20 Secret ChatGPT Codes That Turn One Photo Into a Cinematic Video (Higgsfield Plugin)

The codes run on the Higgsfield plugin inside ChatGPT. Without it, ChatGPT describes the video instead of making it. Takes about thirty seconds.

1. Tap here to get Higgsfield and sign in. New users get a 3 day free trial through that link.
2. Open the ChatGPT app, tap the plus button next to the message box, then Plugins. Search Higgsfield and add it. On desktop it's Settings, then Plugins.
3. Start a new chat. From now on, typing @ in the message box brings Higgsfield up.

Every video you generate inside ChatGPT runs on Higgsfield credits, so the trial is the cheapest way to run through all 20.

⚡ Add Higgsfield to ChatGPT (3 day free trial)

Same three moves for all 20.

1. Attach the photo. Two of the codes take a video instead, I'll flag those.
2. Type @higgsfield, pick it from the menu, then type the code on the same line.
3. Send it.

Thirty seconds to a minute later, boom, the video is sitting in the chat. No prompt writing. No settings.

Two things separate a clean result from a weird one. One subject per photo, fully in frame, on a simple background. And if you want something specific, add one sentence after the code. The code sets the shot, your sentence sets the details.

This is the exact message for the first one in the video. Swap the code for any of the 20 below.

/earthzoom: brings your photo to life, then the camera pulls up and away through the clouds until you're looking down from the stratosphere. Feed it an outdoor location, a building or a landmark.

/thirdpersonview: takes a first person video and re-shoots the same moment from outside, like a crew was following you. Feed it a video of you walking a trail with your full body visible. For the drive in the video, upload the first person clip plus a photo of your car so it knows what to render.

/productexplosion: your product blows apart into floating pieces, as a video, not a still. Feed it one sharp photo of the product, fully visible, simple background.

/fpvdrone: an FPV drone flight through the scene, low, fast and ultra realistic. Feed it a mountain valley, coastal cliffs, a winding road through a forest, or your car on a road.

/flyover: a wide aerial pass over the whole scene. Feed it a landscape, a city skyline, a building or a coastline.

/zoomout: starts tight on the subject and pulls back to reveal where they are. Feed it a clear subject in a recognisable setting, a person in a garden, a car on a road.

/spiralshot: the camera spirals up and around the subject. Feed it a lighthouse on a cliff or a person on a rooftop, with open space around them.

/orbitshot: a full circle around the subject at eye level. Feed it one fully visible subject with clear surroundings, a car, a statue, a person, a product.

/360: a smooth 360 that shows every side. Feed it a sneaker, a perfume bottle or a sports car with clear space around it.

/dollyzoom: the Vertigo shot. Your subject stays locked while the background stretches away behind them. Feed it a person in a long hallway or on a city street with the background running far behind them.

/gimbalshot: a smooth gimbal glide through the space. Feed it a wide scene with depth and room to move, a hallway, a forest path, a big interior.

/tvcommercial: turns the product into a polished TV spot. Feed it a perfume bottle, a sneaker, a beverage can. It needs a product, not a random photo of your cat.

/macrozoom: pushes in until you can see the texture. Feed it a sharp, high resolution photo with fine detail, an eye, a flower, fabric, jewelry.

/macroreveal: opens on one tiny detail, then pulls out to reveal the whole thing. Feed it the full subject with a distinctive detail to start on, a watch face, a flower petal.

/slowmotion: a fast moment stretched into ultra slow motion. Feed it something caught mid action, a water splash, a balloon bursting.

/ugc: a creator style clip, like someone filmed themselves recommending the product at home. Feed it a person holding the product at home with their face and the product both clearly visible.

/glambot: the red carpet slow motion sweep. Feed it a person in a glamorous outfit, mid hair flip or dress twirl, with fabric or hair moving.

/pov: re-shoots the scene through someone else's eyes. Feed it a photo or video with a clear action and setting, then name the perspective after the code, a cyclist, a chef, a dog. For example, /pov from the dog.

Use your own face, or someone who has agreed to be in it.

These two aren't camera moves, they're ad modes. Same photo, same @higgsfield, same one line.

/adcreative: builds a full ad out of one product photo. Feed it a product on a clean background.

/lifestyleshot: drops the product into a real setting, a kitchen counter, a desk, a bathroom shelf. Feed it a plain white background product photo, that is exactly what it's built for.

If a code ever comes back as text instead of a video, make sure Higgsfield is selected with the @, then say it in plain English. "Make an FPV drone video of this" does the same job. The codes are the shortcut.

Paste this into your notes so you're not scrolling back here mid chat. Every code and the photo it wants, one line each.