🎬 Video Inpaint!
Video Inpaint is a semantic segmentation–powered video editing tool that allows you to automatically generate masks from text and regenerate selected regions with custom prompts.
Simply enter a segment prompt, and the system will detect the corresponding object or area in the video, generate a dynamic mask, and inpaint it with your new creative instructions.
No manual masking. No frame-by-frame drawing. Just prompt → mask → generate.
✨ Key Features
🧠 Automatic Mask Generation from Text
Input a segment keyword such as:
person
sky
car
background
hair
clothes
The model analyzes video frames and automatically extracts the matching region as a mask.
🎨 Prompt-Based Regeneration
Once the mask is generated, you can provide a new prompt to replace or modify that region.
Perfect for:
Background replacement
Object removal or substitution
Outfit modification
Scene restyling
Adding new visual elements
You define the idea — the model generates the result.
⚙ Parameter Explanation
🔁 Select Every N Frame
This parameter controls the frame sampling interval — how often the model performs segmentation and inpainting.
1 → Process every frame (highest accuracy, best temporal consistency, highest computation cost)
2 → Process every other frame
3 → Process every third frame
5–8 → Suitable for slow or static scenes, reduces computation load
📌 How to Choose?
Scenario Recommended Value
Fast character movement1–2
Slow camera motion2–4
Mostly static background4–8
Quick testing preview4+
Rule of thumb: Lower value → better temporal stability Higher value → faster processing, but may cause slight flickering in fast-moving areas
If unsure, start with 2 as a balanced default.
🎥 Ideal Use Cases
AI video background replacement
Removing unwanted objects
Clothing or appearance modification
Creative video transformation
Experimental AI visual effects
🌟 Why Use Video Inpaint?
Fully automatic mask generation
Prompt-driven region editing
Flexible performance control
Designed for creative and experimental workflows