Step-by-step guide · Verified product workflow

How to Make an AI Dunk Video From Your Photo

This AI Dunk video tutorial shows how to choose a play, upload one or two authorized photos, select a court and quality, review the vertical result, and download the completed MP4.

Start by choosing the play because it determines the photo count and duration. Add one clear person per photo, choose Auto or a named court, select 480p or 720p, confirm permission, and submit. When the task is complete, inspect and download the vertical MP4.

Open the AI Dunk generator Read the five steps
Verified One-Hand Slam Dunk result
This is the same accepted generation documented below, shown here for a quick look at the solo play’s action and framing. The source portrait, settings, review date, and evidence limits appear in the verified input-and-output record.
Inputs1 or 2 authorized photos
PlaysSolo or two-person face-off
Output8- or 12-second vertical video
Quality480p: 50 credits · 720p: 100 credits

01 · Understand the result

What does the AI Dunk generator make?

This AI Dunk video tutorial also works as an AI dunk basketball tutorial for the guided photo-to-video workflow. AI Dunk turns one or two authorized photos into a 9:16 MP4 at Standard 480p or HD 720p. One-Hand Slam Dunk uses one photo and lasts 8 seconds; Middle Finger Dunk uses two photos and lasts 8 seconds; Dunk on Your Friend uses two photos and lasts 12 seconds.

Your inputWho appears

One dunker photo for the solo play, or one dunker and one defender photo for a face-off.

Selected playWhat should happen

The intended duration, number of people, dunk action, camera framing, and reaction beat.

The result is AI-generated, not real game footage. The action reference shows the target action style; it does not prove that every face, limb, ball path, or court detail will be preserved.

02 · Before you start

What photos should you prepare?

Use a sharp front or three-quarter photo with one person, a clear uncovered face, and enough body visible to guide the basketball pose. Avoid group photos, heavy blur, mirrored screenshots, and phones covering the body.

Upload only photos you own or have permission to use. Do not use public figures, deceptive impersonations, or prohibited images of minors.

Photo requirements for each current play
PlayPhotosWhat they guide
Middle Finger Dunk2 requiredThe featured dunker and defender
Dunk on Your Friend2 requiredThe featured dunker and defender
One-Hand Slam Dunk1 requiredThe featured dunker
AI Dunk generator showing the three play choices and their photo counts
Choose the play first; it sets the number of photos and the 8- or 12-second duration.
AI Dunk dunker upload area with photo quality guidance
Upload one clear person per role. A two-person play adds a separate Defender Photo field.
AI Dunk court cards and Standard 480p and HD 720p quality controls
Select Auto or a named court, then choose 480p for 50 credits or 720p for 100 credits.

One person per photo keeps the role assignment clear. It does not guarantee exact likeness in every generated frame.

03 · Follow the workflow

Make your AI Dunk video in five steps

If you are looking for how to make an AI dunking video, this AI Dunk video tutorial uses the same five steps shown on phone and desktop.

  1. 01

    Choose the dunk play

    Pick Middle Finger Dunk, Dunk on Your Friend, or One-Hand Slam Dunk. The selected play controls whether the form needs one photo or two and whether the resulting video is 8 or 12 seconds.

  2. 02

    Add the required photos

    Upload one clear person per photo. The solo play needs one dunker photo. The two face-off plays need a dunker and a defender. Use only photos you own or have permission to use.

    One personClear faceVisible bodyGood light

    Selecting and reviewing photos does not itself create a video.

  3. 03

    Choose the court

    Keep Auto selected or choose Pro Arena, Office Gym, or Outdoor Community Court. These are product options; final environment consistency still depends on generation.

  4. 04

    Choose quality and confirm permission

    Select Standard 480p for 50 credits or HD 720p for 100 credits. Review the setup and confirm that every featured person is eligible and authorized.

    See the current video packs before choosing a quality.

  5. 05

    Sign in, create, and review

    Sign in if asked, submit the prepared setup, then follow the existing task. After it completes, watch the entire clip and download the MP4 only when the result is suitable.

Create an AI dunk video from your photo

04 · After you submit

Read the task state before trying again

Keep the same account and check the active result or the signed-in My Creations task list. Refreshing a task is different from submitting another generation. Completion time varies by play and provider load, so no fixed number of minutes is promised.

