The torn paper effect AI trend turns one portrait into a still image that looks ripped by hand, with a galaxy, a sunset or a second exposure showing through the tear. It has no sound and is not a video. This guide goes deeper than the page FAQ, with which photo suits which style, why a tear looks fake, how it differs from a double exposure and when to add an AI label.
Key takeaways
- The result is one silent vertical 2:3 still image, not a video.
- Galaxy portal keeps your place and pose; sunset and strips replace your background.
- A believable tear needs fibres, thickness, curled flaps and matching shadows.
- Label the picture as AI where the platform asks, especially when it shows a real person.
- One image is 1 credit, with packs from $6.99 for 10 credits.
In this guide
- Which photo suits which of the four styles?
- Why does the tear look fake or like a cut-out shape?
- How is torn paper double exposure different from torn strips?
- Should I do it in Photoshop or Canva instead of AI?
- Why do hands and small details sometimes come out wrong?
- Why does the AI keep the tear away from my face?
- Why do prompts copied from tutorials drift away from my photo?
- Is the result a video, and can I add a sound?
- Do I have to label a torn paper picture as AI?
- What does it cost, and what if a result fails?
Which photo suits which of the four styles?
Match the style to what your photo shows. Galaxy portal keeps your place and pose, while sunset and double exposure strips replace the background, so those two care less about the scene behind you.
Galaxy portal is the one that keeps your own place and pose, so it rewards a photo where you already stand somewhere interesting and your body is clearly visible. A full or half-length shot with room around you gives the tear space to open.
The sunset style replaces your background with a golden sky showing through the paper, so a plain wall or a messy room behind you does no harm. What matters is that your face is sharp and lit from the front.
The strips style shows close-ups, so it does not need your whole body. A tight, clear face photo is enough, and it is the easiest style to feed with a selfie. Torn pieces on coloured paper sits in between: a clear face and a simple background are the safest start. You can try all four on the torn paper effect AI page.
Why does the tear look fake or like a cut-out shape?
A fake tear is almost always flat. Real ripped paper has rough fibres, visible thickness, curled flaps and shadows, and when a prompt or tool skips those, the edge reads as a smooth cut-out sticker.
The part that sells the effect is the paper itself. A torn edge of real thick paper has a pale, hairy fringe where the layers split, and the flap casts a soft shadow on whatever lies beneath. Leave those out and you get a clean wavy outline, which the eye reads as a digital shape.
Shadows are the second giveaway. They have to match the scene: a golden sunset behind the tear should warm the edge, and a purple galaxy should not leave a grey studio shadow. When the light on the paper disagrees with the light on the person, the picture looks pasted together.
This is why the page builds fibres, thickness, curled flaps and matching shadows into every style, instead of asking for a generic torn look. If you still get a flat result, generating again is cheap at 1 credit, because the tear is different each time.
How is torn paper double exposure different from torn strips?
A true double exposure blends two images into one, while the torn strips look places the same person in separate panels split by white rips. They share a name online, but the structure is different.
A camera double exposure lays one picture over another so the shapes show through each other. The look people call torn paper double exposure does something else: it cuts the picture into panels with white paper rips between them, often with one colour portrait between black and white close-ups.
On videosi this is the double exposure strips style, one of four. It is made by AI from a single photo, not in camera, and it does not need your whole body because the panels are close-ups. If you want the galaxy or sunset showing inside a silhouette instead, that is a different look, and the galaxy portal or sunset styles are the ones to choose.
Should I do it in Photoshop or Canva instead of AI?
Do it by hand if you want full control over every edge and have an hour to spare. Use AI if you want a believable tear from one photo in about a minute with no editing skill.
The effect itself is an old design trick, long made by hand in Photoshop and Canva. The manual route is to cut your subject out, draw or import a ragged paper edge, add shadows, and place the second image behind it. Editors describe masks as a way to hide part of a layer without deleting its pixels, with black areas of a mask hiding pixels and white areas revealing them, so you can reshape the tear later without losing anything (Photopea's guide to masks explains this).
