
Haunted Portrait
A slow head turn, drifting fog, and a gentle push-in. The most reliable recipe for a portrait with a clear face.
Animate a costume portrait into a short, atmospheric video. Keep the face and outfit consistent while the camera moves, the fog drifts, and the light shifts.
Upload the portrait you want to animate. Short, specific motion instructions produce more stable results than dramatic action.

Upload a first frame or paste one public image URL. Private and share-page links will fail.
Upload an ending frame or paste one public image URL if you want the motion to land on a target shot.
Cost 1 credits
Generation Controls
These are the shared controls across all three workflow lanes.
Uses the faster route when available.
Include synchronized audio when the provider supports it.
Nothing generated yet
The safest clips use small movements: a head turn, blinking, flickering light, and drifting fog. Start there, then experiment.

A slow head turn, drifting fog, and a gentle push-in. The most reliable recipe for a portrait with a clear face.

Low fog, falling leaves, flickering lanterns, and a slow lateral camera move.

Flickering candlelight, floating dust, one blink, and a nearly static camera.

A slow forward drift, humming fluorescent lights, and a faint haze in a liminal corridor.

Carved pumpkins pulse warmer, warm light moves across the face, and the subject smiles subtly.
Image-to-video models are best at continuing what is already in the frame. When you ask for a costume change, a new character, or a fast action sequence, the model has to invent too much and the face starts to drift. A five-second clip with one clear movement looks expensive because it is believable.
The recipe that works most often is: hold the composition, move one thing, move the camera slightly, and keep the lighting consistent. A head turn plus drifting fog is enough for a TikTok or Reels loop. You can always generate a second clip with a different movement and cut them together.
Use a portrait with a clear face, a visible costume silhouette, and some background depth. A flat cut-out on a white background gives the model fewer cues and often produces a floating, pasted-on look. If your still image is too plain, generate a new one with an environmental background first.
Keep the resolution reasonable. A very large image is slower to process and does not improve the motion. A 1024- or 1536-pixel-wide portrait is usually enough for a vertical crop.
The best use is not a standalone film. Use the clip as a moving hook: the first two seconds of a TikTok, a background loop for a Halloween party invitation, or a subtle animated header on a product page. Because the clip is short, it stays under most platform limits and is easy to re-cut into vertical, square, or landscape formats.
If you need a longer video, generate several five-second clips from the same portrait with different motion recipes, then join them with a simple crossfade. Keep the same costume and lighting across clips so the sequence still feels continuous.
One still image in, one short clip out. No timeline editor required.
Use the image you already generated. Make sure the face is sharp and the costume is clearly visible.
Start with a head turn and drifting fog. Add camera movement only after the first clip looks stable.
Watch for face drift, extra fingers, and sudden cuts. Regenerate with simpler motion if anything warps.
These prompts describe one continuous shot. Keep the duration short and the motion specific.
Use the provided image as the exact visual reference and starting composition. Create a continuous five-second cinematic Halloween portrait animation. Keep the subject’s recognizable face, costume, hairstyle, and body proportions consistent. The subject slowly turns their head toward the camera with a subtle mysterious expression. A gentle breeze moves loose hair strands and the edges of the clothing. Soft fog drifts through the moonlit garden. The camera performs a slow, smooth push-in. Maintain realistic motion, stable anatomy, and consistent lighting. Do not change the person’s identity, replace the costume, add characters, or introduce scene cuts. No text or watermarks.
Animate the provided image into a five-second atmospheric Halloween shot. Preserve the subject and composition. Low fog rolls across the ground, a few autumn leaves drift past the lens, and the background lanterns flicker softly. The camera makes a very slow lateral move. Keep motion subtle and realistic, with stable anatomy and consistent lighting. Do not add people, change the face, or cut the scene. No text or watermarks.
Create a five-second cinematic animation from the provided image. Keep the subject and costume consistent. Candle flames flicker, warm light shifts across the face, and faint dust motes float in the air. The subject blinks once and breathes naturally. The camera holds steady with a barely perceptible push-in. Realistic motion, stable anatomy, consistent identity. No scene cuts, no extra characters, no text or watermarks.
Animate the provided image into a five-second uneasy Backrooms-style shot. Preserve the subject and architecture. The camera drifts forward slowly, fluorescent lights hum and flicker once, and a thin haze moves through the corridor. The subject turns slightly toward the camera. Keep motion minimal and realistic with stable anatomy. Do not add characters or sudden cuts. No text or watermarks.
Animate the provided image into a five-second warm Halloween scene. Preserve the subject and composition. The carved pumpkins glow brighter and pulse gently, warm light moves across the subject’s face, and light fog drifts behind them. The subject smiles subtly and turns their head a few degrees. Slow, smooth camera push-in. Realistic motion and consistent identity. No extra people, no scene cuts, no text or watermarks.
Practical answers about length, motion quality, and formats.
The hosted video workflow is designed for short clips. Five seconds is the sweet spot for image-to-video because it gives the model enough time for one clear movement without inventing a new scene.
Audio support depends on the selected video model. The prompt recipes on this page focus on silent clips that you can pair with trending audio in TikTok, Reels, or Shorts.
Fast movement, a large camera move, or a request to change the costume mid-clip forces the model to invent new angles. Reduce the motion to a small head turn, keep the camera nearly still, and use a sharper source image.
Yes. Use a photo with one clear subject if possible. For couples, a gentle head turn from both people is harder to keep stable; start with a nearly static camera and a small environmental movement such as fog or candlelight.
Choose vertical 9:16 for TikTok, Reels, and Shorts. Use 16:9 for a website header or YouTube. Generate the still image in the same aspect ratio as the final video to avoid awkward cropping.
Generate the costume portrait first, then animate it. Or edit a group photo and turn that into a clip.
Upload a still, choose a subtle motion recipe, and get a clip you can post tonight.