What to do at each stage
What you seeWhat to do
Queued / preparing referencesThe task is assembling the authorized inputs. Keep the existing task open and do not submit a duplicate.
GeneratingThe video model is processing the selected play. Wait or return to the same task later.
FinalizingThe result is being validated, combined with authorized template audio, and prepared for delivery.
CompletedWatch the full result, inspect the people and action, then download only if it is suitable.
Terminal failureThe reserved credits are released. Check the task message and input before deciding whether to try again.
Submission reviewThe task is under submission review. Do not submit the same setup again until that review is resolved.

05 · Verified input and output

What did one real AI Dunk generation produce?

This AI Dunk video tutorial separates the source from the finished file. The source portrait below is an AI-generated fictional adult chef. The video is the actual accepted AI Dunk result generated from that exact input on October 26, 2026. It used One-Hand Slam Dunk, Pro Arena, Standard 480p, and an 8-second output.

AI-generated fictional chef portrait used as the One-Hand Slam Dunk input
Input · AI-generated fictional adult · one authorized test file
Output · actual unedited AI Dunk generation · 480×854 · 8 seconds · H.264 video with AAC audio

The review found one featured player, a visible professional arena, and no visible text or watermark in sampled frames. The file passed the publishable sample review. This is one controlled result, not a success-rate claim or a promise that every face, hand, ball path, court, or landing will look the same.

06 · Review the result

Watch the whole video before sharing

Use this AI Dunk video tutorial checklist at the beginning, during the airborne action, and after landing. Check the featured people. Confirm that the ball, hoop, and ending remain readable; listen for the expected audio; and look for extra limbs, body overlap, sudden clothing changes, brand marks, captions, or a visible watermark.

The face changes

Use a sharper, front or three-quarter photo with an uncovered face. Generation can still change details during turns, jumps, or contact.

The ball or hand looks wrong

Review the gather, airborne contact, rim finish, and landing. Basketball motion places hands, ball, bodies, and hoop geometry close together, so artifacts can appear.

The defender overlaps the dunker

Use separate photos with one person in each. A clear source helps role assignment, but it cannot guarantee clean body separation in every frame.

The court differs from the choice

A court option guides the intended environment. Lighting, crowd, floor markings, and background details may still vary.

A second render is a separate attempt and may fail differently. Change the source photo only when the review points to a specific input problem.

07 · Quick answers

Common AI dunk video questions

Is one photo enough?

Yes for One-Hand Slam Dunk. Middle Finger Dunk and Dunk on Your Friend each need one dunker photo and one defender photo.

Which play needs two photos?

The Middle Finger Dunk generator and Dunk on Your Friend need two separate authorized photos. One-Hand Slam Dunk needs one photo.

Do I need a written prompt?

No. The guided creator uses the selected play, photo roles, court, and quality instead of a written prompt. If you want to control the action manually in another compatible tool, use the AI dunk prompt guide.

Do I need a reference video?

No. The public creator does not ask you to upload a video. The selected play supplies the intended action inside the product workflow.

Can I make it on a phone?

Yes. The form stacks the play, photo, court, and quality controls for a mobile screen. Use an original photo rather than a compressed screenshot when possible.

Does the output include sound?

Completed AI Dunk files include the authorized template audio as an AAC audio track. The provider-generated audio is not used.

Does AI Dunk add a watermark?

No visible watermark was observed in the accepted test samples reviewed on October 27, 2026. AI output can vary, so inspect the completed file before sharing it.

Can I download the MP4?

Yes. After a task reaches completed status, watch the result and download the vertical MP4 from your signed-in account.

How long does generation take?

There is no fixed published completion time. Timing varies by the selected play and provider load, so follow the existing task instead of submitting a duplicate.

What happens to credits when generation fails?

A terminal failure releases the credits reserved for that generation. If submission status is unclear, wait for review rather than submitting the same setup again.

Registration includes no free credits. The smallest current one-time pack is $6.99. Review the AI Dunk video pricing before generating.

Review the finished file before sharing, label AI-generated content where a platform requires it, and follow the terms and privacy policy.

Your turn

Choose the play.
Add your photos.

Start with one or two authorized photos, depending on the play. Choose the court and quality only after the input roles are clear.

Create an AI Dunk video from your photo