We thank Shauly and Wolmer for their thoughtful and insightful letter regarding our recent article, “Use of Text-to-Image Artificial Intelligence Model in Preoperative Counseling for Lip-Lift Procedures.”1 Their engagement reflects the evolving and multidisciplinary interest in the integration of artificial intelligence (AI) into surgical counseling and patient education. We appreciate their recognition of our work’s contribution toward reshaping the patient consultation experience. We fully agree that the current limitations of DALL·E2, particularly in terms of anatomical accuracy, scar simulation, and functional modeling, highlight the importance of viewing text-to-image models as adjuncts rather than replacements for established 3-dimensional imaging systems. In fact, as of the time of this writing, DALL·E2 has been deprecated and replaced by newer models, such as DALL·E3, which paradoxically does not currently reproduce the same capacity for iterative, inpainting-based modeling seen in its predecessor. As the authors rightly point out, it remains to be determined whether these limitations are unique to DALL·E2 or generalizable across current text-to-image AI. In our experience and as supported by current literature, even next-generation models, such as Midjourney-v7, continue to fall short in replicating clinically relevant anatomic precision, particularly in applications requiring subtle 3-dimensional facial modeling.2 This suggests that the challenges may reflect broader constraints inherent to this class of models at present. We appreciate the authors’ reference to commercial platforms, such as VECTRA and Crisalix. While VECTRA provides powerful photogrammetric and volumetric tools, the ability to simulate a lip lift remains highly dependent on the user’s manual manipulation and familiarity with the system.3 Crisalix offers a more consumer-facing interface and may be useful in certain applications, such as breast augmentation, where volumetric changes are more generalized.4 However, in our hands, Crisalix may not offer the level of fine detail required for simulating nuanced facial procedures. Ultimately, whether using professional-grade systems or emerging AI tools, all simulation platforms remain limited by their inability to reliably capture the biological variability of healing, scarring, and tissue remodeling that ultimately shape the postoperative outcome. As artificial intelligence and 3-dimensional morphing systems continue to mature, we envision a hybrid workflow in which the structural precision of engineered imaging platforms is complemented by the flexible, patient-facing strengths of generative AI. Importantly, as the authors noted, these tools must be implemented with transparency. AI-generated visualizations should be framed as educational aids, not definitive forecasts of surgical outcomes. In our own practice, we have begun evaluating additional platforms beyond DALL·E, including Midjourney and Adobe Firefly. These tools introduce new possibilities for dynamic and contextual rendering, and we look forward to assessing their role in enhancing the surgical counseling process. Finally, we appreciate the authors’ suggestion to expand the application of these technologies beyond aesthetic lip surgery. We agree that domains such as breast reconstruction, craniofacial procedures, and gender-affirming surgery, where individualized aesthetic and psychosocial outcomes are of paramount importance, may particularly benefit from innovations that help bridge expectation and reality. We echo the call for future studies incorporating patient-reported outcomes, qualitative assessments, and validation against real-world surgical results. DISCLOSURE The authors have no financial support of conflicts of interest to declare in relation to the content of this communication.
Huang et al. (Wed,) studied this question.