Application of the Inverse Packet Discrete Wavelet Transform for Deterministic Image Synthesis in Computer Graphics and Generative Design

Relevance: modern stochastic generative models (GANs, diffusion models) provide high-quality image synthesis; however, their probabilistic nature precludes precise spatial control over composition and detail hierarchy, limiting their application in tasks requiring result reproducibility and predictability. Consequently, there is a need to develop deterministic synthesis algorithms that combine controllability of the structural organization of generated data with potential for practical application in computer graphics and generative design.

Purpose: demonstration of the potential of the deterministic image synthesis algorithm based on the inverse packet discrete wavelet transform (IPDWT) for solving practical tasks in computer graphics and generative design that require reproducibility, interpretability, and controllability of the generation process.

Methods: analytical review, mathematical modeling, modified IPDWT procedure, hierarchical synthesis from an 8×8 pixel binary mask, experimental implementation in gaming environments.

Results: five variants of practical application were developed and analyzed: a structural control tool for generative AI ("wavelet brush"), procedural landscape generation with corrective masks, creation of seamless textures, surface detailing of low-polygon 3D models, and wavetable generation for sound synthesis.

Novelty. For the first time, the applicability of the deterministic image synthesis algorithm as a structural control tool for computer graphics and generative design tasks is investigated, demonstrated through five practical scenarios: generation of conditioning maps for generative neural networks, procedural terrain generation, creation of seamless textures, texturing of low-poly models, and wavetable generation.

Practical Significance. The obtained results allow forming controlled image structures with guaranteed reproducibility, allow creating seamless textures without additional boundary post-processing due to internal periodicity, and allow automating the creation of wavetables with controlled spectral evolution of timbre. The proposed approach opens new opportunities in computer graphics, the gaming industry, and digital design, where interpretability and controllability of the generation process are crucial.

1. Yu Y., Zhan F., Lu S., Pan J., Ma F., Xie X., Miao C. WaveFill: A Wavelet-based Generation Network for Image Inpainting. Proceedings of the International Conference on Computer Vision, ICCV, 10–17 October 2021, Montreal, Canada. IEEE; 2021. DOI:10.1109/ICCV48922.2021.01385

2. Mattar W., Levy I., Sharon N., Dekel S. Wavelets Are All You Need for Autoregressive Image Generation. arXiv:2406.19997. 2024. DOI:10.48550/arXiv.2406.19997

3. Zhang L., Rao A., Agrawala M. Adding Conditional Control to Text-to-Image Diffusion Models. Proceedings of the International Conference on Computer Vision, ICCV, 01–06 October 2023, Paris, France. IEEE; 2023. DOI:10.1109/ICCV51070.2023.00355

4. He J., Hirn M. Texture synthesis via projection onto multiscale, multilayer statistics. arXiv:2105.10825. 2021. DOI:10.48550/arXiv.2105.10825

5. Brochard A., Zhang S., Mallat S. Generalized Rectifier Wavelet Covariance Models For Texture Synthesis. arXiv:2203.07902. 2022. DOI:10.48550/arXiv.2203.07902

6. Buchatsky A.N., Prokhorov K.Yu., Rogozinski G.G. Procedural Generation of Height Maps and Landscapes: Fractal Methods and Their Application. Vestnik of Volga Tech. Ser.: Radio Engineering and Infocommunication Systems. 2025;1(65): 18–33. (in Russ.) DOI:10.25686/2306-2819.2025.1.18. EDN:FADQBC

7. Prokhorov K.Yu., Buchatsky A.N., Rogozinsky G.G. Structural Transformations in Inverse Packet Wavelet Transform: Spatial Shift Model and Group Classification of Reconstructions. Radioengineering. 2026. (Accepted for Publication)

8. Bau D., Strobelt H., Peebles W., Wulff J., Zhou B., Zhu J.Y., et al. Semantic photo manipulation with a generative image prior. ACM Transactions on Graphics. 2019;38(4):59. DOI:10.1145/3306346.3323023

9. Liu N., Li S., Du Y., Torralba A., Tenenbaum J.B. Compositional Visual Generation with Composable Diffusion Models. Proceedings of the 17th European Conference on Computer Vision, ECCV, 23–27 October 2022, Tel Aviv, Israel. Lecture Notes in Computer Science, vol.13677. Cham: Springer; 2022. p.423–439. DOI:10.1007/978-3-031-19790-1_26

10. Perlin K. An Image Synthesizer. Proceedings of the 12th Annual Conference on Computer Graphics and Interactive Techniques, SIGGRAPH '85, 22–26 July 1985, San Francisco, USA. New York: Association for Computing Machinery; 1985. p.287–296. DOI:10.1145/325334.325247

11. Worley S. A Cellular Texture Basis Function. Proceedings of the 23rd Annual Conference on Computer Graphics and Interactive Techniques, SIGGRAPH '96, 04–09 August 1996, New Orleans, USA. New York: Association for Computing Machinery; 1996. p.291–294. DOI:10.1145/237170.237267

12. Stam J. Aperiodic Texture Mapping. 2003. URL: https://www.researchgate.net/profile/Jos-Stam/publication/2927727_Aperiodic_Texture_Mapping/links/0f317536cf9c4342dc000000/Aperiodic-Texture-Mapping.pdf [Accessed 05.08.2026]

13. Cohen M., Shade J., Hiller S., Deussen O. Wang Tiles for image and texture generation. ACM Transactions on Graphics. 2003;22(3):287–294. DOI:10.1145/1201775.882265

14. Dong W., Zhou N., Paul J.-C. Optimized Tile-Based Texture Synthesis. Proceedings of Graphics Interface, 28–30 May 2007, Montreal Canada. New York: Association for Computing Machinery; 2007. p.249–256. DOI:10.1145/1268517.1268558

15. Fruehstueck A., Alhashim I., Wonka P. TileGAN: Synthesis of Large-Scale Non-Homogeneous Textures. ACM Transactions on Graphics. 2019;38(4):58. DOI:10.1145/3306346.3322993

16. Rodríguez-Pardo C., Garces E. SeamlessGAN: Self-Supervised Synthesis of Tileable Texture Maps. IEEE Transactions on Visualization and Computer Graphics. 2022;29(6):2914–2925. DOI:10.1109/TVCG.2022.3143615

17. Rogozinski G.G., Prokhorov K.Yu., Gorbunova I.B. A Method for Wave Table Generation in Sound Synthesis Tasks Based on Two-Dimensional Wavelet Transform. In: Proceedings of the St. Petersburg International Conference and St. Petersburg Interregional Conference on Regional Informatics and Information Security. St. Petersburg: St. Petersburg Society of Informatics, Computer Engineering, Communication Systems and Control Publ.; 2024. p.483–486. (in Russ.)

18. Cook R.L., DeRose T. Wavelet Noise. ACM Transactions on Graphics. 2005;24(3):803–811. DOI:10.1145/1186822.1073264