Before you start
- Attach your reference photo and replace the details in square brackets.
- Each prompt describes a complete slide instead of a vague before-and-after effect.
- Use these in your image editor; the studio’s built-in AI uses preset themes rather than free-text prompts.
The formula: reference + edit + details to preserve
To make AI vs reality photos, start by naming the exact transformation: a different jacket, a specific hairstyle or a small accessory on your pet. Then list the parts that should stay stable. This gives you a clear request against which to judge the output.
Avoid stacking conflicting styles. A prompt that asks for a casual snapshot, a luxury campaign and cinematic lighting may reinvent the scene when you only wanted a new shirt. A precise change is also easier to explain with a one-line confession when it disappears.
The prompts below are original example storyboards. Copy them into an image editing tool that accepts a reference photo, and replace the bracketed details before generating. They are starting points, not guarantees of face consistency or a particular result.
Storyboard 1: an everyday portrait, three confessions
Choose a photo with the face and upper body visible. This storyboard changes the background, jacket and hairstyle. The first slide includes all three edits, the second keeps two, and the third keeps one. Your original photo supplies the fourth slide.
Attach the same reference to all three requests. Use the same jacket and hairstyle descriptions each time to reduce differences in how the model interprets them.
01 · The full edit
Edit the reference photo. Preserve the person’s identity, expression, pose, framing and proportions. Change only these three elements: replace the background with [a studio with a plain blue wall], replace the upper garment with [a navy jacket], and change the hairstyle to [a high ponytail]. Keep the lighting consistent with the photo. Do not add text, logos, jewelry or other people.
02 · Restore the background
Edit the same reference photo. Preserve the original background, identity, expression, pose, framing and proportions. Change only the upper garment to [a navy jacket] and the hairstyle to [a high ponytail]. Keep the original lighting. Do not alter facial features or add objects or text.
03 · Only the hairstyle remains
Edit the same reference photo. Change only the hairstyle to [a high ponytail]. Preserve the original clothing, original background, identity, expression, pose, framing and lighting. Do not smooth the skin or change facial features. Do not add text or accessories.
Storyboard 2: your pet’s imaginary life
For a pet, choose recognizable accessories and keep the animal’s anatomy intact. One storyboard could show a gold cushion, a scarf and a medal. The cushion disappears first, then the medal, and the original photo finally reveals the pet without the scarf. The joke is the invented story, not a different animal.
This prompt prepares the first frame. For the next two, remove one transformation from the list and ask for that part of the original to stay unchanged. Check the ears, paws, coat markings and direction of the gaze.
Pet · The full edit
Edit this photo of my pet. Preserve its species, anatomy, fur, markings, expression, pose and framing. Add a small blue scarf and a round medal with no lettering. Replace only the surface the pet is resting on with a gold cushion; preserve the rest of the background. Use lighting consistent with the reference. Do not add human hands, extra limbs, text or logos.
Short captions that follow the visible change
Write the caption after approving the image. If the output changed something other than what you planned, fix the edit or revise the storyboard. Do not keep a confession that no longer matches the photograph. The opening line creates curiosity; the later lines point to the change.
These are editable examples. Match the voice to the person or pet in your photo. “I’m so sorry, I lied” can introduce the first slide, but there is no required wording. A natural sentence is more useful than forcing the same phrase into every frame.
| Slide | Portrait | Pet |
|---|---|---|
| 1 · The edited version | I’m so sorry, I lied. | Sorry, his life isn’t really like this. |
| 2 · First reveal | That place wasn’t real either. | The throne came from AI. |
| 3 · Second reveal | The jacket was AI too. | He didn’t win that medal either. |
| 4 · Original | This is actually me. | This is actually my dog. |
What to fix before opening the studio
If the face changes, return to the source and narrow the request to one visible edit. More descriptive words do not always help; they can introduce a new interpretation. If the background shifts, ask to preserve its lines, perspective and object positions.
Reject an attempt when an accessory cuts through hair or a paw. A caption cannot repair broken anatomy. If all three images look nearly identical, strengthen one concrete edit instead of changing the whole visual style.
Once approved, save the versions in a clear order and place the original last. The manual editor lets you review the captions and export the four slides. Music selection and publishing happen in TikTok.
Put the story together
One original. Your photo reveal.
Start with one photo and a template, or use existing edits in manual mode. Review the three edits, keep the original last and download a captioned MP4 or PNGs. Export is free; AI generation uses credits.
Open Perdón Studio