In this paper, we propose a feature-based multi-frequency component fusion as a network module for the attention mechanism, in response to the prevalence of current image super-resolution models for the most basic residual network module in the network. The performance of the feature-based multi-frequency fusion image super-resolution model is evaluated using peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) to verify its operability for image enhancement in graphic design. Aiming at the dynamic demand for logo design in the era of smart media, the evaluation system of dynamic graphic design is constructed. Combined with the evaluation scoring of the samples by the dynamic graphic design evaluation system, a new strategy for dynamic graphic design expression is proposed. In the evaluation system of dynamic graphic design, the order of attributes is perceptual attributes, aesthetic attributes, cognitive attributes and perceptual experience attributes. Therefore, the perceptual aspects of graphic presentation need to be prioritized in design creativity. Combined with the score of 75.68 in the sample design, it is proposed that the visual effects of graphics and information visualization should be emphasized in the design of new media motion graphics.