What is the Tiramisu Fridge AI generator?
Tiramisu Fridge AI turns one photo into the viral fridge-POV meme: upload a portrait, and the model replaces the tiramisu dessert in a fixed refrigerator clip while keeping the camera, scooping motion, shelves, and original audio.
This is a strict subject replacement on a fixed reference clip — not a text-prompt recreation. Identity comes from your photo; timing and fridge POV stay locked to the sample.
The preview plays the After example (sample swap) first, and both full clips — Before (original reference) and After — are listed underneath, so you can compare before you generate.
The reference clip’s audio is kept. You can still swap tracks in TikTok, Reels, Shorts or CapCut after download, and label the video as AI-made when the platform asks.
How this generator works
Upload one photo and press Generate. We charge 600 credits per render, send your image plus the bundled Tiramisu Fridge reference to Wan 3.0 reference-to-video, and return a finished MP4 in the browser.
The reference clip is fixed and hosted with Vocalove, so every run follows the same fridge POV and scooping motion. You do not upload a video.
Rendering usually takes several minutes. Keep the tab open while the job runs — the download button appears as soon as the file is ready.
How to make a Tiramisu Fridge AI video
- 1
Step 1: Upload a clear, front-facing photo
Use a sharp JPG, PNG or WebP under 10 MB. Face clearly visible works best for the fridge-tray swap; avoid heavy filters and group photos.
- 2
Step 2: Preview Before / After, then check the cost
The preview plays the After sample, with both full Before and After clips listed below. The generate button shows 600 credits before you submit.
- 3
Step 3: Generate, preview and download
Press Generate and keep the tab open while the render runs. Preview your MP4, download it with the original audio, and mark the clip as AI-made where required.
Each render costs 600 credits. Compare plans on the pricing page.
Photo requirements for the fridge swap
- One person, face clearly visible and not covered.
- Even lighting and a simple background.
- No screenshots, stickers or watermarks over the face.
- Formats: JPG, PNG or WebP, up to 10 MB.
Only upload photos of yourself or people who gave you permission to animate them. Do not use images of public figures or anyone else’s likeness without consent.
Ready to generate? Back to the tool

