The solidified microstructure and carbide precipitation behavior in an S390 high-speed steel processed by electron beam melting (EBM) have been fully characterized. The as-EBM microstructure consists of discontinuous network of very fine primary carbides dispersed in auto-tempered martensite matrix together with a limited amount of retained austenite. The carbide network consists of M 2 C/M 6 C and MC carbides. Both the columnar and near-equiaxed grain structures were found in as-EBM microstructure and the presence of inter-dendritic eutectic carbides assisted in revealing the dendritic solidification nature. The top-layer microstructure observation confirmed that the columnar dendritic structured grains were located adjacent to the micro-melt pool boundary, indicating an epitaxial growth with the average growth direction parallel to the maximum thermal gradient. At the center of the micro-melt pool, the near-equiaxed grains were developed by dendritic growth parallel to the beam traveling direction. The carbide decomposition was revealed by scanning transmission electron microscopy and confirmed by transmission Kikuchi diffraction. The MC carbides (rich in V followed by W) nucleated at the interface between M 2 C (W, Fe, Mo, and Co in the order of significance) and the matrix and then grew from the outside inward, but their nucleation might occur from the M 2 C carbide itself. The thermal effect induced by the adjacent scan lines seems to trigger a solid-state phase transformation of MC fi M 2 C + c-Fe. The elemental migration was theoretically calculated and compared with the experimental results. The high hardness of~65 HRC and good transverse rupture strength of~2500 MPa in as-EBM S390 means that EBM processing can be used to fabricate highly alloyed tool steels. With the help of the post-processing heat treatment, the best Rockwell hardness of 73.1±0.2 HRC and transverse rupture strength of 3012±34 MPa can be obtained.

We advocate the use of differential visual shape metrics to train deep neural networks for 3D reconstruction. We introduce such a metric which compares two 3D shapes by measuring visual, image-space differences between multiview images differentiably rendered from the shapes. Furthermore, we develop a differentiable image-space distance based on mean-squared errors defined over Hard-Net features computed from probabilistic keypoint maps of the compared images. Our differential visual shape metric can be easily plugged into various reconstruction networks, replacing the object-space distortion measures, such as Chamfer or Earth Mover distances, so as to optimize the network weights to produce reconstruction results with better structural fidelity and visual quality. We demonstrate this both objectively, using well-known visual shape metrics for retrieval and classification tasks that are independent from our new metric, and subjectively through a perceptual study.

No abstract

No abstract