Search for: All records

The popularization of Text-to-Image (T2I) diffusion models enables the genera- tion of high-quality images from text descriptions. However, generating diverse customized images with reference visual attributes remains challenging. This work focuses on personalizing T2I diffusion models at a more abstract concept or category level, adapting commonalities from a set of reference images while creating new instances with sufficient variations. We introduce a solution that al- lows a pretrained T2I diffusion model to learn a set of soft prompts, enabling the generation of novel images by sampling prompts from the learned distribution. These prompts offer text-guided editing capabilities and additional flexibility in controlling variation and mixing between multiple distributions. We also show the adaptability of the learned prompt distribution to other tasks, such as text- to-3D. Finally we demonstrate effectiveness of our approach through quantitative analysis including automatic evaluation and human assessment. 

Neutron elastic scattering cross sections on natural carbon serve as a reference standard in the incident energy range 10 eV to 1.8 MeV. The 2017 standards evaluation [1, 2] is 0.5 to 2.0% higher in that energy range than the 2006 standards evaluation [3]. In addition the ENDF/B-VIII.0 release split the natural carbon cross sections into the isotopes¹²C,¹³C, and¹⁴C for the first time. These details call for the re-measurement of the¹³C cross sections in sensitive regions. Ten elastic scattering angular distributions were recently measured for incident neutron energies between 0.5 and 3.25 MeV at the University of Kentucky Accelerator Laboratory (www.pa.uky.edu/accelerator/) using nanosecond pulsed beams and time-of-flight techniques. An overview of neutron production and detection, the new digital data acquisition system, and data analysis will be presented. Results are compared with data from previous measurements and database evaluations. 

Abstract Euplectella aspergillummarine sponge spicules are renowned for their remarkable strength and toughness. These spicules exhibit a unique concentric layering structure, which contributes to their exceptional mechanical resistance. In this study, finite element method simulations were used to comprehensively investigate the effect of nested cylindrical structures on the mechanical properties of spicules. This investigation leveraged scanning electron microscopy images to guide the computational modeling of the microstructure and the results were validated by three-point bending tests of 3D-printed spicule-inspired structures. The numerical analyses showed that the nested structure of spicules induces stress and strain jumps on the layer interfaces, reducing the load on critical zones of the fiber and increasing its toughness. It was found that this effect shows a tapering enhancement as the number of layers increases, which combines with a threshold related to the 3D-printing manufacturability to suggest a compromise for optimal performance. A comprehensive evaluation of the mechanical properties of these fibers can assist in developing a new generation of bioinspired structures with practical real-world applications. 

Possessing a unique combination of properties that are traditionally contradictory in other natural or synthetical materials, Ga-based liquid metals (LMs) exhibit low mechanical stiffness and flowability like a liquid, with good electrical and thermal conductivity like metal, as well as good biocompatibility and room-temperature phase transformation. These remarkable properties have paved the way for the development of novel reconfigurable or stretchable electronics and devices. Despite these outstanding properties, the easy oxidation, high surface tension, and low rheological viscosity of LMs have presented formidable challenges in high-resolution patterning. To address this challenge, various surface modifications or additives have been employed to tailor the oxidation state, viscosity, and patterning capability of LMs. One effective approach for LM patterning is breaking down LMs into microparticles known as liquid metal particles (LMPs). This facilitates LM patterning using conventional techniques such as stencil, screening, or inkjet printing. Judiciously formulated photo-curable LMP inks or the introduction of an adhesive seed layer combined with a modified lift-off process further provide the micrometer-level LM patterns. Incorporating porous and adhesive substrates in LM-based electronics allows direct interfacing with the skin for robust and long-term monitoring of physiological signals. Combined with self-healing polymers in the form of substrates or composites, LM-based electronics can provide mechanical-robust devices to heal after damage for working in harsh environments. This review provides the latest advances in LM-based composites, fabrication methods, and their novel and unique applications in stretchable or reconfigurable sensors and resulting integrated systems. It is believed that the advancements in LM-based material preparation and high-resolution techniques have opened up opportunities for customized designs of LM-based stretchable sensors, as well as multifunctional, reconfigurable, highly integrated, and even standalone systems.