# How to Make a Witchcore AI Video From One Photo > How to make a witchcore AI video from one photo: pick a clear portrait, follow the phone-to-cauldron story, and inspect your clip before posting. - URL: https://trendai.video/blog/witchcore-ai-video-photo-guide - Published: 2026-10-15 - Updated: 2026-10-17 - Category: TikTok Trends - Tags: AI video template, Halloween video, witchcore - Author: Trendai.video A witch in a black hat ends a phone call, stirs a cauldron and reaches toward a floating black cat. To make a Witchcore AI video, upload one clear portrait to the [Witchcore template](https://trendai.video/trends/witchcore-video), generate the scene and check the face through every cut. This is a 1960s-style little story, not a generic costume filter. ## The phone call sets up the joke The [Witchcore template page](https://trendai.video/trends/witchcore-video) shows a 20-second scene in 4:3. Its built-in story starts with a rotary phone in a red parlor, moves to a cauldron and ends with the black cat in the air. The spoken phone line in the sample turns down plans for the day. Then the cat makes the witch's actual plans fairly obvious. That order matters. Don't skip it. If you cut away before the cauldron, the joke reads as a vintage costume portrait with no punch line. If you start with the cat, you spoil the reveal. Keep the phone beat when making a short teaser for your feed, even if you later trim the scene to fit a faster edit. The page labels its displayed clip as a sample output. It's a reference for the scene, not evidence that your portrait will preserve every eyelash or sound exactly like the preview. Listen to your own download before you write captions quoting any dialogue. Check it first. A generated voice, music bed or cat sound is something to check, not to promise. The [Haunted Elevator photo guide](https://trendai.video/blog/haunted-elevator-halloween-ai-photo) is another Halloween portrait idea, but it ends on a still instead of a cauldron scene. ## Pick a portrait that survives the hat The live form asks for **one clear photo of one person with the face visible**. There isn't a second slot for a cat, a costume or a friend. Your picture supplies the person; the setting and animal come from the template. Pick a head-and-shoulders shot with both eyes visible and enough light across the face. A hat brim and flickering room already add shadows, so a selfie taken in near darkness makes the most recognizable part of the video harder to read. Don't use a group photo and hope the model chooses the right friend. Skip sunglasses, heavy beauty filters and hair covering half the face. Both eyes matter. A photo where you're turned a little toward the lens will usually give the edit more to work with than a profile in which one eye disappears. The page says the scene follows the photographed person's hair and face, but it's still smart to inspect each close shot in the actual export. You don't need to buy a witch hat for the input shot. The costume is part of the fictional scene. No prop needed. Keep your original portrait straightforward rather than stacking makeup filters on top of what the generator will add. The cover image shows a blond person at a steaming pot and a black cat between red curtains; it is the template poster, not a personalized result. ## Make it in the template 1. Open the [Witchcore creation page](https://trendai.video/trends/witchcore-video) and upload a photo you own or have permission to use. The form shows one required photo upload and the scene's fixed 20-second, 4:3 format. 2. Check the credit amount on the live button before you generate. The page displays a price, but use the amount shown to you at creation time rather than a number copied from an older guide. 3. Generate, preview the result and download it if the face, cat and phone beats hold together. Watch the full clip with sound, then once muted: silent autoplay should still make the setup and payoff clear. The form calls this an image-to-video template. It isn't a place to upload your own audio or rewrite the phone exchange. If you want a different line, treat that as a later edit with audio you have the right to use; don't assume the template will speak custom words. ## Frame the vintage scene for a feed The 4:3 canvas is wider than a vertical story slot. The parlor, cauldron and hovering cat use that width. Width matters here. A hard center crop could remove the cat precisely as the person reaches for it. If you're posting to a tall surface, keep the entire frame inside a vertical canvas with a background or a deliberate border before trying a crop that cuts off the punch line. A caption can play off the visual without pretending the incident happened: “I said I had plans. The cat was the plan.” Put any on-screen text away from the hand-to-cat interaction. Watch the export at phone size, not just in a desktop preview; the bright red curtains can pull the eye away from a smaller face. There's a clean contrast with the [Witch Flying on Broom template](https://trendai.video/trends/witch-flying-on-broom-video). That one keeps the outfit from your photo and stages a roadside takeoff in a vertical clip. Witchcore changes the setting into a vintage interior and tells a longer phone-to-cat story. Choose based on the beat you want, not just the word “witch.” ## Check the result before it goes public Pause where the hand meets the phone, where the stirring stick enters the pot and when the finger touches the cat. Those are the spots where a generated object can slide or a hand can lose its shape. If a frame looks wrong, use a different clear portrait rather than treating the sample poster as a guarantee. Get consent from anyone whose face you upload, and present the scene as a fictional AI edit. Don't assign a real person's identity to the cat or claim the phone call is genuine. For platform disclosures, follow the synthetic-media controls offered at upload. A good last pass is simple: play the downloaded file on your phone with captions visible, then make sure the cat still appears inside the frame.