Guide
Oh You Can Make It Cap AI Prompts: The Full 11-Shot Prompt
These oh you can make it cap AI prompts let you rebuild the 15-second studio meme in any video tool that accepts reference images. Copy the full text below, attach your photos and run it. The rest of the guide explains why each line is there, and which fixes came out of our own test runs.
Last updated: Oct 29, 2026
Key facts
- The oh you can make it cap AI prompt in this guide covers 15 seconds of video in 11 shots, with one person in each shot.
- Two people take turns, and the clip ends on the first one.
- A test plan with only 7 shots caught just four of the meme's ten cuts, and it ended on a ghosted double.
- One test run invented a two-person split screen at the end, so the text bans it.
- Adding a reference video fixed the ratio at 4:3 in our test and made one face drift from its photo.
The oh you can make it cap AI prompt, shot by shot
Below is the oh you can make it cap AI prompt we use, lightly rewritten for public use. It covers the studio, the 11 shots with their timings, the audio and the face rules. Paste it into a video model that accepts reference images and, optionally, reference audio.
Attach the photo of person A first and the photo of person B second. The prompt calls them image 1 and image 2. Keep that order, or the roles swap.
Realistic live-action footage in a bright seamless white photo studio. Soft, even light. Medium shots from the knees up, camera at chest height.
Length: 15 seconds. 11 shots. Every shot has one person in frame, alternating between A and B, starting and ending on A.
Every cut is a hard cut, except two quick dissolves at 12.2 s and 14.45 s.
Never use a split screen. Never put two people in one frame. No face close-ups.
A is the person in image 1. B is the person in image 2.
Shot 1 (0.00-1.93 s) A: grins, raises one arm high and points off-camera in disbelief.
Shot 2 (1.93-3.17 s) B: turns sideways and spins, looks back over the shoulder with a smug face, one hand holds a small object.
Shot 3 (3.17-5.07 s) A: bites a finger, trying not to laugh, eyes wide.
Shot 4 (5.07-6.27 s) B: faces the camera with both hands at chest height, talking with confidence.
Shot 5 (6.27-8.63 s) A: covers the mouth with one hand and laughs, shoulders shaking.
Shot 6 (8.63-10.00 s) B: holds the small object and keeps talking to someone off-camera.
Shot 7 (10.00-11.47 s) A: raises a hand beside the face, laughing and talking back.
Shot 8 (11.47-12.20 s) B: stands with the small object and a smug look. A quick dissolve leads to shot 9.
Shot 9 (12.20-13.77 s) A: grins and gestures with one hand, giggling.
Shot 10 (13.77-14.45 s) B: talks with the small object in hand.
Shot 11 (14.45-15.00 s) A: a quick dissolve back to A, standing full body and talking until the end. The video ends on A alone, never on B.
Audio: an upbeat meme beat. A voice says "oh you can make it cap" around the spin. Real stifled laughter in the A shots.
Keep each face, hairstyle, skin tone and outfit identical to the reference image. No text, no captions, no logos, no watermark.How to run the oh you can make it cap AI prompt
The prompt names no aspect ratio and no quality level. Those belong in the settings of your tool. Run the oh you can make it cap AI prompt unchanged the first time. Then change one thing per run, so you know what caused each difference.
- Upload the photo of A as image 1 and the photo of B as image 2.
- Paste the prompt and set the length to 15 seconds.
- Pick 16:9 or 9:16, and turn on sound if the tool has that option.
- Generate the clip, then check the cuts against the shot times.
Using only one photo
With one photo, change a single line. Replace the sentence that names the two people with the sentence in the block below. The shot list stays as it is.
The result still has two roles, but both wear your face. In our test with one portrait, the roles were easy to tell apart.
A and B are both the same person from image 1, one person playing both roles.Why each part is written this way
Every line of this oh you can make it cap AI prompt answers a problem from our tests. Knowing the reason helps you edit the wording without breaking it.
The white studio and the framing
A plain white seamless studio with soft light gives the model nothing to invent. There are no props, no extra people and no hard shadows to flicker between frames.
Knee-up framing shows the arms and the torso. The comedy lives there: pointing, a hand over the mouth, a spin.
Eleven shots, a single person per shot
The shot count matters most. Our first plan listed only 7 shots. The result hit 4 of the 10 cuts in the meme, and the ending showed one person dissolving into a copy of themselves.
Listing all 11 shots with start and end times gives the model a cut to aim for. Each role returns every other shot, and the last one belongs to A. The ghost in our test was B dissolving into B, so the prompt says never to end on B.
Hard cuts and two short dissolves
Naming only two dissolves tells the model that every other transition is a plain cut. When a model blended cuts on its own, our results showed ghosted frames mid-clip.
