Video from a photo using AI: how to make it in 2026
What is video from a photo using AI and how it works
The technology is called image‑to‑video. You give the neural network one image (or several), and it fills in the missing frames: animates the background, adds camera movement, makes hair sway, water flow, and a person turn their head or smile. Essentially, the model “imagines” how the scene could continue in time and draws it frame by frame.
It’s important not to confuse this with two older formats:
- Slideshow – simply switching static photos to music with transitions. The images don’t move; no neural network is needed.
- GIF – a short, looping animation without sound and usually of low quality.
But video from photos using AI is real motion inside the frame: realistic physics, smooth dynamics, sometimes sound. The difference is roughly like flipping through an album versus a living photograph from “Harry Potter”.
Which neural networks can do this in 2026:
- Kling 2.6 / 3.0 – currently the best image‑to‑video engine for motion realism and face preservation. Version 2.6 was released on December 3, 2025, version 3.0 in February 2026.
- MiniMax Hailuo 2.3 (October 28, 2025) – leader in correct physics (falls, fabric, liquids) and also fast.
- Google Veo 3 – the only mainstream model that generates video with sound immediately, so it’s often used for talking avatars.
- ByteDance Seedance – good for complex scenes with multiple objects.
We did a detailed breakdown of portrait animation in a separate article – how to animate a photo with AI. Here we’ll focus on practice: how to make a video from a photo from start to finish.
How to make a video from a photo using AI: step‑by‑step guide
I’ll show using the example of the Russian service revideo.ai – it works without a VPN, with payment via SBP and a Russian interface, and the first video can be made for free even without registration. The algorithm is almost the same everywhere, so the guide is universal.
Step 1. Open the service and select the mode. Go to revideo.ai and choose “Video from photo” (image‑to‑video). Registration is not needed for the first video.
Step 2. Upload your photo. A regular JPG or PNG works fine. It’s better to use a clear image without strong noise or overexposure – more on requirements below.
Step 3. Describe what should happen. In the prompt field, write in Russian what motion you want: “light wind sways the leaves, camera slowly zooms in” or “the girl turns her head and smiles”. The service automatically translates the prompt for the model – no need to think in English.
Step 4. Choose the model and format. For most tasks, Kling v2.6 Turbo Pro is enough – it’s the default. If you need fast and free, choose Hailuo 2.3. Set the format right away: 16:9 for YouTube, 9:16 for Reels, Shorts, and marketplace product cards (there is a ready‑made preset).
Step 5. Start generation and download the result. Wait from 30 seconds to a couple of minutes. The finished clip downloads in one click. On the free plan, there will be a watermark and 720p; on the paid plan, clean 1080p.
Want to try right now? Make your first video from a photo on revideo.ai for free – without VPN, without a card, and without registration.
If the first result isn’t perfect – that’s normal. Change the wording of the prompt, add details about speed and direction of motion, and regenerate. Usually the second or third try hits the mark.
Where to make a video from a photo: 2026 services (comparison)
There are many services, but for Russia half are ruled out because of VPN and foreign cards. I’ll break it down honestly – with pros and cons.
| Service | No VPN | Payment from Russia | Russian language | Models | Price |
|---|---|---|---|---|---|
| revideo.ai | Yes | SBP, Russian cards | Yes | Kling, Veo 3, Hailuo, Seedance | from 799 RUB/month |
| Alisa (videogen) | Yes | Yandex Plus | Yes | proprietary | Plus subscription |
| Shedevrum | Yes | free | Yes | proprietary | free |
| Renderforest | Partially | foreign card | Partially | proprietary | from $9/month |
| Pollo / Luma / Pika | No (VPN) | foreign card | No | various | from $10/month |
Let’s go into more detail.
1. revideo.ai – the optimal choice for Russia. The main advantage is accessibility: works without VPN, payment via SBP and Russian cards, interface and prompts in Russian (with auto‑translation). Servers in Russia (152‑FZ compliant), photos are deleted after 24 hours, support responds in Russian. Under the hood – the best global models (Kling, Veo 3, Hailuo, Seedance), includes a 9:16 preset for marketplaces and Reels. First video is free. This is one of those cases where you don’t have to choose between “convenient” and “high‑quality”.
2. Alisa (videogen) from Yandex. Generates short clips, available via Yandex Plus subscription. Convenient if you already have Plus, but has fewer capabilities for serious work (model selection, long scenes, talking head).
3. Shedevrum. A completely free Yandex app. A great entry point to try the genre, but quality and control are more modest than specialized services. From the same ecosystem – Kandinsky and GigaChat from Sber.
4. Renderforest. A well‑known Western service, but it is closer to a slideshow constructor than to real image‑to‑video, and requires a foreign card.
5. Foreign services – Pollo, Luma, Pika, Runway. Technologically strong, but almost inaccessible from Russia: need VPN and a foreign card. Sora from OpenAI is blocked in Russia, and its free access was closed on January 10, 2026 – if you were specifically looking for it, check out Sora alternatives in Russia without a VPN.
6. Mobile apps (various “animators” from stores) – convenient for fun, but often have hidden subscriptions, watermarks, and weak models.
Can you make a video from a photo for free
Yes, and this is the most common question. But it’s important to understand what “free” means in practice.
What is truly free in Russia:
- Shedevrum – truly free, no hidden limits, right in the app.
