{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,16]],"date-time":"2026-06-16T22:52:00Z","timestamp":1781650320904,"version":"3.54.5"},"reference-count":46,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,4,13]],"date-time":"2026-04-13T00:00:00Z","timestamp":1776038400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,5,30]],"date-time":"2026-05-30T00:00:00Z","timestamp":1780099200000},"content-version":"vor","delay-in-days":47,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Discov Artif Intell"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>\n                    Text-to-image diffusion models can generate high-quality images from individual prompts, yet it remains unclear how reliably they can fuse\n                    <jats:italic>two<\/jats:italic>\n                    distinct concepts into a single coherent image in a strictly zero-shot setting. Inspired by human combinational creativity, we operationalize visual concept blending here as conditioning a diffusion model on a pair of text prompts and ask whether the resulting image preserves recognizable traits of both concepts while remaining visually coherent. We study four inference-time blending strategies that intervene at different stages of the diffusion process, including embedding-space interpolation, mid-denoising prompt switching, timestep-wise prompt alternation, and layer-wise conditioning within the denoising network. We evaluate these strategies across diverse blend categories, spanning object\u2013object blends, compound concepts, style\u2013content combinations, and architectural landmarks, and we complement the analysis with a user study involving 100 participants. Results indicate that no single strategy dominates across all categories: timestep-scheduling approaches are often preferred for producing recognisable and well-integrated hybrids, while embedding- and layer-based interventions exhibit characteristic strengths and failure modes in specific settings. Finally, we analyse practical control factors such as prompt ordering, blend ratio, and random seed, showing how they can substantially alter outcomes and providing concrete levers for steering blends more predictably in creative applications.\n                  <\/jats:p>","DOI":"10.1007\/s44163-026-01134-1","type":"journal-article","created":{"date-parts":[[2026,4,13]],"date-time":"2026-04-13T05:41:28Z","timestamp":1776058888000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Blending concepts with text-to-image diffusion models"],"prefix":"10.1007","volume":"6","author":[{"given":"Lorenzo","family":"Olearo","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Giorgio","family":"Longari","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alessandro","family":"Raganato","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rafael","family":"Pe\u00f1aloza","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Simone","family":"Melzi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,4,13]]},"reference":[{"key":"1134_CR1","volume-title":"The way we think: conceptual blending and the mind\u2019s hidden complexities","author":"G Fauconnier","year":"2003","unstructured":"Fauconnier G, Turner M. The way we think: conceptual blending and the mind\u2019s hidden complexities. New York: Basic Books; 2003a."},{"key":"1134_CR2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-65602-1","author":"R Confalonieri","year":"2018","unstructured":"Confalonieri R, Pease A, Schorlemmer M, Besold T, Kutz O, Maclean E, et al. Concept invention foundations, implementation, social aspects and applications: foundations, implementation. Soc Aspects Appl. 2018. https:\/\/doi.org\/10.1007\/978-3-319-65602-1.","journal-title":"Soc Aspects Appl"},{"key":"1134_CR3","volume-title":"Conceptual Spaces: The Geometry of Thought","author":"P Gardenfors","year":"2004","unstructured":"Gardenfors P. Conceptual Spaces: The Geometry of Thought. A Bradford book. Cambridge: MIT Press; 2004."},{"issue":"1\u20133","key":"1134_CR4","doi-asserted-by":"publisher","first-page":"69","DOI":"10.1007\/S10472-016-9524-8","volume":"82","author":"R Confalonieri","year":"2018","unstructured":"Confalonieri R, Eppe M, Schorlemmer M, Kutz O, Pe\u00f1aloza R, Plaza E. Upward refinement operators for conceptual blending in the description logic EL++. Ann Math Artif Intell. 