AI Hug Video Trend: Turn Two Photos Into an Embrace

The AI hug video is the quietest wave to take over Reels and TikTok — no flex, no punchline, just two separate photos animated until the people in them embrace. A grandmother meets her grandchild again. Someone hugs their own younger self. A long-distance couple finally closes the gap. The trend first went mass-market when Luma's Dream Machine turned "two photos in, one hug out" into a one-minute job — which is why so many people still search it as "luma ai hug" — and it resurfaces every time a new image-to-video model gets better at holding two faces steady in one scene.

New here? The homepage overview sums up the AI hug trend on one screen, and the trends hub tracks every embrace variant as it breaks.

What is the AI hug video trend?

The format is a five-to-ten-second clip built from stills, not footage: you feed one or two photos into an image-to-video model, describe the embrace, and the model animates the moment two people come together and hug. Nothing is filmed. The people can be you and a friend, past and present versions of the same face, or someone you have not seen in years.

That last case is the engine of the whole trend. Most viral AI formats run on spectacle — cinematic night drives, celebrities swapped into famous scenes. The hug runs on feeling. The comment section is people tagging the person they would make one about, which is a stronger share trigger than any effect, and it is why dedicated hug generators with preset styles — warm hug, tearful reunion, bear hug — keep appearing as fast as the trend cycles back.

Why this trend keeps coming back

Watch the pattern across waves and the anatomy is always the same: one recognizable visual, one tool moment, and a participation hook where the output stars you. The hug trend has the strongest version of that third piece — the output stars you and someone you actually miss. Every time a new model ships with better multi-image reference support, creators rediscover the format, and a fresh round of tutorials, presets and scene libraries appears almost overnight. Template marketplaces now sell the hug as a one-click style, and general-purpose models take it as a plain prompt.

Which tools make the AI hug video?

Dedicated hug generators

A wave of single-purpose sites does exactly one thing: upload two photos, pick a hug style, get a clip. They are the fastest route to the baseline look, but read the fine print — most stamp a watermark on free exports and cap resolution, which is noticeable the moment the clip is posted full-screen.

General image-to-video models

Luma's Dream Machine put this exact use case on the map, and today any model that accepts two reference images can stage the embrace — Luma, Hailuo, Kling and Veo-class models all handle it. The general models are where quality lives, because you control the prompt instead of picking from a style list.

The fastest route on this site

The generator runs this trend natively: sign in, switch to the Custom Scene tab, upload your two photos, and paste a hug prompt like the template below. A two-photo hug video costs 10 credits, and a text-only dress rehearsal — same prompt, no photos — costs 6 if you want to test the wording first. There is no editing step: the clip comes out ready to post.

Step by step: from two photos to one embrace

1. Pick two strong photos. Front-facing, well-lit, both subjects at similar scale. Mismatched lighting and wildly different distances from the camera are the number-one cause of warped faces and broken hands.
2. Open the generator and switch to Custom Scene. Upload both photos — left and right — so the model knows who is who.
3. Paste a hug prompt. Describe who is walking toward whom, where, and with what mood. The template below covers it in one line.
4. Generate, review, iterate. The embrace itself is easy; hands and the moment of contact are where models stumble. Change one variable at a time — hug style, setting, camera — instead of rewriting the whole prompt.
5. Post with the right audio. A soft piano or slowed track under a hug clip is half the reach. Export, attach the sound in TikTok or Reels, publish while the wave is still warm.

Copy-paste hug prompt template

The community's template shape, with the variables in brackets:

two people slowly walk toward each other and share a warm embrace,
[setting], soft natural light, steady medium shot,
gentle camera push-in, emotional and heartwarming mood

Swap one variable at a time so you can see what each edit does:

• Hug style: warm gentle hug, tearful reunion, big bear hug, lift-and-spin
• Setting: a sunlit kitchen, an airport arrivals hall, a rainy street, the family living room
• Camera: medium shot that holds both faces, slow push-in, over-the-shoulder as they close the distance
• Mood: heartwarming, bittersweet, joyful

Variants worth trying

• Hug your younger self. Two photos of the same face at two ages, walking toward each other. It is the most reposted variant of the trend and works precisely because everyone has a version of it.
• The reunion you could not have. Grandparents and grandchildren, deployed partners, friends split by continents. These are the clips that travel furthest — treat the subject with the directness it deserves and let the moment carry it.
• Memorial tributes. Some of the most-shared hug videos animate a photo of someone who is gone. If you make one, keep the motion gentle and the staging dignified, and make it for the family first, the feed second.
• Fan meets idol. Fun until it is not: a celebrity hug video uses a real person's likeness. Keep it clearly playful, never present it as real, and never attach a brand or endorsement to it.

Red lines before you post

• Likeness and consent. Only use photos you have the right to use, of people who would be comfortable with the result — that goes double for anyone grieving and triple for minors.
• Label it as AI. TikTok and Instagram both require AI-generated content to be flagged; an unlabeled "reunion" that reads as documentary is the fastest way to lose the audience this trend earns you.
• Do not pass it off as real. The format works because viewers know it is AI and feel something anyway. The moment it pretends to be real news, it becomes a different problem.

Frequently asked questions

Can I make an AI hug video for free?

The cheapest test is a text-only run — describe the two people in the prompt and the model invents both, at 6 credits. The real trend, with your two photos, costs 10 credits per clip, and the Starter pack covers three two-photo hugs. Dedicated hug generators have free tiers, but watermark the export.

Do I need two separate photos?

No. One photo with both people already in frame animates fine, but two single-person photos give the model a cleaner beat — the walk toward each other — which is the moment that makes viewers feel it.

How long should the clip be?

Five to ten seconds. Long enough to land the embrace, short enough to loop — and the loop is what the algorithm rewards.

Which photos work best?

Front-facing, well-lit, similar scale and distance from the camera. Faces roughly level with each other save the model from inventing body geometry, which is where most bad hug clips go wrong.

Where should I post it?

TikTok and Instagram Reels, with an emotional or slowed audio track. The trend spreads through those two platforms, and the sound choice carries a large share of the reach — posting a hug clip on generic audio leaves most of the wave on the table.