A set of physics-based imaging benchmarks for testing the speed of simulating digital image correlation and infrared thermography using pyvale.
The benchmark cases are describes below and z case can be retrieved using its index in the case list or by using its string tag as follows:
import imagebenchmarks as ib
(case_ident,case_mesh,case_camera) = ib.load_benchmark_by_index(case_index)
(case_ident,case_mesh,case_camera) = ib.load_benchmark_by_tag(case_tag)The case list can also be retrieved using the following: case_list = ib.load_case_list(). The mesh data (nodal coordinates, connectivity table and field to render) for the case is stored as a pyvale.CameraMesh data class and the camera data for the case is stored as a pyvale.CameraData data class. Each pyvale.CameraData.field_to_render contains 8 time steps to render along with a zero frame at the start.
The benchmarks should be run as follows:
- Average render time for a single frame rendering the last field30 times sequentially. This case only allows parallelisation within a single frame.
- Average total time to render all 8 frames 30 times. This case allows for rendering the 8 frames in parallel.
Simulation: Simple Multi-Physics Cube
A 10mm cube undergoing thermo-mechanical loading as shown in the figure below. The cube is fixed on its bottom edge and the top edge is displaced 1mm in tension over 20 time steps. A heat flux is applied to the top surface and heat transfer coefficient is applied to the bottom surface inducing a temperature gradient in the same direction the is loaded mechanically. The displacement and temperature fields are output for all nodes. This simulation only contains a minimum number of elements for testing rendering algorithms on different element types including higher order elements. The mesh has 1 division per edge for tetrahedral and 2 divisions per edge for hexahedral resulting in a total of 24 3D elements per simulation. When the surface mesh is extracted there are 4 elements per face of the cube giving 24 surface elements for both tetrahedral and hexahedral meshes.
 |
|---|
| Paraview rendering of the cube test case for the tetrahedral mesh showing the temperature fieldpv. |
Simulation: Mechanical Plate
A 3D thin plate with a hole in the center loaded in tension as shown in the figure below. Uses tetrahedral 4-node linear elements with the surface mesh extracted to 3 node linear triangles. Higher order and quadrilateral meshes are included in the later benchmarks denoted: 'quadtri' for quadratic triangles, 'linquad' for linear quadrilaterals and 'quadquad' for quadratic quadrilaterals. The vertical displacement field disp_y is to be rendered as the image. Plate mesh bounding box: X [0,100], Y [0,130], Z [0,2] mm.
 |
|---|
| Paraview rendering of the plate test case for the coarsest mesh showing the vertical displacement field. |
Cameras
The following cameras form the benchmark list:
- "1Mpx" = 1280x960 pixels, pixel size = 5.3x5.3 micron
- "5Mpx" = 2464x2056 pixels, pixel size = 3.45x3.45 micron
- "24Mpx" = 5328x4608 pixels, pixel size = 2.74x2.74 micron
All cameras use a 50mm focal length lens. Two pixel subsampling (anti-aliasing) cases are analysed per camera: 1) 1 sample per pixel, 2) 2x2 subsamples per pixel.
Camera Position
The camera is automatically positioned such that the center of the pixel array points at the center of the mesh. For the plate simulation the camera is rotated -30 degrees about the Y axis to view the plate at an angle.
The imaging distance is also automatically set based on the 'crop' / 'no crop' condition. For the 'crop' condition the camera is positioned to include 75% of the bounding box enclosing the mesh projected onto the camera viewing plane. For the no crop condition the camera is placed to include 105% of the bounding box enclosing the mesh projected onto the camera viewing plane (i.e. a 5% buffer is placed around the projected bounding box to ensure all elements are within the field of view).