The trade-off is time and skill. Hand-made tears give you exact control, but convincing fibres and matching shadows take practice. The AI route gives you less control over where the flaps fall, but your face comes from your photo and the paper physics is handled for you. Many people try the AI result first and move to manual editing only if they need a specific layout. The torn paper effect AI page is the quick route.
Why do hands and small details sometimes come out wrong?
Because the AI decides where every flap and shadow falls, and fine details such as fingers are the hardest thing for image models to keep consistent. The tear is different each time, so a bad hand does not mean the next result will be bad.
Check the picture at full size before you post it. Look at hands, rings, strands of hair near the tear, and the point where the paper edge meets the body. Small errors there are common in AI images and are easy to miss on a phone screen.
If a detail is wrong, generate again instead of trying to repair it. A photo where your hands are hidden or out of frame, such as a close face shot, avoids the problem entirely, which is another reason the strips style is forgiving.
Why does the AI keep the tear away from my face?
The torn edge is kept clear of the face on purpose, so the tear never cuts away your identity. If paper covered part of your face, the picture would stop looking like you.
Face match is the weak point of every AI portrait. Your photo is the only source for face, skin tone, age, hair and build, and anything that hides or distorts the face makes a match harder. Keeping the rip around the person rather than through the eyes and mouth protects the likeness.
A sharp, front-lit photo with your whole face visible gives the best match. Sunglasses, heavy filters and group shots lower it. Even then, face match is never perfect, and small differences can appear in all four styles.
Why do prompts copied from tutorials drift away from my photo?
Because they were written for someone else's photo. A prompt that describes a stranger's look or setting tells the model to build that stranger, so the result pulls away from you.
Most tutorials we watched use CapCut's AI photo editor with a pasted prompt. Those prompts often describe a particular hairstyle, outfit or room. Paste one with your own photo and the model has two conflicting sources.
Two other problems show up. Some models add a stamp or caption the prompt mentioned, such as a creator logo or a title, which then has to be erased by hand. And beauty retouching creeps in by default, changing the face shape, which is what makes the result stop looking like you.
The fix is to keep the identity rule strict: your photo is the only source for the person, and the prompt only describes the paper, the style and what shows through. That is the structure behind the page, and the full wording stays private.
Is the result a video, and can I add a sound?
No, it is a still image with no sound. The page returns one vertical 2:3 picture per generation, with no animated tear and no audio.
The trend came from CapCut tutorials, and many of those posts are screen recordings with their own voiceover, which can give the impression of a video effect. There is no single trend song either: the posts we watched use voiceover or unrelated tracks.
If you post the picture as a slideshow or a short clip, you choose the music yourself. The reference clips on the page are results made with other tools, shown muted, and they were not made by videosi.
Do I have to label a torn paper picture as AI?
If the platform asks, yes, and the rules depend on whether the image looks realistic. A stylised torn paper portrait may count as non-realistic, but it is safest to label it when it shows a real person.
YouTube, for example, requires creators to disclose AI-generated or altered content that seems realistic, and says that non-realistic or minor edits do not require disclosure. You answer a question about AI use during upload in YouTube Studio. The details are in YouTube's disclosure policy.
Other platforms have their own wording and change it often, so read the current rule where you post. Use the picture only with consent from the person shown, who must be an adult, and do not use it to mislead anyone about a real person.
What does it cost, and what if a result fails?
One torn paper effect image costs 1 credit. Credit packs start at $6.99 for 10 credits and do not expire, and plans start at $29.90 a month.
There is no free generation on this page, and purchases are final. A failed or blocked generation returns its credits automatically, so you are not charged for a result that never arrives. It usually finishes in about a minute.
Because the tear is different every time, some people make two or three versions and pick the best one. At 1 credit each, that is a reasonable way to avoid a weak edge or a wrong hand. Start on the torn paper effect AI page with one clear photo.
