How to Make an AI Wink Video From One Photo (VdoBloom Wink Effect)
Animate a still portrait into a wink clip on VdoBloom: the Playful, Flirty and Cute templates, the exact credit cost per model tier, the photo that works best, and step-by-step settings.
To make an AI wink video, upload one clear front-facing photo to the VdoBloom Wink tab, pick the Playful, Flirty or Cute template, and generate — the cheapest working setup is Runway Gen-3 at 720p for 5 seconds, which costs 10 credits ($0.50), and new accounts start with 10 free credits so the first wink clip costs nothing. The Wink effect is a single-image animation: you give it a still portrait, and the model produces a short clip in which the subject closes one eye and returns to a smile. There is no timeline, no keyframing and no face rigging to set up.
If the photo is of a real person who is not you, you must have that person’s consent before uploading it and before publishing anything the effect produces. That is not a formality — a wink is a personal, expressive gesture, and putting one on someone’s face without permission is the fastest way to get a clip taken down.
What the Wink effect actually does — and what it does not
Honest description first. Wink is one of VdoBloom’s direct effect types: your uploaded image is handed straight to the chosen video model together with an effect prompt, and the model generates every frame from scratch. It is not a surgical edit of your original photo. The first frame usually looks close to your upload, and then the model re-draws the face for each subsequent frame.
What follows from that:
- The wink itself is reliable. Eye closure on one side is a motion these models handle well.
- Fine identity detail can drift — freckles, a mole, the exact shape of an eyebrow, jewellery near the face.
- Teeth and the inside of the mouth are the weakest area. A closed-mouth smile is safer than a wide grin.
- There is no audio in the clip. Wink is a silent expression effect.
The photo that works best
The single biggest quality lever is the input image, not the model. A good Wink source photo is:
- Front-facing or near front-facing. A three-quarter turn works; a hard profile does not, because one eye is hidden and the model has to invent it.
- Both eyes open and visible. No sunglasses, no hair across the eye, no heavy fringe shadow.
- Framed head-and-shoulders. The face should fill roughly a third of the frame. A full-body shot leaves too few pixels on the eyes.
- Evenly lit and sharp. Soft window light beats a harsh flash. Motion blur in the source becomes smeared motion in the output.
- Already in your target aspect ratio. Upload a 9:16 crop for Reels, Shorts and TikTok; 16:9 for YouTube and web headers.
If your only photo is low resolution or badly cropped, fix it before you animate it. Re-crop and clean it in the AI image editor, or generate a fresh portrait in text to image and animate that instead. A better still is always cheaper than three failed video generations.
The three Wink templates
The tab ships with three preset prompts. Selecting one replaces the prompt box — you can still edit the text afterwards.
| Template | What it is for | Prompt it loads |
|---|---|---|
| Playful | The default. Friendly, light, safe for brand and creator content. | playful wink, charming expression, natural smile, soft lighting, cinematic |
| Flirty | Warmer and more confident — dating profiles, fashion, lifestyle. | flirty wink, seductive expression, confident smile, warm lighting, cinematic |
| Cute | Brighter and more energetic — youth-facing and meme content. | cute wink, fun expression, bright smile, natural lighting, cinematic |
The only real differences are the adjective on the wink, the adjective on the smile, and the lighting word — which tells you how to write your own: keep the structure, swap the descriptors.
Credit cost per model and tier
The Wink tab defaults to Runway Gen-3, deliberately: at 720p for 5 seconds it is 10 credits, which is exactly the free signup balance. One credit is $0.05 at the Lite rate ($15 for 300 credits), so the dollar figures below are the real cost of a single clip.