2018;82(1\u20133):69\u201399. https:\/\/doi.org\/10.1007\/S10472-016-9524-8.","journal-title":"Ann Math Artif Intell"},{"key":"1134_CR5","unstructured":"Porello D, Kutz O, Righetti G, Troquard N, Galliani P, Masolo C. A toothful of concepts: towards a theory of weighted concept combination. In: Simkus M, Weddell GE, editors. Proceedings of the 32nd international workshop on description logics (2019)"},{"key":"1134_CR6","unstructured":"Sohl-Dickstein J, Weiss E, Maheswaranathan N, Ganguli S. Deep unsupervised learning using nonequilibrium thermodynamics. In: International conference on machine learning. PMLR; 2015. p. 2256\u20132265."},{"key":"1134_CR7","first-page":"6840","volume":"33","author":"J Ho","year":"2020","unstructured":"Ho J, Jain A, Abbeel P. Denoising diffusion probabilistic models. Adv Neural Inf Process Syst. 2020;33:6840\u201351.","journal-title":"Adv Neural Inf Process Syst"},{"key":"1134_CR8","doi-asserted-by":"crossref","unstructured":"Rombach R, Blattmann A, Lorenz D, Esser P, Ommer B. High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE\/CVF conference on computer vision and pattern recognition. 2022. p. 10684\u201310695","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"1134_CR9","unstructured":"Melzi S, Pe\u00f1aloza R, Raganato A. Does stable diffusion dream of electric sheep? CEUR Workshop Proc. 2023;3511:1\u201311 (CEUR-WS)."},{"key":"1134_CR10","unstructured":"Olearo L, Longari G, Melzi S, Raganato, A, Pe\u00f1aloza R. How to blend concepts in diffusion models (2024). Preprint at arXiv:2407.14280"},{"key":"1134_CR11","doi-asserted-by":"crossref","unstructured":"Sun Z, Zhang Z, Zhang Y, Lu M, Lischinski D, Cohen-Or D, Huang H. Creative blends of visual concepts (2025)","DOI":"10.1145\/3706598.3713683"},{"key":"1134_CR12","doi-asserted-by":"publisher","first-page":"57","DOI":"10.14428\/rec.v19i19.48413","volume":"19","author":"G Fauconnier","year":"2003","unstructured":"Fauconnier G, Turner M. Conceptual blending, form and meaning. Recherches en communication. 2003b;19:57\u201386.","journal-title":"Recherches en communication"},{"issue":"2","key":"1134_CR13","doi-asserted-by":"publisher","first-page":"133","DOI":"10.1207\/s15516709cog2202_1","volume":"22","author":"G Fauconnier","year":"1998","unstructured":"Fauconnier G, Turner M. Conceptual integration networks. Cogn Sci. 1998;22(2):133\u201387. https:\/\/doi.org\/10.1207\/s15516709cog2202_1.","journal-title":"Cogn Sci"},{"key":"1134_CR14","doi-asserted-by":"publisher","first-page":"105","DOI":"10.1016\/j.artint.2017.11.005","volume":"256","author":"M Eppe","year":"2018","unstructured":"Eppe M, Maclean E, Confalonieri R, Kutz O, Schorlemmer M, Plaza E, et al. A computational framework for conceptual blending. Artif Intell. 2018;256:105\u201329. https:\/\/doi.org\/10.1016\/j.artint.2017.11.005.","journal-title":"Artif Intell"},{"key":"1134_CR15","doi-asserted-by":"publisher","unstructured":"Kutz O, Bateman J, Neuhaus F, Mossakowski T, Bhatt M. In: Besold TR, Schorlemmer M, Smaill A, editors. E Pluribus Unum. Paris: Atlantis Press; 2015. p. 167\u2013196. https:\/\/doi.org\/10.2991\/978-94-6239-085-0_9.","DOI":"10.2991\/978-94-6239-085-0_9"},{"key":"1134_CR16","doi-asserted-by":"publisher","unstructured":"Hedblom MM, Kutz O, Neuhaus F. Image schemas in computational conceptual blending. Cogn Syst Res. 2016;39:42\u201357.https:\/\/doi.org\/10.1016\/j.cogsys.2015.12.010. (From human to artificial cognition (and back): new perspectives of cognitively inspired AI systems).","DOI":"10.1016\/j.cogsys.2015.12.010"},{"key":"1134_CR17","unstructured":"Si M. Word embedding for analogy making. In: ICCC. 2022. p. 282\u2013286."},{"key":"1134_CR18","doi-asserted-by":"publisher","DOI":"10.1016\/j.aei.2025.103427","volume":"67","author":"B Wang","year":"2025","unstructured":"Wang B, et al. From analogy to innovation: a creative conceptual design approach leveraging large language models. Adv Eng Inform. 2025;67:103427.","journal-title":"Adv Eng Inform"},{"key":"1134_CR19","doi-asserted-by":"crossref","unstructured":"Wang S, Petridis S, Kwon T, Lisle L, Mo BH, Ma C, Watkins O, Hearst MA. PopBlends: strategies for conceptual blending with large language models. In: Proceedings of the 2023 CHI conference on human factors in computing systems. 2023. p. 1\u201319.","DOI":"10.1145\/3544548.3580948"},{"issue":"11","key":"1134_CR20","doi-asserted-by":"publisher","first-page":"1394","DOI":"10.1080\/09544828.2024.2356707","volume":"35","author":"L Chen","year":"2024","unstructured":"Chen L, et al. A foundation model enhanced approach for generative design in combinational creativity. J Eng Des. 2024;35(11):1394\u2013420.","journal-title":"J Eng Des"},{"key":"1134_CR21","unstructured":"Cho W, Zhang Y, Chen Y-Y, Inouye DI. Imagine for me: creative conceptual blending of real images and text via blended attention. 