as the recommended value.

For this kind of content movement can mostly involve the camera. A slow push toward a breakfast, a pan over a terrace or a discreet movement around a product can make the scene feel more alive without changing its substance.

A longer duration is not automatically better. Fifteen seconds may make sense for a space that calls for slow, continuous movement. For a table or a single product, seven seconds can be more than enough.

When to keep just the photograph

Some Extras don't gain much from video. This is especially true when the offer is simple, functional or without a significant visual component: late checkout, early check-in, parking, transfer, luggage storage, extra cleaning, simple equipment rental or laundry service.

In these cases the guest mostly needs to know what is offered, when it's available, how much it costs, how to request it and which conditions apply. A clear photograph, paired with an accurate title, price and description, can work better than a generic animation.

Movement shouldn't become an editorial obligation. A showcase built on photographs and a few selected videos can be more readable than a screen where every element tries to grab attention at once.

The choice depends on what needs to be demonstrated

A useful criterion is to distinguish between demonstrating, evoking and informing.

A real video is recommended when you need to demonstrate how a service is performed, the real presence of a person, a preparation, a sequence, an activity or the interaction with the guest.

A generated video can be suitable when you want to evoke atmosphere, depth, relaxation, care, elegance, desirability or attention around a product or setup.

A photograph can be enough when the decision depends mostly on price, timing, availability, rules, practical use or how to request.

A few concrete examples

In-room massage

Real video: the most reliable choice when a good clip of the treatment and the required permissions exist. Generated video: Pika can animate a photo with people already present, requesting a slow, natural gesture. Photo: still valid when you don't want to process the people's faces or bodies.

Romantic breakfast

Real video: useful to show the preparation or the service. Generated video: Seedance can create a slow camera push toward the table. Photo: enough if the setup is already clear and price is the main deciding factor.

Tasting

Real video: preferable to show the service moment, the toast or the products' presentation. Generated video: can enhance a static composition of bottles, glasses and local products. Photo: suitable when the offer mostly needs to show what is included.

Late checkout

Real video: rarely needed. Generated video: may feel artificial or uninformative. Photo: generally enough, together with timing, availability and price.

Romantic in-room setup

Real video: useful when you want to document the preparation. Generated video: Seedance can add a delicate camera movement while keeping bed, decorations and lights intact. Photo: can already communicate everything, especially when the setup is careful and recognisable.

Local experience or tour

Real video: often the best choice, because it shows a real place, activity and atmosphere. Generated video: can animate a good image, but shouldn't simulate actions or landscapes that aren't actually available. Photo: still suitable when no usable clip exists.

Even a generated video starts from the owner's content

Generation is not fully autonomous. StayFast uses the Extra's photograph, title, description, selected model, duration and the owner's directions. The initial prompt suggestion is prepared locally: it doesn't use a second AI model and it doesn't consume fal.ai credits.

The suggestion changes with the video type. For Pika it can describe a short or complete gesture. For Seedance it can describe a slower or faster camera movement depending on the selected duration. The owner can edit the text up to 600 characters. If they edit it, StayFast doesn't silently replace it when the model, duration or preset changes: a confirmation is requested before restoring a suggestion.

The generated result should be treated as a proposal

An AI video shouldn't be considered correct only because the generation technically succeeded. Before publishing, the owner should review people, hands and faces, objects, furniture, colours, any text that may have appeared, proportions, continuity and commercial consistency with the Extra.

That's why the result initially stays as a preview. The owner can watch it, discard it, keep the previous video, possibly generate a new version or accept it through an explicit action. StayFast does not publish the result automatically.

Every new generation may result in a new direct charge on the tenant's connected fal.ai account. No automatic fallback runs between Pika and Seedance.

Photos with people require more care

A photograph with people can be very effective at presenting an experience, but it requires more responsibility. The owner should verify they have the right to use the image and that recognisable people are aware of its intended use. This applies to publishing the photo, uploading a real video and using the photo as a source for a generated video.

AI generation doesn't remove the rights attached to the original content. When in doubt, it can be more prudent to use an image produced directly by the property, non-recognisable people, details of hands, products or surroundings, a scene without people, or a real video made with the appropriate permissions.

Common mistakes to avoid

Animating every Extra: if everything moves, nothing really stands out. It's better to select the Extras that benefit most from movement.

Asking the model to invent: the prompt shouldn't ask for new people, missing rooms, un-photographed products, promotional text or details the property doesn't offer. The video should enhance the image, not replace reality.

Always picking the maximum duration: a longer video usually costs more, takes more time and can introduce more distortions. Duration should follow the required action.

Publishing without watching the whole video: an error may only appear in the final seconds. The preview should be reviewed from start to end.

Using video to compensate for an incomplete offer: visual content doesn't replace price, description, conditions, timing, availability or how to request. The Extra must be understandable even without playing the video.

A simple rule for deciding

Use real video when you need to show what actually happens; use generated video when a good photo needs movement; keep the photo when clarity and information matter more than animation.

There isn't one right format for every Extra. There is the format most consistent with the content, with the resources available and with what the guest needs to decide.

StayFast's role

StayFast doesn't force the owner to pick a single path. It lets you upload your own video, generate a video from the photo, keep a simple image, choose Pika or Seedance, select the duration, edit the suggestion, review the preview, approve the result and display the content in the guest showcase and the Extras catalogue.

The value isn't only in generation. It's in the ability to choose the most suitable content and bring it all the way to the guest without splitting editor, video and commercial offer apart.

To understand how the whole path works technically, read <a href="/blog/how-an-extra-becomes-a-video-owner-to-guest-showcase">How an Extra becomes a video: from the owner's content to the guest showcase</a>. To explore the value of movement in presenting Extras, read <a href="/blog/photo-to-video-property-extras">From photo to video: making your property's Extras feel more alive</a>.