| Model | Tier | Credits | Cost | Best use |
|---|---|---|---|---|
| Runway Gen-3 | 720p / 5s | 10 | $0.50 | Default. Test the photo here first. |
| Runway Gen-3 | 1080p / 5s | 20 | $1.00 | The keeper version once the 720p test looks right. |
| Runway Gen-3 | 1080p / 10s | 30 | $1.50 | Longest, sharpest Gen-3 option. |
| Runway Gen-4 Turbo | 5s | 20 | $1.00 | Cleaner faces than Gen-3, still cheap. |
| Runway Gen-4.5 | 5s | 48 | $2.40 | Best identity retention of the Runway tier. |
| Kling V2.5 Turbo | 5s | 40 | $2.00 | Very natural micro-expression; no quality or aspect control. |
| PixVerse v5.5 | 720p / 5s | 35 | $1.75 | Stylised and social-first looks. |
| Vidu Q3 Turbo | 720p / 5s | 20 | $1.00 | Cheap alternative when Runway output feels stiff. |
Runway Gen-3 is the only option in the list that exposes both a quality switch (720p / 1080p) and the full aspect-ratio set of 16:9, 4:3, 1:1, 3:4 and 9:16, at 5 or 10 seconds. Kling V2.5 Turbo has neither a quality nor an aspect control — you get duration only. Full spec sheets sit on the Runway Gen-3 page and the Kling V2.5 Turbo page.
Step by step
- Open the Wink tab.
- Upload your photo under Upload Photo. One image only — Wink is a single-image effect.
- Pick a template: Playful, Flirty or Cute. The prompt box fills in.
- Edit the prompt if you want a specific eye or a specific mood. Example: playful wink with the left eye, closed-mouth smile, soft window light, shallow depth of field, cinematic.
- Leave the model on Runway Gen-3 for the first attempt. Set quality to 720p and duration to 5 seconds — 10 credits.
- Set the aspect ratio to match where the clip is going. 9:16 for vertical feeds.
- Generate, then watch the clip at full size before spending anything more.
- If the face held up, re-run the same prompt at 1080p, or on Gen-4.5, for the version you actually publish.
That test-then-commit order matters. A 720p Gen-3 test is 10 credits; a Gen-4.5 clip is 48. Finding out on the cheap run that your source photo has a hair strand across the eye saves you 38 credits.
Prompting a better wink
Four things move the needle, in order of impact:
- Name the eye. Left or right. Without it the model picks, and it may pick differently on each re-run.
- Name the mouth. Closed-mouth smile, slight smile, or wide grin. Leaving it unspecified is how you get wandering teeth.
- Keep the lighting word from the template. Soft, warm or natural lighting anchors the render to your photo instead of relighting the face.
- Say what should not move. Head still, camera static, no background change. Short clips look better when only the eye moves.
What does not help: piling on adjectives, or asking for camera moves in a 5-second expression clip.
When to pick a different effect
Wink is narrow by design. If you want the head to turn, a full-body movement, or a change of setting, a different tab fits better. Blowing Kiss covers the same flirty register with more motion. Hair Flip and Twirl move the whole upper body. If what you actually want is a photo that simply comes alive with breathing and small head motion, use the general image-to-video tab rather than an effect preset. And if you are only experimenting, start on the free generator before you buy credits.
One more practical note: clips generated on the free tier carry a watermark. Any paid plan downloads clean, with commercial rights — the details are on the no-watermark page and the pricing page. Credit packs start at $2.49 for 75 credits and never expire, which is seven 720p Gen-3 winks with change left over.
Frequently asked questions
How much does one AI wink video cost?
10 credits, or $0.50, on the default setup: Runway Gen-3 at 720p for 5 seconds. Your first one is free because new accounts start with 10 credits and no card. At 1080p it is 20 credits ($1.00).
Can I choose which eye winks?
Yes, by writing it into the prompt — for example, playful wink with the right eye. There is no separate toggle for it, and if you leave it out the model chooses on its own.
Does the Wink effect need two photos?
No. Wink takes exactly one image. The two-photo effects on VdoBloom are the interaction ones, like Handshake and Bridal Carry.
Why does the face look slightly different from my photo?
Because the model regenerates every frame rather than editing your original. A sharper, better-lit, closer-cropped source photo reduces the drift, and Runway Gen-4.5 at 48 credits holds identity noticeably better than Gen-3 at 10.
Can I use a photo of someone else?
Only with that person’s consent. Uploading a stranger’s portrait, or a public figure’s, to make them wink is not something we support, and explicit or illegal content is blocked outright regardless of model.
Does the clip have sound?
No. Wink outputs silent video, so add music in your editor afterwards.
Ready to try it?
Create your first AI video in minutes — no credit card required.
Start Creating Free →