RESUMEN
Compared to traditional vat photopolymerization 3D printing methods, pixel blending technique provides greater freedom in terms of user-defined lighting sources. Based on this technology, scientists have conducted research on 3D printing manufacturing for elastic materials, biologically inert materials, and materials with high transparency, making significant contributions to the fields of portable healthcare and specialty material processing. However, there has been a lack of a universal and simple algorithm to facilitate low-cost printing experiments for researchers not in the 3D printing industry. Here, we propose a mathematical approach based on morphology to simulate the light dose distribution and virtual visualization of parts produced using grayscale mask vat photopolymerization 3D printing technology. Based on this simulation, we develop an auto-correction method inspired by circle packing to modify the grayscale values of projection images, thereby improving the dimensional accuracy of printed devices. This method can significantly improve printing accuracy with just a single parameter adjustment. We conducted experimental validation of this method on a vat photopolymerization printer using common commercial resins, demonstrating its feasibility for printing high precision structures. The parameters utilized in this method are comparatively simpler to acquire compared to conventional techniques for obtaining optical parameters. For researchers in non-vat photopolymerization 3D printing industry, it is relatively user-friendly.
RESUMEN
With the development of mobile Internet technology, firms need to complete the entire process of consumer targeting, ad content generation, and ad display in a very short time window. Therefore, computational advertising, such as native ads on social media platforms, has become the mainstream of online advertising with its automation and personalization features. However, computational advertising faces some problems when using artificial intelligence technology to generate content. First, the images should have a significant enough impact on consumers and be easy to adjust to save computational power at the same time; second, the iteration of the computational advertising system relies on consumer behaviors or advertising effectiveness, and firms need to learn the relationship between ad design and consumer behaviors. Under the above two problems, this paper selects visual distance as the main variable, and images can be adjusted by cropping to save computational power. This paper incorporates image design and ad effectiveness metrics into the construal level theory framework, under which the effectiveness metrics can be quickly determined. Following previous studies, we use click-through rate (CTR) to represent the early stage of the sales funnel and a higher construal level and CVR (conversion rate) to represent the later stage of the sales funnel and a lower construal level. Therefore, visually distant images bring distant psychological distance or higher construal level, which can get higher CTR; visually proximate images bring near psychological distance or lower construal level, which can bring higher CVR. These findings suggest that firms can improve the efficiency of their advertising systems and gain more revenue by understanding consumer psychological states.