The best photo for The Box AI trend is one sharp head-and-shoulders photo of one person, face forward, in even light, with your hair visible and nothing over your face. That single photo has to hold up in three very different scenes: a dim, close seated shot at a cash table, a wide night shot in an industrial lot where two cars crash behind you, and a front-on podium shot at the end. A photo that only works close up, or only works from far away, will break in one of them. This guide checks each scene against the photo you plan to use.
Got the photo? Make the video
Add one photo for the lead rapper, check the credit quote and make the 20-second The Box AI trend video.
What the photo has to do
The Box template on Oakgen is a Kling V4 edit of the real reference clip. It keeps the moves, the camera, the cuts and the timing, and replaces the lead with the person in your photo. The result is a 20-second landscape 16:9 clip in 720p or 1080p.
There is one required upload slot, the lead rapper, plus an optional Background slot for a location photo, covered at the end of this guide. Your photo's face, hair and outfit are what the model uses. The reference lead in the reel is an AI look-alike of basketball star Luka Dončić in a yellow number 77 kit. The template never keeps that look-alike, and the reference is AI-made, not real footage of Luka Dončić or anyone else it resembles.
So the photo has three jobs: a clear face for two close, front-on moments, a hairline and build that read in a wide night shot, and an outfit for every scene, because there is no costume.
Scene by scene: what each shot asks of your photo
The clip was inspected on 18 November 2026. It runs 20.04 seconds at 24 frames per second. Here is what each part does to your photo.
1. The cash table (0 to about 4 s)
A dim room. The lead sits at a glass table stacked with banded cash and lip-syncs "I been movin' 'em out". A man in a grey hoodie counts money, and there is a short close-up of him.
This is the first face people see, and the light is low. A photo in soft daylight, facing a window, gives the model a clean face to carry into the shadows. A dim or orange photo gives it little.
Here is where this breaks: a photo with a hand, a cup or a phone near the mouth. The opening is a lip-sync, and the mouth needs to be clear.
2. The TV and the night lot (about 4 to 16 s)
An old CRT TV shows an empty lot, and the clip cuts into it. In the lot, the lead wears narrow black visor glasses, dances and shadowboxes. Two cars race in from both sides and collide behind them around 10 to 12 seconds. One flips in a fireball in slow motion while the lead walks toward the camera without looking back.
This is the longest stretch, and the lead is small in frame for much of it. Your face matters less here than your silhouette: hair, head shape, shoulders and outfit colour. A photo cropped tight to the chin gives the model your face but no idea of your build or clothes. Include your shoulders and the top of your outfit.
The visor glasses belong to the reference scene. Do not wear sunglasses in your photo to match them. Sunglasses hide your eyes in the two close scenes, which costs more than it gains in the lot.
3. The podium (about 16 to 20 s)
A press-room podium in front of a blue curtain. The lead taps their chest on "I'm a 2020 president candidate". This is the ending, the frame people screenshot, and it is front-on.
A three-quarter or side profile photo is the usual failure here. The model has to invent the half of your face it never saw, and the podium shows it straight on. Face the camera in your photo.
Photo types and what to expect
This table is based on how the template works, not on generated results. No outputs were generated for this article.
| Photo type | Cash table (close, dim) | Night lot (wide) | Podium (front-on) | Verdict |
|---|---|---|---|---|
| Head and shoulders, face forward, daylight | Clear face | Hair and shoulders read | Clear face | Best choice |
| Tight selfie, chin to forehead | Clear face | No outfit or build to follow | Clear face | Usable, outfit is guessed |
| Full-length photo from a distance | Face too small | Outfit and build read well | Face too small | Weak in two scenes |
| Side profile or three-quarter | Half the face guessed | Fine | Front view guessed | Avoid |
| Sunglasses or tinted lenses | Eyes hidden | Fine | Eyes hidden | Avoid |
| Hat or hood up | Hairline hidden | Head shape changes | Hairline hidden | Avoid |
| Night or club photo, harsh flash | Muddy face in a dim scene | Fine | Flat, overlit face | Retake in daylight |
| Group photo, several faces | Model may pick the wrong face | Same | Same | Crop to one face |
If your only photo is a tight selfie, use it, and expect the outfit in the lot to be a guess.
Your outfit replaces the basketball kit
The reference lead wears a yellow number 77 basketball kit. Your video does not. The template uses the outfit in your photo for every scene, and it asks for no team lettering, numbers, seals or logos.
That has two practical effects:
- Pick the outfit on purpose. You wear it at the cash table, in the lot and at the podium. A plain dark hoodie or a plain jacket fits all three. A suit makes the podium funnier and the lot more absurd. A jersey of your own team will have its lettering and numbers asked away, so it may come out blank.
