Article: Semantic-guided multi-scale diffusion network for immersive tourism content generation Journal: International Journal of Information Quality (IJIQ) 2026 Vol.4 No.5 pp.1 - 20 Abstract: This paper proposes an immersive tourism content generation model based on a semantic-guided multi-scale diffusion network (SGMSD-Net) to address semantic consistency, panoramic structure preservation, and high-resolution detail restoration in digital cultural tourism engineering. Built on the diffusion generation framework, SGMSD-Net integrates semantic modulation, multi-scale diffusion, and multi-objective optimation. Semantic embeddings and feature-wise linear modulation (FiLM) are used to dynamically guide the generation process, while a hierarchical diffusion module progressively restores structural and textural details from coarse to fine scales. A joint loss combining semantic consistency and perceptual constraints is further introduced to improve realism and semantic alignment. Experiments on SUN360 and Matterport3D show that SGMSD-Net outperforms existing methods in FID, LPIPS, and CLIP score. Subjective evaluations also confirm its advantages in clarity, spatial consistency, and visual appeal, demonstrating its potential for intelligent cultural tourism, virtual tours, and digital heritage reconstruction. Inderscience Publishers - linking academia, business and industry through research

Title: Semantic-guided multi-scale diffusion network for immersive tourism content generation

Authors: Lun Wei; Afia Zafar; Muhammad Attique Khan

Addresses: College of Art and Design, Ma'anshan University, Ma'anshan, 243100, Anhui, China ' National University of Computer and Emerging Sciences (Fast), Islamabad, Pakistan ' Department of Computer Science, College of Computer Engineering and Science, Prince Mohammad Bin Fahd University, Eastern, Saudi Arabia

Abstract: This paper proposes an immersive tourism content generation model based on a semantic-guided multi-scale diffusion network (SGMSD-Net) to address semantic consistency, panoramic structure preservation, and high-resolution detail restoration in digital cultural tourism engineering. Built on the diffusion generation framework, SGMSD-Net integrates semantic modulation, multi-scale diffusion, and multi-objective optimation. Semantic embeddings and feature-wise linear modulation (FiLM) are used to dynamically guide the generation process, while a hierarchical diffusion module progressively restores structural and textural details from coarse to fine scales. A joint loss combining semantic consistency and perceptual constraints is further introduced to improve realism and semantic alignment. Experiments on SUN360 and Matterport3D show that SGMSD-Net outperforms existing methods in FID, LPIPS, and CLIP score. Subjective evaluations also confirm its advantages in clarity, spatial consistency, and visual appeal, demonstrating its potential for intelligent cultural tourism, virtual tours, and digital heritage reconstruction.

Keywords: semantic-guided diffusion model; multi-scale image generation; immersive tourism; panoramic scene synthesis; digital cultural heritage reconstruction.

DOI: 10.1504/IJIQ.2026.157237

International Journal of Information Quality, 2026 Vol.4 No.5, pp.1 - 20

Received: 15 Jun 2026
Accepted: 28 Jul 2026
Published online: 23 Oct 2026 *

Title: Semantic-guided multi-scale diffusion network for immersive tourism content generation

Keep up-to-date