AI Dolly Zoom Effect: The Vertigo Shot From One Photo
The VdoBloom AI dolly zoom effect turns one still photo into the vertigo shot: the person stays the same size in frame while the space behind them stretches away or presses in. You upload a picture, pick a pace, and the tab renders a short clip with the face, clothes, framing and light kept as they were shot. It is built for creators, editors and marketers who want the most recognisable camera trick in film without a dolly track, a zoom lens or a crew.
Is it free? New accounts get 10 free credits with no card. The tab opens on Seedance 2 Fast at 5 seconds, 720p and 9:16, which costs 50 credits, because that model holds faces and framing best while the camera moves. To spend only the free credits, switch the model picker to Runway Gen-3 (10 credits for 5 seconds at 720p) or Seedance 1.5 Pro (8 credits for 5 seconds at 720p) before you generate. Free-tier results play with a VdoBloom watermark; a paid plan unlocks the clean MP4.
Three templates set the pace. Vertigo is the classic steady pull-back with a matching zoom-in. Rush In plays the same move fast, so the background lurches away in one punchy beat and then settles. Slow Reveal draws it out so gradually that the room seems to open behind the subject almost without being noticed. Each template is an editable prompt, so you can say which way the camera travels, how far, and what the person does while it happens.
How it works
- 1
Upload one clear photo
A front-facing shot with the face sharp and plenty of visible depth behind the subject: a street, a corridor, a room with furniture receding into it. A dolly zoom works by stretching the background, so a subject flat against a wall gives it almost nothing to move.
- 2
Confirm consent
Tick the box confirming the photo is of you or of an adult who agreed to be animated. Generation stays locked until it is checked.
- 3
Pick a pace
Vertigo for the steady classic, Rush In for a fast snap, Slow Reveal for a long, calm opening-out. The three change how fast the background moves, not what the move is.
- 4
Direct the shot in the prompt
Add what a preset cannot know: whether the background should fall away or close in, whether the person walks toward the camera, what the setting is. Change one clause at a time so you can see what each instruction does.
- 5
Choose model, length and frame
The tab defaults to Seedance 2 Fast, 5 seconds, 720p, 9:16. Drop to 480p (24 credits) or switch to Runway Gen-3 or Seedance 1.5 Pro for cheaper test takes, and pick 16:9 if the clip is going into a wide edit, where there is more background to stretch.
- 6
Generate and watch it full size
Rendering runs on the server, so you can close the tab. The move is subtle at thumbnail size: watch the clip full-screen and check that the person stays the same size from the first frame to the last.
Why use this tool
The person holds, the background moves
Every template locks the subject in place, same size and same sharpness, while the space behind them expands or compresses. That matched pull-back and zoom-in is the whole illusion, and it is why the result reads as a dolly zoom rather than a plain zoom.
Identity and framing kept from the photo
The prompts tell the model to keep the same face, hair, clothes, expression, framing and lighting as your upload and not to re-light, re-style or re-frame the shot. The camera move is added; the person is left alone.
A face-holding default model
The tab opens on Seedance 2 Fast, the same default used for the crying effect, because it holds faces and composition steady while the camera travels. You can still pick any image-to-video model in the picker.
Three paces, all editable
Vertigo, Rush In and Slow Reveal each ship their own prompt, and the prompt box stays open, so direction, distance, speed and the subject's action are all yours to change.
One credit balance for every model
Test the framing on an 8- or 10-credit model, then re-run the take you like on Seedance 2 Fast or Runway Gen-4.5 without a second subscription. Failed generations refund their credits automatically.
Vertical or wide from the same photo
Choose 9:16 for Reels, TikTok and Shorts or 16:9 for a YouTube or film edit before you generate, so the composition is built for the screen it is going to.
What is a dolly zoom?
A dolly zoom, also called the vertigo effect, a zolly or a contra-zoom, is a camera move where the camera travels toward or away from the subject while the lens zooms the opposite way. The two movements cancel out at the subject, so the person stays the same size in frame, while everything behind them changes perspective: the background appears to rush away when the camera pulls back and zooms in, and to crowd forward when the camera pushes in and zooms out.
Alfred Hitchcock's Vertigo (1958) made the shot famous, and Steven Spielberg used it on the beach in Jaws (1975). It reads as unease, realisation or a sudden shift in perspective because the brain sees the subject hold still while the world around them warps. Shooting it for real takes a track, a zoom lens and a careful operator keeping the subject's size constant. Here the model generates the whole move from one still image.
Which model to use for a dolly zoom
The tab opens on Seedance 2 Fast because face and framing stability matter more on this effect than on most: if the person drifts, grows or changes shape, the illusion breaks. Cheaper models are useful for checking whether your photo has enough depth before you pay for the default. Prices below are for one clip, from VdoBloom's credit table.