- revideo.ai – 1 video per day without registration (720p, with watermark). An excellent way to test quality before purchasing.
- Alisa – as part of a Yandex Plus subscription, which many already have paid for.
Where free modes usually have a catch:
- Watermark. The service logo in the corner is almost standard for free tiers. Tolerable for social media, not for business.
- Low resolution. Free is most often 720p; clean 1080p is already paid.
- Limits. One or two videos per day, queues, length restrictions.
If you need a neural network for video from photos for free just once (a greeting, a test) – the free mode of revideo or Shedevrum will suffice. If you make content regularly, a paid plan is cheaper: on revideo’s “Start” plan (799 RUB/month, ~20 videos), one video costs about 40 RUB – many times cheaper than a freelancer, and no watermarks.
Where video from photos is used
An animated image is not a toy but a working tool. Here’s where it actually delivers results.
Social media. Reels, Shorts, and Stories love motion: an animated photo holds attention better than a static one and gets more views. The 9:16 format is the standard for vertical feeds.
Marketplaces. The most monetizable use. Video on a Wildberries or Ozon product card increases conversion and trust: you can show the product in motion, from different angles, “in real life”. Just upload a product photo and set a gentle rotation or zoom. The vertical 9:16 fits perfectly into the card – revideo has a ready‑made preset for this.
Greetings and personal clips. Animating an old family photo, making a moving birthday card – touching and fast.
Advertising and portfolios. Dynamic creatives from a single image for targeted ads, animated works in a designer’s or photographer’s portfolio – all of this used to require a videographer budget.
Making product cards for Wildberries or Ozon? Create a video from a photo in 9:16 format on revideo.ai – the marketplace preset is already built in, payment via SBP.
A separate big topic is animating a face so that the person speaks. That’s a talking avatar from a photo – Veo 3 with sound works well for such tasks.
Which photo works best and how to get the best result
Video quality depends 80% on the source image and the prompt. A few practical rules.
Photo requirements:
- Sharpness. A crisp image without blur or strong noise – the neural network finds it easier to “build” motion.
- Good lighting. Even light is better than deep shadows and overexposure.
- One main subject. The clearer what is in the center of the frame, the more stable the result. With a crowd, the motion may “melt”.
- Sufficient resolution. Small pixelated images yield poor results.
Prompt tips:
- Describe motion, not the picture. Not “beautiful girl by the sea”, but “the girl turns her head toward the camera, wind blows her hair, waves roll onto the shore”.
- Specify camera movement. “Slow zoom in”, “orbit around the object”, “smooth pan” – these strongly affect the feel of the clip.
- Set the tempo. “Gentle, smooth motion” vs “dynamic scene” will give different results.
- Don’t overload. Too many actions in one short clip – and the scene will fall apart. One or two movements over 5 seconds.
And most importantly – experiment. Regenerating costs pennies, and the second or third version is almost always better than the first.
In a nutshell: the key points
- Video from a photo using AI (image‑to‑video) is real motion inside the frame, not a slideshow or GIF.
- Best models of 2026: Kling 2.6/3.0 (realism), Hailuo 2.3 (physics, speed), Veo 3 (with sound).
- For Russia, revideo.ai is optimal: no VPN, SBP payment, Russian language, servers in Russia, first video free.
- Free is possible via Shedevrum and revideo’s free mode (720p + watermark); for business you need a paid plan without watermark.
- Main applications – Reels/Shorts/Stories and Wildberries/Ozon marketplace product cards in 9:16 format.
- Result depends on a clear photo and a prompt that describes motion and camera.
Frequently asked questions
How to make a video from a photo for free?
In Russia, the Shedevrum app (no limits) and revideo.ai’s free mode work for free – 1 video per day without registration, at 720p with a watermark. That’s enough to try the genre or make a one‑off clip. For watermark‑free quality at 1080p, you’ll need a paid plan.
Which neural network makes videos from photos?
The best image‑to‑video models in 2026 are Kling 2.6/3.0 (realistic motion and face preservation), MiniMax Hailuo 2.3 (physics and speed), and Google Veo 3 (generation with sound). In Russia, the easiest way to access them is via revideo.ai, which brings all these engines together in a single Russian‑language interface.
Can I make a video from several photos?
Yes. In addition to animating a single image, there is a multi‑image mode – a slideshow video from several photos with smooth transitions. On revideo, it is available on the “Pro” plan (1,190 RUB/month) together with video clips from prompts.
What format is the video in?
Usually MP4. You set the aspect ratio yourself: 16:9 for YouTube, 1:1 for feeds, and 9:16 for Reels, Shorts, Stories, and marketplace product cards. Resolution – 720p on the free plan and 1080p on paid plans.
Is it suitable for marketplaces (Wildberries/Ozon)?
Yes, this is one of the most demanded applications. Video on a product card increases conversion and buyer trust. For Wildberries and Ozon, you need vertical 9:16 – revideo has a ready‑made preset for this; just upload a product photo and set a gentle motion or zoom.
How much does it cost to make a video from a photo?
Free – on Shedevrum and revideo’s free mode. Paid on revideo: “Start” plan 799 RUB/month (~20 videos, 1080p, no watermark) – about 40 RUB per clip; “Pro” 1,190 RUB/month (~40 videos plus multi‑image and video clips). This is many times cheaper than ordering animation from a freelancer.