2025. Preprint at arXiv:2506.24085."},{"key":"1134_CR22","unstructured":"Ge S, Parikh D. Visual conceptual blending with large-scale language and vision models. 2021. Preprint at arxiv:2106.14127"},{"key":"1134_CR23","unstructured":"Leemhuis M, Kutz O. The boat, the house and the in-between: conceptual blending as betweenness relation. In: Hedblom MM, Kutz O, editors. Proceedings of the eighth image schema day. CEUR workshop proceedings, vol. 3888. CEUR-WS.org, Bozen-Bolzano. 2024. https:\/\/ceur-ws.org\/Vol-3888\/Paper_5.pdf"},{"key":"1134_CR24","unstructured":"Wang CJ, Golland P. Interpolating between images with diffusion models. 2023. Preprint at arXiv:2307.12560."},{"key":"1134_CR25","doi-asserted-by":"crossref","unstructured":"Li J, Zhang Z, Yang J. Tp2o: creative text pair-to-object generation using balance swap-sampling. In: European conference on computer vision. Springer; 2024. p. 92\u2013111.","DOI":"10.1007\/978-3-031-73220-1_6"},{"key":"1134_CR26","doi-asserted-by":"crossref","unstructured":"Kothandaraman D, Lin M, Manocha D. Financial models meets generative art: Black-scholes-inspired concept blending in text-to-image diffusion. In: Proceedings of the 33rd ACM international conference on multimedia. 2025. p. 12209\u201312217.","DOI":"10.1145\/3746027.3758146"},{"key":"1134_CR27","unstructured":"Park T, Zhu J-Y, Wang O, Lu J, Shechtman E, Efros AA, Zhang R. Swapping autoencoder for deep image manipulation. In: Advances in neural information processing systems. 2020"},{"key":"1134_CR28","doi-asserted-by":"crossref","unstructured":"Alaluf Y, Garibi D, Patashnik O, Averbuch-Elor H, Cohen-Or D. Cross-image attention for zero-shot appearance transfer. 2023.","DOI":"10.1145\/3641519.3657423"},{"key":"1134_CR29","unstructured":"Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser \u0141, Polosukhin I. Attention is all you need. Adv Neural Inf Process Syst. 2017;30"},{"key":"1134_CR30","first-page":"97766","volume":"37","author":"H Qiyuan","year":"2024","unstructured":"Qiyuan H, Wang J, Liu Z, Yao A. Aid: attention interpolation of text-to-image diffusion. Adv Neural Inf Process Syst. 2024;37:97766\u201399.","journal-title":"Adv Neural Inf Process Syst"},{"key":"1134_CR31","unstructured":"Xiong Z, Zhang Z, Chen Z, Chen S, Li X, Sun G, Yang J, Li J. Novel object synthesis via adaptive text-image harmony. 2024."},{"key":"1134_CR32","unstructured":"Zhou Y, Shen H, Wang H. FreeBlend: advancing concept blending with staged feedback-driven interpolation diffusion. 2025. Preprint at arXiv:2502.05606"},{"key":"1134_CR33","unstructured":"Ramesh A, Dhariwal P, Nichol A, Chu C, Chen M. Hierarchical text-conditional image generation with clip latents. arXiv preprint 1(2), 3 (2022). arXiv:2204.06125"},{"key":"1134_CR34","unstructured":"Podell D, English Z, Lacey K, Blattmann A, Dockhorn T, M\u00fcller J, Penna J, Rombach R. SDXL: improving latent diffusion models for high-resolution image synthesis. In: The twelfth international conference on learning representations. 2024. https:\/\/openreview.net\/forum?id=di52zR8xgf."},{"key":"1134_CR35","unstructured":"Kingma DP, Welling M. Auto-encoding variational bayes. 2022. arXiv:1312.6114"},{"key":"1134_CR36","doi-asserted-by":"crossref","unstructured":"Ronneberger O, Fischer P, Brox T. U-net: convolutional networks for biomedical image segmentation. In: Medical image computing and computer-assisted intervention\u2013MICCAI 2015: 18th international conference, Munich, Germany, October 5\u20139, 2015, Proceedings, Part III 18. Springer; 2015. p. 234\u2013241.","DOI":"10.1007\/978-3-319-24574-4_28"},{"key":"1134_CR37","unstructured":"Radford A, Kim JW, Hallacy C, Ramesh A, Goh G, Agarwal S, Sastry G, Askell A, Mishkin P, Clark J et al. Learning transferable visual models from natural language supervision. In: International conference on machine learning. PMLR; 2021. p. 8748\u20138763."},{"key":"1134_CR38","unstructured":"Zhao W, Bai L, Rao Y, Zhou J, Lu J. UniPC: a unified predictor-corrector framework for fast sampling of diffusion models. 2023. Preprint at arXiv:2302.04867"},{"key":"1134_CR39","unstructured":"Karthik S, Roth K, Mancini M, Akata Z. If at first you don\u2019t succeed, try, try again: faithful diffusion-based text-to-image generation by selection. 2023. Preprint at arXiv:2305.13308"},{"key":"1134_CR40","unstructured":"Ho J, Salimans T. Classifier-free diffusion guidance. In: NeurIPS 2021 workshop on deep generative models and downstream applications. 