Generative artificial intelligence (GenAI) enables rapid packaging concept exploration, but consistent design across product variants remains difficult to specify and verify. This article develops a constraint-based workflow that separates invariant brand elements, variant-specific imagery and colors, and production constraints. The workflow was illustrated with a three-product tea series (Jasmine, Citrus, Berry) and examined in a six-image exploratory generation exercise using Adobe Firefly. Each product was generated once using a narrative prompt and once using a structured-field prompt carrying the same design information. Paired requests specified identical seeds, a 3:4 aspect ratio, and 1MP resolution. A single-observer visual inspection found the intended ingredient motifs, green ornamental frames, empty brand zones, and distinct accent bands were broadly represented in all six images. Unrequested pseudotext appeared in all three structured-prompt images but in none of the three narrative-prompt images. Proportions and motif scale also varied across the family. This small descriptive study does not establish the general superiority of either prompt format, but it demonstrates a traceable approach to packaging design constraints and identifies reasons that artificial intelligence (AI) concepts require human typesetting and technical finishing before production.
References
[1] Cao, P., Zhou, F., Song, Q., Yang, L. (2025) Controllable generation with text-to-image diffusion models: a survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(4), 4771-4791.
[2] Oppenlaender, J. (2024) A taxonomy of prompt modifiers for text-to-image generation. Behaviour & Information Technology, 43(15), 3763-3776.
[3] Li, H., Xue, T., Zhang, A., Luo, X., Kong, L., Huang, G. (2024) The application and impact of artificial intelligence technology in graphic design: a critical interpretive synthesis. Heliyon, 10(21), e40037.
[4] Gao, Z., Li, Z. (2026) A tea packaging design method based on artificial intelligence generated content. Journal of Engineering Design, 37(2), 339-371.
[5] Qi, J., Xu, W., Zhu, Q., Chu, X., Wang, Y. (2026) Collaborative control in diffusion models for precise image generation: a survey. Mathematics, 14(15), 2737.
[6] Yang, F., Yang, S., Butt, M. A., Van De Weijer, J. (2023) Dynamic prompt learning: addressing cross-attention leakage for text-based image editing. Advances in Neural Information Processing Systems, 36, 26291-26303.
[7] Gong, X. (2023) Enhancing brand consistency via integration of emotional graphic design and cloud service interface system. The Design Journal, 26(4), 673-684.
[8] Huang, H., Lee, K. (2026) Development of an integrated branding platform based on generative AI and a comparative usability study between designers and non-designers. Archives of Design Research, 39(1), 199-222.
[9] Chong, L., Lo, I. P., Rayan, J., Dow, S., Ahmed, F., Lykourentzou, I. (2025) Prompting for products: investigating design space exploration strategies for text-to-image generative models. Design Science, 11, e2.
[10] Chen, J., Huang, Y., Lv, T., Cui, L., Chen, Q., Wei, F. (2023) Textdiffuser: diffusion models as text painters. Advances in Neural Information Processing Systems, 36, 9353-9387.
Share and Cite
Yu, Q. (2026) Application of Generative Artificial Intelligence to Visual Consistency in Product Packaging Design. Scientific Research Bulletin, 3(5), 1-8. https://doi.org/10.71052/srb2024/FJHS9377