How to Animate a Photo with AI: 9 Ways, Including Free Ones and No VPN

"Animate a photo" is the most searched AI query in the Russian-speaking web: almost seventy thousand searches a month. People usually want one of three things: make an old family photo smile, turn a portrait into a short clip for stories, or bring a product image to life. All three work the same way: upload the picture, describe what should happen, and get a few seconds of video half a minute later.

Below is what actually works in September 2026, what is genuinely free (not "free for the first three tries"), what opens from Russia without VPN gymnastics, and most importantly how to phrase the request so the face stays the same face. That last part is why most first attempts go straight to the bin.

Short version: what to pick

• Fast, free, no extra sign-ups: Alice AI by Yandex. Four seconds at 480p, but it works from Russia and asks for nothing.
• Quality and up to 15 seconds: Kling. The best motion physics available, free credits every day, though they cover roughly one generation.
• Russian interface and a Russian account: Kandinsky by Sber. Ten generations a day, up to ten seconds.
• You want the person to speak: DreamFace or HeyGen. That is no longer "animate a photo" but a talking avatar, covered in a separate guide.

Free and available from Russia

Alice AI (Yandex)

The simplest option in the country. Open the Alice AI app, the Yandex app or Yandex Browser, upload a photo, write what should happen. Twenty to thirty seconds later you get a four-second clip at 480p in the same aspect ratio as the source. Four seconds and 480p is not much for serious work, but it is enough for stories, a greeting or simply seeing how this looks. No VPN, no foreign cards, no registration beyond the Yandex ID you already have.

Kandinsky (Sber)

Sign in with Sber ID, ten free generations a day, clips up to ten seconds, Russian interface, with a choice of camera movement and lighting. Motion quality is below Kling, but for a free Russian option with ten seconds it is the best there is.

Kling

The Chinese model that has topped motion-physics rankings for a year and a half: fabric, hair, water, walking. It is the one to pick when the animated photo must not look like jelly. Free tier gives 66 credits a day, roughly one generation. Length from three to fifteen seconds. From Russia the site opens unreliably, the app may be missing from local stores and Russian cards do not work; see the Kling guide. Russian aggregators offer Kling for roubles without any of that.

PixVerse, Hailuo, Pika

Three international services with free limits: PixVerse gives two or three generations a day, Hailuo up to three a day, Pika around eighty credits on sign-up. All open from Russia, none has a Russian interface, and free clips carry a watermark. Good for trying different engines on one photo and picking the one that understood the task.

Remini and Nero AI: template animation

A different class: not "describe what happens" but "pick a ready scenario", a smile, a head turn, a hug. Remini gives five generations a day and speaks Russian; Nero AI gives five trial animations. The result is predictable and samey, but you do not need to write prompts. A good fit for old family photos, where a light smile is exactly what is needed.

Aggregators: everything in one place, paid in roubles

If you use this regularly, it is easier to take one service with Kling, Seedance, Veo and others inside, paid in roubles, than to hop between ten sites. In our experience Russian aggregators differ mainly in price per generation and in how quickly they add new model versions. How to choose one is in the aggregator breakdown.

Free limits are for finding out which model suits you. After that, one subscription with several models beats three separate ones.

Why the face drifts and how to avoid it

The most common complaint: "I animated grandma, and on the video it is no longer grandma." The model draws the frames between the source and the next one, and with strong movement it loses its anchor. A few rules remove most of the problem.

1. Small movement. Ask for a smile, a head turn, a glance at the camera, light wind in the hair. Do not ask for "stands up and leaves". The smaller the movement, the truer the face.
2. Static camera. Add "static camera" to the prompt. Push-ins and pans force the model to redraw the face at a new scale, and that is where it drifts.
3. One hero. On group photos the model mixes up who does what. Crop into separate frames.
4. Sharp source. Run old shots through restoration first (Remini does that too), then animate. From a blurred face the model invents its own.
5. Shorter. Make the first attempts at minimum length. Five seconds with the right face beat fifteen of which two are usable.

How to write the prompt: examples

Most services understand Russian, but Kling, Pika and Hailuo work more precisely in English. The structure is the same: who is in frame, what they do, how the camera behaves, what must not appear.

Old portrait: "The woman slowly smiles and slightly turns her head toward the camera, blinks. Static camera. Face and features unchanged." Product: "A mug on a table, light steam rising, window light from the left, static camera." Full-length story clip: "Light wind moves the hair and clothes, the person shifts weight from one foot to the other, looks at the camera."

In our studio every prompt gets one more line about the image: film softness, grain, no sharpening, no HDR. It removes the plastic "AI look" by which such clips are recognised from the first frame. To get the same effect, append: "soft film look, fine grain, no sharpening, no HDR".

Step by step: animate a photo in Alice in a minute

1. Open the Alice AI app (or Yandex, or Yandex Browser) and sign in with Yandex ID.
2. Tap the paperclip in the chat and upload the photo. A chest-up or waist-up portrait with a clear face and no harsh shadows works best.
3. Write what should happen: "smiles and slightly turns the head, static camera". One action, not three.
4. Wait twenty to thirty seconds. If the face drifted, do not rewrite everything: reduce the movement and try again.
5. Save the clip. It is 4 seconds at 480p: fine for stories, not for a TV screen.

Old and black-and-white photos: a different order

With old photos the mistake is almost always the same: animating them as they are. The model gets a blurred, scratched face and completes it as it sees fit, producing a similar but different person. The right order: restore first, then animate. Scan or reshoot under even light, run through face restoration aiming for sharp eyes and a clear outline, skip colourisation (it adds another layer of errors), and animate with minimal movement. For a family archive a smile and a blink are enough, and they move people more than "stood up and walked away".

Frequently asked questions

Can I animate a photo completely free and without registration?

Without registration, practically no: every service asks at least for a Yandex ID, Sber ID or Google sign-in. Completely free, yes: Alice AI, Kandinsky and the daily limits of Kling and PixVerse. The restriction everywhere is length and resolution, not the number of tries.

Which AI animates photos best?

By motion quality in September 2026: Kling 3.0, then Seedance 2.5 and Veo 3.1. By simplicity and availability from Russia: Alice AI and Kandinsky. If the result is for work rather than stories, start with Kling through an aggregator.

Why does the face change on the video?

The model draws intermediate frames and loses anchor features with large movement or camera motion. Minimal movement, a static camera and a sharp source fix most of it. No model removes the problem entirely.

Can I make the person on the photo talk?

That is a separate technology, a talking avatar with lip sync. DreamFace, HeyGen and others do it. See our HeyGen guide.

Can an animated photo be used in advertising?

Technically yes; legally check two things. First, the service terms: free tiers often forbid commercial use. Second, the rights to the image and the consent of the person on it. Animation changes nothing in ordinary law: a stranger in an ad without consent is not allowed, animated or not.

When four seconds is not enough

An animated photo is one scene. A business video is a dozen such scenes that must match in light, hero and style, and that is not solved with a button. How separate generations become a finished video is in our step-by-step AI video guide, and if you want to learn to do it yourself on your own task, there is AI video training where this is exactly what we cover.