- Avoid big logos and printed text. Lettering on a shirt is exactly what the template asks to remove. A plain top gives you a cleaner result than a shirt that loses half its print.
Solid colours are the safer pick for a dark lot with fire behind you.
Glasses, hats and beards
Glasses. If you wear clear prescription glasses every day, keep them on. People know you with them. Pick a photo where the lenses do not reflect light over your eyes. Sunglasses and tinted lenses are the problem, not frames.
Hats and hoods. Take them off. The model follows your hairline and head shape in all three scenes, and a cap hides both.
Beards and moustaches. Keep them as they are. A beard is part of your face, and a clear, front-on photo shows its shape. A very new or patchy beard can blur in the dim cash-table scene, so a photo in good light matters more.
Groups, couples and pets
The Box template has one role, the lead rapper, and is written for one person. The man counting money is part of the reference scene, not a slot you can fill.
- A group photo risks the model taking the wrong face. Crop it to one person before you upload, or pick a solo photo.
- A couple photo has the same problem. Make the video for one of you. If you want a duo, the Bandit Rap Duo trend and the 8 Mile Rap Battle trend each take two photos.
- A pet photo is accepted by some other one-photo Kling trends on Oakgen, but we have not confirmed it for The Box. The clip is built on a person's chest tap, dance and walk. Use a photo of a person.
Use your own photo or one of someone who agreed to appear, never a photo of Luka Dončić or another public figure.
Choosing a background photo (optional)
The Background slot is for a place, not a person. Leave it empty and you keep the dim room, the night lot and the press room from the reference. Add a location photo and the template tries to rebuild all three scenes in that one place, at the same time of day and in the same light, while the table of cash, the old TV, the two cars and the fireball, and the podium with its microphone stay. It is a rebuild of the place, so check the result more closely than with the original settings.
What makes a good background photo:
- Wide and landscape. The clip is 16:9, so a landscape photo of the whole place gives the model the most to work with.
- Enough open floor for the car scene. Two cars collide and one flips behind you, so the place needs a large open stretch of ground. An indoor photo becomes a large open floor of that room.
- Empty of people. No one from the photo is added to the video, and people in it only get in the way.
- No readable signs. Shop names, posters and logos are asked to stay unreadable, so a place without them looks cleaner.
- Even light. All three scenes take their light from this one photo, so avoid a frame that is half in deep shadow.
A background photo adds no cost.
Run your best photo through all three scenes
One photo, 20 seconds, 16:9. The template page shows the credit quote before you generate.
A 30-second photo check
Before you upload, check for one face looking at the camera, both eyes and the whole mouth visible, daylight or soft indoor light, your shoulders and outfit in frame, an outfit you would wear at a podium, and no filter. Six yeses and you are ready. Then choose 720p or 1080p on the The Box trend page. The page shows the credit quote before you generate, and plan details are on pricing. I would run 720p first to check the face in the cash-table and podium shots, then make the 1080p version you post.
About the sound
Your video keeps the original soundtrack, Roddy Ricch's "The Box". It is put back on the finished file after generation and is never sent to the model. The mouth follows the reference clip, not the audio. Oakgen does not license the song for your posts, so check each platform's music rules before you share.
How we put this together
On 18 November 2026 we inspected the reference clip frame by frame and read the template settings. We did not generate any videos for this article, so the table is a prediction from how the template works, not a set of results.
Disclosure: Oakgen is our product, and The Box template is an Oakgen template.
Related reading
- How to make The Box AI trend video
- The Box AI trend explained
- The Box music video, scene by scene
- The Box AI trend caption ideas
- AI rap video trends compared
- 2020 president candidate AI video
- Car crash walk away AI video
- Luka Dončić The Box AI video
FAQ
How many photos does The Box AI trend need?
One is required. It replaces the lead rapper in every scene. The look-alike in the reference reel is never kept. A second, optional Background photo moves all three scenes to your place.
What is the best photo for The Box AI trend?
A sharp head-and-shoulders photo of one person, face forward, in even light, with hair fully visible and nothing covering the face: no hat, no sunglasses and no hand near the chin.
Do I need a yellow basketball kit?
No. Your outfit replaces the kit in every scene. The template asks for no team lettering, numbers, seals or logos, so the reference kit is never copied.
Can I wear glasses in my photo?
Clear prescription glasses you always wear are fine if the lenses show your eyes. Avoid sunglasses and tinted lenses, because the cash table and the podium are close, front-on shots where your eyes carry the likeness.
Can I use a group photo or a pet photo?
The template is written for one person in one role. Crop a group photo down to one face, or use a solo photo instead. We have not confirmed pet photos for this template, so use a photo of a person.
Did Oakgen generate the examples in this guide?
No. No outputs were generated for this article. The advice comes from how the template works and from the reference clip, which was inspected on 18 November 2026.