2021."},{"key":"1134_CR41","unstructured":"Martins P, Urbancic T, Pollak S, Lavrac N, Cardoso A. The good, the bad, and the aha! blends. In: Toivonen H, Colton S, Cook M, Ventura D, editors. Proceedings of the sixth international conference on computational creativity, ICCC 2015; 2015. pp. 166\u2013173. computationalcreativity.net, Park City, Utah."},{"key":"1134_CR42","unstructured":"Radford A, Kim JW, Hallacy C, Ramesh A, Goh G, Agarwal S, Sastry G, Askell A, Mishkin P, Clark J, et al. Learning transferable visual models from natural language supervision. In: Proceedings of international conference on machine learning. PMLR; 2021. p. 8748\u20138763."},{"key":"1134_CR43","unstructured":"Park, D., Kim, S., Moon, T., Kim, M., Lee, K., Cho, J.: Rare-to-frequent: unlocking compositional generation power of diffusion models on rare concepts with LLM guidance. In: International conference on learning representations. 2025."},{"key":"1134_CR44","doi-asserted-by":"crossref","unstructured":"Liu N, Li S, Du Y, Torralba A, Tenenbaum JB. Compositional visual generation with composable diffusion models. In: European conference on computer vision. Springer; 2022. p. 423\u2013439.","DOI":"10.1007\/978-3-031-19790-1_26"},{"key":"1134_CR45","unstructured":"Huang L, Chen D, Liu Y, Shen Y, Zhao D, Zhou J. Composer: creative and controllable image synthesis with composable conditions. In: Proceedings of the 40th international conference on machine learning. 2023. p. 13753\u201313773."},{"key":"1134_CR46","doi-asserted-by":"crossref","unstructured":"Ruiz N, Li Y, Jampani V, Pritch Y, Rubinstein M, Aberman K. Dreambooth: fine tuning text-to-image diffusion models for subject-driven generation. In: Proceedings of the IEEE\/CVF conference on computer vision and pattern recognition. 2023. p. 22500\u201322510.","DOI":"10.1109\/CVPR52729.2023.02155"}],"container-title":["Discover Artificial Intelligence"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s44163-026-01134-1","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s44163-026-01134-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s44163-026-01134-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,16]],"date-time":"2026-06-16T21:54:44Z","timestamp":1781646884000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s44163-026-01134-1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,13]]},"references-count":46,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["1134"],"URL":"https:\/\/doi.org\/10.1007\/s44163-026-01134-1","relation":{},"ISSN":["2731-0809"],"issn-type":[{"value":"2731-0809","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,13]]},"assertion":[{"value":"10 October 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"6 March 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"13 April 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"Not applicable. Ethical review and approval were not required for this study because it involved an anonymous questionnaire administered to competent adult university students, posed minimal risk, and did not involve the collection or processing of personal data or special-category data. In accordance with the EU General Data Protection Regulation (Reg. (EU) 2016\/679, Recital 26) and the Italian Privacy Code (Legislative Decree 196\/2003, as amended by Legislative Decree 101\/2018), anonymous information falls outside the scope of data-protection obligations. The study did not fall under healthcare or clinical research frameworks requiring ethics-committee review. Informed consent was obtained from all participants. Participation was voluntary and limited to adults (\n                      \n                      18 years). Prior to participation, instructors explained the purpose of the study in class and participants were informed through an introductory screen in the questionnaire. Only students who agreed to participate were included. No personal or identifying data were collected, and participants could discontinue participation at any time.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval and consent to participate"}},{"value":"Not applicable. The study did not collect personal or identifiable data, and all reported results are based solely on anonymous responses.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent for publication"}},{"value":"The authors declare no conflict of interest.","order":4,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"487"}}