Shot boundary detection
Shot boundary detection is the automatic process of finding the exact points in a video where one camera shot ends and the next begins. It detects hard cuts, where the picture changes instantly, and gradual transitions such as fades, dissolves and wipes. Each detected boundary is reported as a timestamp or frame number, giving a precise map of how the video was edited.
Why it matters
Shot boundaries are the building blocks of most automated editing. Thumbnails, highlights, scene grouping and training datasets all start from clean cuts. Misplaced boundaries produce clips that start mid-motion or include a flash of the previous shot, which looks careless to viewers.
How it works
- The system compares each frame with the one before it, using color, edges or learned features.
- A sudden large difference between two frames signals a hard cut.
- A slow, steady change over several frames signals a fade, dissolve or wipe.
- Each boundary is reported as a frame number or timestamp.
What is the difference between a hard cut and a gradual transition?
A hard cut switches from one shot to the next between two frames, so the change is instant. A gradual transition, such as a fade or dissolve, blends the two shots over several frames. Hard cuts are easy to detect because the jump is large. Gradual transitions are harder because each frame differs only slightly from the last.
Is shot boundary detection the same as scene detection?
No. Shot boundary detection finds every camera cut. Scene detection groups those shots into larger story units, such as one conversation or one location. A two minute scene might contain twenty shots. Most scene detection systems run shot boundary detection first and then decide which shots belong together.
Where it's used
News publishers and creator platforms split footage at each camera cut to build highlight reels and thumbnails quickly, and generative AI video companies use the single-shot clips to train their models.