The two dissolves sit at 12.2 seconds and 14.45 seconds, because the meme itself softens those two moments.
No split screen and no close-ups
One test run ended with a split-screen close-up of two faces. Nothing in our prompt had asked for it. We added three bans and one positive rule: every shot shows exactly one person.
Our later example, made with this wording, had no shared frames and no ghosting.
Locking the faces
One line tells the model to match each person's look to the reference image: face, hair, skin and clothes. Fast cuts make faces drift, and this line pushes back. The photos do the real work, though.
In our example run, both people matched their photos in all 11 shots. The same line also bans text, captions, logos and watermarks.
The sound
The audio line asks for an upbeat beat, the spoken line and stifled laughter in the shots of person A. Treat it as a request, since models follow it loosely.
Our own tool adds two steps around it. It passes a reference soundtrack to the model. After generation it lines the sound up with the cuts and levels it to about -14 LUFS. In one run the sound moved 0.35 seconds earlier to fit.
How to prepare the reference images
Photos decide most of the likeness, and no wording can rescue a weak photo. Use the rules below for both images. They match what our own photo check accepts.
Check your tool before you upload people. In our testing, one video tool refused photos of real people as reference images, while another accepted them.
- One adult per photo, facing the camera.
- A sharp face and even light, with no heavy filter.
- Head and upper body in frame, so the outfit carries over.
- A big, clear face for A, because the laughing shots are the closest.
- A clear upper body for B, because the spin shows hair and clothes.
Common problems and the fix
Most failures in our runs fell into a few groups. Each has a short fix in the wording or in the settings.
- The ending splits the frame between two faces. Keep the bans, and keep the line that ends the clip on A alone.
- A ghost double appears near a cut. Keep the full shot list with times, and do not add extra dissolves.
- Cuts land late or are missing. Count the shots in your prompt, since a 7-shot plan gave us only four matching cuts.
- A face drifts away from its photo. Use a bigger, sharper photo with one adult in it.
- The last half second turns into a wide shot that shows the dark edge of the studio. This appeared in one of our four runs.
- The frame ratio is not what you asked for. A reference video can lock the ratio to its own shape.
How to check the cuts in your result
Open the clip in an editor and compare it to the timings. A scene-detection tool can find the cuts for you, but the setting needs care.
In a white studio the frame barely changes at a cut, because both people stand against the same background. A threshold of 0.2 missed most cuts in our clips. At 0.08, the tool found them. Expect a few false hits and a few misses, and check each cut by eye.
Vertical 9:16 and other formats
No ratio appears in the wording. Choose vertical or horizontal in the tool, and leave the wording alone. Our own tool sends the same words for both formats.
A horizontal clip can also become vertical in an editor. Stack it, or place it over a blurred copy of itself. Our video page offers both versions as downloads.
A reference video is the one setting that changes the shape. In our test it forced the ratio to match the reference, which was 4:3. For that reason we do not use one.
If you would rather skip oh you can make it cap AI prompts
Our tool runs the same prompt for you. Upload one or two photos on the home page, pick a vertical or horizontal format, and choose 480p or 720p. A 480p video uses 3 credits and a 720p video uses 6. At current pack prices that is $4 to $6 and $8 to $12.
The step-by-step guide shows real timings. The meme guide explains each of the 11 shots. If a video comes out wrong, see why a video fails and how to fix it.
FAQs
What are oh you can make it cap AI prompts?
Oh you can make it cap AI prompts are written instructions for recreating the 15-second white-studio meme with a video model. The prompt on this page lists 11 single-person shots with times, audio notes and face-keeping rules.
Which tools can use this prompt?
The oh you can make it cap AI prompt above suits any tool that takes reference images, with or without reference audio. A tool that takes text only cannot use your photos, so the faces will not match.
How many shots should an oh you can make it cap prompt have?
An oh you can make it cap prompt should list all 11 shots. Our 7-shot plan caught only four cuts, and its ending showed a ghost of person B.
Why are split screen and close-ups banned?
Split screen and face close-ups are banned because a run of ours added two faces side by side at the very end. The bans, plus the rule of one person per shot, stopped it in our later runs.
Do I need a reference video?
A reference video is not needed. In our test it put the cuts closer to the meme, but the frame locked to 4:3 and a face looked less like its photo.
Does the prompt change for a vertical video?
The same AI video prompt works for vertical clips. Set 9:16 in the ratio setting of your tool, and keep the wording as it is.
Do I have to write a prompt in your tool?
Oh You Can Make It Cap AI does not ask you for a prompt. You add a single photo or a pair, pick the format, and the tool sends the prompt for you.