Why does the video claim a 33.3% storage reduction in the face measurement example?

Conditions

• Each original sample stores 3 floating-point coordinates.
• Each compressed sample stores 2 floating-point coordinates.

Reasoning, step by step

1. Identify the original representation: 3 floats per face.
2. Identify the compressed representation: 2 floats per face.
3. Calculate the difference in the number of stored values: $3 - 2 = 1$.
4. Divide the difference by the original number of values: $1 / 3$.
5. Convert the fraction to a percentage: 33.3%.

Example

The video displays a bottom banner reading 'Storage Reduction: -33.3% (3 floats -> 2 floats per face)' while showing the side-by-side coordinate compression.

Common misconceptions

• Assuming the storage reduction applies to the entire dataset's file size without considering metadata or overhead.
• Believing that reducing dimensions always reduces storage by 50% regardless of the original dimensionality.

Watch the explanation

Explore next

• How are the two compressed coordinates $z_1$ and $z_2$ obtained from the original 3D vector?
• What does the reconstruction formula $\hat{\mathbf{x}} = \mathbf{W}\mathbf{z}$ represent?