Table 5 Comparison of computational efficiency and visual quality across different NST models.

From: Applying deep learning for style transfer in digital art: enhancing creative expression through neural networks

Model type

Processing time (s)

Memory usage (MB)

Content score

Style score

Efficiency

Full iterative

15.2

1500

5.0

5.0

Low

Feed-forward (AdaIN)

2.3

800

4.5

4.5

High

Lightweight CNN

1.5

500

4.0

4.0

Very high