Research on intelligent image processing and video enhancement algorithm for film post-production
Abstract
With the increasing expectations of the audience for immersive experience, film post-production is facing many challenges, including the contradiction between labor cost and efficiency, and the conflict between artistic expression and technical adaptation. In order to solve these problems, this paper proposes a comprehensive algorithm framework, which aims to improve the technical performance and artistic expression ability of film post-production through key technical breakthroughs such as dynamic scene perception super-resolution network, spatio-temporal consistency enhancement module and Controllable generative adversarial networks (CGAN). Dynamic scene-aware super-resolution network adaptively allocates computing resources according to depth of field through attention mechanism, and optimizes detail enhancement in different regions. Spatio-temporal consistency enhancement module adopts block spatio-temporal attention mechanism and optical flow estimation, which solves the flicker artifact problem between consecutive frames in long shots and ensures the fluency of video. CGAN embeds artistic guidance parameters such as Look-up table (LUT) parameters and film grain strength into the generation process, so that the generated content is more in line with the director's artistic intention. The experimental results show that the framework is superior to the existing advanced methods in objective indicators, such as higher PSNR and SSIM values, and the calculation amount is significantly reduced. In the subjective evaluation, experts highly recognized the artistic controllability and motion consistency of the algorithm. In addition, the algorithm shows excellent performance in the film 4K restoration project, significantly reducing the need for manual correction, and showing a higher degree of detail reduction in complex scenes. The research in this paper provides an efficient and artistic intelligent image processing and video enhancement solution for film post-production.
Fields of Study & Subjects
Advanced Technologies in Various Fields
Generative Adversarial Networks and Image Synthesis
Digital Media and Visual Art
Publication details
- Journal / source
- IET conference proceedings.
- Volume
- 2025
- Issue
- 34
- Pages
- 282–286
- Publication date
- 1 January 2026
- Language
- EN