In this work we present Reynolds-averaged Navier-Stokes (RANS) simulations of the flow past the constant design shape of a leading-edge inflatable (LEI) wing. The simulations are performed with a steady-state solver using a k − ω SST turbulence model, covering a range of Reynolds numbers between 105 ≤ and ≤ 15 × 106 and angles of attack varying between −5° and 24°, which are representative for operating conditions in airborne wind energy applications. The resulting force distributions are used to characterize the aerodynamic performance of the wing. We found that a γ − R ¯ e θ transition model is required to accurately predict the occurrence of stall up to at least Re= 3 × 106. The work highlights similarities with the flow past a two-dimensional LEI airfoil, in particular, with respect to flow transition and its influence on the aerodynamic properties. The computed values of the lift and drag coefficients agree well with in-flight measurements acquired during the traction phase of the LEI wing operation. The simulations show that the three-dimensional flow field exhibits a significant cross flow along the span of the wing.

From left to right: input photo (4096 × 4096 pixels), SVBRDF maps (albedo, normal, roughness, height, ambient occlusion) reconstructed with our method, rendering of the reconstructed material and rendering with a constant albedo to visualize the extracted geometry.

No abstract

We present a novel U-Attention vision Transformer for universal texture synthesis. We exploit the natural longrange dependencies enabled by the attention mechanism to allow our approach to synthesize diverse textures while preserving their structures in a single inference. We propose a multi-stage hourglass backbone that attends to the global structure and performs patch mapping at varying scales in a coarse-to-fine-to-coarse stream. Further completed by skip connection and convolution designs that propagate and fuse information at different scales, our U-Attention architecture unifies attention to microstructures, mesostructures and macrostructures, and progressively refines synthesis results at successive stages. We show that our method achieves stronger 2× synthesis than previous work on both stochastic and structured textures while generalizing to unseen textures without fine-tuning. Ablation studies demonstrate the effectiveness of each component of our architecture.