AI 视频生成器

CCTV Tracking Overload

模型

提示词

CCTV TRACKING OVERLOAD — VIRAL COMPUTER-VISION EFFECT

Use the uploaded image as the exact starting scene. Preserve the subject's identity, clothing, environment, composition and original camera perspective.

The underlying footage remains photoreal and recognizable throughout.

Human movement is completely candid and natural, like ordinary surveillance or observational footage. No staged acting, fashion posing or exaggerated reaction.

NORMAL OPENING:
Begin with completely clean footage.

The subject casually moves within the existing scene.

After a moment, a small computer-vision tracking box snaps precisely around the main subject.

A tiny tracking point locks onto their face.

The box follows them perfectly as they move.

Then the system begins analyzing more information.

Additional small tracking markers appear on the eyes, nose, shoulders, elbows, hands, hips, knees and feet.

Thin skeletal pose-estimation lines connect the major joints.

The tracking remains accurately attached to the moving body.

ESCALATION:
The system begins detecting other objects in the environment.

Bounding boxes rapidly snap around:

people,
vehicles,
bags,
chairs,
doors,
windows,
plants,
street objects,
furniture,
and anything else visible in the supplied scene.

Small technical labels, coordinates, confidence values and tracking IDs appear beside detected objects.

The system becomes increasingly aggressive.

Every moving object receives motion vectors.

Faces receive dense landmark meshes.

Hands receive finger tracking.

The environment receives perspective lines and depth measurements.

A subtle grid maps the ground plane.

The camera's computer vision appears to be understanding the entire physical scene in real time.

TRACKING OVERLOAD:
The analysis rapidly becomes absurd.

The system starts tracking things that do not need tracking.

Individual shoes.

Individual pieces of clothing.

Hair movement.

Shadows.

Reflections.

Leaves.

Tiny background objects.

Edges of buildings.

Every visible element receives its own bounding box, points, vectors or measurement lines.

The underlying real-world scene continues moving naturally while hundreds of perfectly synchronized analytical graphics accumulate over it.

The screen becomes an increasingly dense machine-vision visualization without completely hiding the original footage.

FINAL BEAT:
At maximum overload, nearly EVERYTHING visible is being simultaneously tracked.

The frame is filled with rapidly updating boxes, skeletons, landmarks, vectors, coordinates and depth indicators, all accurately locked to the real scene.

Then the tracking system suddenly crashes.

Every graphic disappears simultaneously.

The clean original footage remains.

The person casually continues their normal behavior as though nothing happened.

CAMERA:
Keep the source-image camera behavior natural and observational. The tracking graphics must remain accurately attached to their targets despite camera or subject movement.

AESTHETIC:
Real computer vision.
Object detection.
Pose estimation.
Facial landmarks.
Motion tracking.
Depth analysis.
Technical diagnostic overlay.

NOT futuristic holograms.
NOT sci-fi HUD.
NOT cyberpunk interface.

The effect should feel like a real machine-vision system becoming impossibly obsessive.

PROGRESSION:

CLEAN REAL FOOTAGE
→ ONE SUBJECT TRACKED
→ BODY ANALYZED
→ OBJECTS DETECTED
→ ENTIRE WORLD TRACKED
→ TOTAL ANALYTICAL OVERLOAD
→ INSTANTLY CLEAN AGAIN.

3462 / 50000

参考图片（可选）

画面比例

时长

分辨率

生成音频2 倍积分Basic+

所需积分50

视频预览

加载中...