Quick answer
Two prompts. The portrait one goes into Gemini or ChatGPT with your selfie and gives a still of you as a witch among moonlit trees. The spin one goes into an image-to-video tool and asks for the change on camera. Witch Transformation AI does the spin from one photo with no prompt.
Both prompts keep your face and hair, which is what makes it your witch rather than a stock one. Both are written to work for anyone; nothing in them assumes a gender.
The witch portrait
Keep the face, hair and skin tone of the person in this photo. Dress them as an elegant witch: a tall black pointed hat sitting on top of their hair, a long black cloak with a high collar. A soft green glow rises from one open palm. Moonlit birch forest, low mist, a full moon behind the trees. Realistic photo, cool blue night light with green light on the face. Vertical 4:5.
“Sitting on top of their hair” is the line people leave out and regret: without it, the hat often replaces the hair. The green palm light gives the face a color that a plain night scene doesn’t.
The spin video
Vertical 9:16 video, 8 seconds. The person in the photo begins as they are, in everyday clothes, facing the camera in soft light. They turn once, and as they turn a tall black pointed hat and a flowing black cloak swirl into place. Green sparks curl up from their hands. When they stop, they stand as an elegant witch in a misty forest under a full moon. Enchanting, not scary. Keep the face exactly as in the photo. No text.
- “Begins as they are”: no before, no transformation. Tools happily start in costume otherwise.
- “They turn once”: the turn hides the outfit change, the way the cut does in the filmed version.
- “Green sparks”: the magic makes it a spell instead of a wardrobe swap.
- “Enchanting, not scary”: keeps the light soft and the face friendly.
Three other witches
Swap one of these into either prompt in place of the forest.
Storybook
A green tint and a broom.
their skin takes on a faint green tint and they hold a crooked wooden broom
Potion kitchen
Indoors, by a cauldron.
a candlelit stone kitchen with a bubbling iron cauldron, green steam curling up
White witch
Light rather than dark.
a silver-white cloak and hat, pale blue sparks, a snowy pine forest
Which tool for which prompt
Chat apps make the portrait; they don’t turn it into the spin. A video tool makes the spin from the second prompt, usually after a few attempts, since keeping your face through a turn is hard. Witch Transformation AI has the spin, the timing and the forest set up and charges 80 credits per video.
Common problems
My hair or face changes
Add “same hair, same face”, and say the hat sits on top of the hair, not instead of it.
It starts with the witch already dressed
Give the opening to you as you are: “begins in everyday clothes”, as its own sentence before the turn.
The clothes blur instead of swirling on
Keep “they turn once”. A single turn gives the tool a place to make the change.
FAQ
What prompt makes an AI witch transformation?
For a still: keep your face and hair, add a tall pointed hat on top of the hair, a long black cloak and green light from one palm, in a moonlit forest. For video: start in everyday clothes and turn once. Both are above.
Can men do the witch transformation?
Yes. The prompts are written for anyone, and the template works the same on any face.
Will ChatGPT make the video?
Chat apps are reliable for the still. The spin needs a video tool or a ready template.
Do I need a prompt on Reelfie?
No. Add one photo of yourself to the template and it handles the rest.
Make your first one
Witch Transformation AI: Become a witch in one spin.
Try Witch Transformation AIYour photos are deleted 7 days after upload and never used to train AI.