| Model | Clip | Credits | Best for |
|---|---|---|---|
| Seedance 2 Fast (tab default) | 5 s, 720p | 50 | Final takes where the face and framing must hold |
| Seedance 2 Fast | 5 s, 480p | 24 | Cheaper preview on the default model |
| Seedance 1.5 Pro | 5 s, 720p | 8 | Fits inside the 10 free credits; quick framing test |
| ByteDance V1 Pro Fast | 5 s, 720p | 7 | Cheapest test of a new photo |
| Runway Gen-3 | 5 s, 720p | 10 | Exactly the 10 free credits |
| Kling 2.6 image-to-video | 5 s | 22 | Mid-price alternative |
| Runway Gen-4.5 | 5 s | 48 | Higher-fidelity alternative to the default |
Read more: Seedance 2 Fast model page · Seedance 1.5 Pro model page · Runway Gen-3 model page
What a dolly zoom costs
Credits come from free signup credits, one-time packs that never expire, or a monthly subscription that resets each cycle. The Lite pack is 300 credits for $19.50 one-time, or $15 a month on a subscription. Here is what that buys on this tab:
| Plan | Credits | Seedance 2 Fast 5 s 720p (50) | Runway Gen-3 5 s 720p (10) | Seedance 1.5 Pro 5 s 720p (8) |
|---|---|---|---|---|
| Free signup | 10 | 0 | 1 | 1 |
| Lite, one-time $19.50 | 300 | 6 | 30 | 37 |
| Lite, $15 per month | 300 per month | 6 | 30 | 37 |
Where a dolly zoom works
The shot is strongest when it carries one idea: a realisation, a reveal, a moment where the world shifts around one person. Use it once per edit and it lands; use it three times and it turns into a gimmick.
- Openers and hooks: the first second of a Reel or Short, where the background lurch stops the scroll.
- Reaction beats: the 'wait, what?' moment in a comedy skit or story video.
- Music videos and lyric edits: a slow reveal under a held note.
- Short films and trailers: a classic film-language beat without hiring a crew.
- Product and brand content: a founder, chef or presenter holding still while the shop or kitchen opens out behind them.
- Travel and real estate: a person in front of a landmark or a long hallway, where there is plenty of depth to stretch.
Tips for a clean dolly zoom
- Start with depth. Corridors, streets, staircases, rows of trees and long rooms give the move something to stretch. Plain walls and studio backdrops do not.
- Keep one clear subject in the middle third of the frame, with the face reasonably large and sharp.
- Avoid heavy backlighting and motion blur in the source photo; both make the subject's edges harder to hold.
- Say the direction in the prompt: 'the background falls away' for the Vertigo feel, 'the background closes in' for the reverse.
- Match the pace to the length. Rush In works at 4 to 5 seconds; Slow Reveal needs more time to be felt.
- Test on Seedance 1.5 Pro or Runway Gen-3 first, then re-run the winning prompt on Seedance 2 Fast.
What it can and cannot do
The effect animates the photo you give it: it can add the camera move, hold the person and invent the parallax in the background. It cannot recover depth that is not in the picture, so a cropped headshot against a flat backdrop gives a weak result. There is no separate direction toggle or intensity slider on this tab; direction and strength are set through the template and the prompt. Clip length, resolution and aspect ratio depend on the model you pick. Upload only your own photos or photos of adults who have agreed to be animated, never of children, and label the clip as AI-generated where the platform asks.
Frequently asked questions
What is a dolly zoom effect?
A dolly zoom, also called the vertigo effect, zolly or contra-zoom, is a shot where the camera moves toward or away from the subject while the lens zooms the opposite way. The subject stays the same size in frame while the background appears to stretch or compress. VdoBloom generates the move from one still photo.
Is the AI dolly zoom effect free?
You can make one free. New accounts get 10 free credits with no card. The tab's default, Seedance 2 Fast at 5 seconds and 720p, costs 50 credits, so for a free clip switch the picker to Runway Gen-3 (10 credits) or Seedance 1.5 Pro (8 credits) before generating. Free-tier results are watermarked and downloads carry the watermark until you are on a paid plan.
How much does a dolly zoom video cost?
It depends on the model. At 5 seconds and 720p: Seedance 2 Fast 50 credits (24 at 480p), Runway Gen-4.5 48, Kling 2.6 image-to-video 22, Runway Gen-3 10, Seedance 1.5 Pro 8 and ByteDance V1 Pro Fast 7. The exact cost is shown before you generate. The Lite pack is 300 credits for $19.50 one-time.
Will my face change in the video?
The prompts tell the model to keep the same face, hair, clothes, expression, framing and lighting as your upload, and the default model is chosen because it holds faces well while the camera moves. A face that started small or blurred can still soften, so use a sharp, front-facing photo and watch the result full size.
What kind of photo works best?
One clear subject with visible depth behind them: a street, a hallway, a staircase, a room with furniture receding into it. The effect stretches the background, so a person standing flat against a wall gives it almost nothing to move. Even light and a reasonably large face help.
Can I make the background close in instead of fall away?
Yes, through the prompt. The templates are written for the classic pull-back with a zoom-in, where the background falls away. Rewrite the prompt to say the camera pushes in while zooming out so the background closes in around the person, and keep the line that holds the person the same size.
How long should a dolly zoom clip be?
Long enough for the move to read: about 4 to 8 seconds for most shots. Rush In works at the short end, Slow Reveal needs more time, and Vertigo sits between. The tab opens at 5 seconds; available lengths depend on the model you pick.
How is this different from the camera-move presets on image to video?
The camera-move tiles on the image-to-video tab add a move as one line inside a bigger prompt. This tab is a dedicated effect with its own three templates, a prompt written for the contra-zoom and a face-holding default model, so it is the faster route when the dolly zoom is the whole shot.
How long does it take to generate?
Usually a few minutes, longer for higher resolutions and heavier models. The job runs on the server and lands in your Creations library, so you can close the tab or queue several paces from the same photo.
Can I use the clip commercially?
Under VdoBloom's terms you keep the content you upload or generate and you are responsible for it, which covers ads, social posts and your own site. You must have the rights to the photo, including consent from anyone recognisable in it, and you should label the result as AI-generated where a platform asks.