The list of all benchmark cases is stored in a json file which can be retrieved using the imagebenchmark.get_case_list() function. The current benchmark list is given below. Each benchmark uses
[
"case0_plate_lintri_1Mpx_1subsamp_nocrop_11776elems",
"case1_plate_lintri_1Mpx_2subsamp_nocrop_11776elems",
"case2_plate_lintri_1Mpx_1subsamp_crop_11776elems",
"case3_plate_lintri_1Mpx_2subsamp_crop_11776elems",
"case4_plate_lintri_5Mpx_1subsamp_nocrop_11776elems",
"case5_plate_lintri_5Mpx_2subsamp_nocrop_11776elems",
"case6_plate_lintri_5Mpx_1subsamp_crop_11776elems",
"case7_plate_lintri_5Mpx_2subsamp_crop_11776elems",
"case8_plate_lintri_24Mpx_1subsamp_nocrop_11776elems",
"case9_plate_lintri_24Mpx_2subsamp_nocrop_11776elems",
"case10_plate_lintri_24Mpx_1subsamp_crop_11776elems",
"case11_plate_lintri_24Mpx_2subsamp_crop_11776elems",
"case12_plate_lintri_1Mpx_1subsamp_nocrop_65296elems",
"case13_plate_lintri_1Mpx_2subsamp_nocrop_65296elems",
"case14_plate_lintri_1Mpx_1subsamp_crop_65296elems",
"case15_plate_lintri_1Mpx_2subsamp_crop_65296elems",
"case16_plate_lintri_5Mpx_1subsamp_nocrop_65296elems",
"case17_plate_lintri_5Mpx_2subsamp_nocrop_65296elems",
"case18_plate_lintri_5Mpx_1subsamp_crop_65296elems",
"case19_plate_lintri_5Mpx_2subsamp_crop_65296elems",
"case20_plate_lintri_24Mpx_1subsamp_nocrop_65296elems",
"case21_plate_lintri_24Mpx_2subsamp_nocrop_65296elems",
"case22_plate_lintri_24Mpx_1subsamp_crop_65296elems",
"case23_plate_lintri_24Mpx_2subsamp_crop_65296elems",
"case24_plate_lintri_1Mpx_1subsamp_nocrop_250496elems",
"case25_plate_lintri_1Mpx_2subsamp_nocrop_250496elems",
"case26_plate_lintri_1Mpx_1subsamp_crop_250496elems",
"case27_plate_lintri_1Mpx_2subsamp_crop_250496elems",
"case28_plate_lintri_5Mpx_1subsamp_nocrop_250496elems",
"case29_plate_lintri_5Mpx_2subsamp_nocrop_250496elems",
"case30_plate_lintri_5Mpx_1subsamp_crop_250496elems",
"case31_plate_lintri_5Mpx_2subsamp_crop_250496elems",
"case32_plate_lintri_24Mpx_1subsamp_nocrop_250496elems",
"case33_plate_lintri_24Mpx_2subsamp_nocrop_250496elems",
"case34_plate_lintri_24Mpx_1subsamp_crop_250496elems",
"case35_plate_lintri_24Mpx_2subsamp_crop_250496elems",
"case36_plate_linquad_1Mpx_1subsamp_nocrop_5888elems",
"case37_plate_linquad_1Mpx_2subsamp_nocrop_5888elems",
"case38_plate_linquad_1Mpx_1subsamp_crop_5888elems",
"case39_plate_linquad_1Mpx_2subsamp_crop_5888elems",
"case40_plate_linquad_5Mpx_1subsamp_nocrop_5888elems",
"case41_plate_linquad_5Mpx_2subsamp_nocrop_5888elems",
"case42_plate_linquad_5Mpx_1subsamp_crop_5888elems",
"case43_plate_linquad_5Mpx_2subsamp_crop_5888elems",
"case44_plate_linquad_24Mpx_1subsamp_nocrop_5888elems",
"case45_plate_linquad_24Mpx_2subsamp_nocrop_5888elems",
"case46_plate_linquad_24Mpx_1subsamp_crop_5888elems",
"case47_plate_linquad_24Mpx_2subsamp_crop_5888elems",
"case48_plate_quadtri_1Mpx_1subsamp_nocrop_11776elems",
"case49_plate_quadtri_1Mpx_2subsamp_nocrop_11776elems",
"case50_plate_quadtri_1Mpx_1subsamp_crop_11776elems",
"case51_plate_quadtri_1Mpx_2subsamp_crop_11776elems",
"case52_plate_quadtri_5Mpx_1subsamp_nocrop_11776elems",
"case53_plate_quadtri_5Mpx_2subsamp_nocrop_11776elems",
"case54_plate_quadtri_5Mpx_1subsamp_crop_11776elems",
"case55_plate_quadtri_5Mpx_2subsamp_crop_11776elems",
"case56_plate_quadtri_24Mpx_1subsamp_nocrop_11776elems",
"case57_plate_quadtri_24Mpx_2subsamp_nocrop_11776elems",
"case58_plate_quadtri_24Mpx_1subsamp_crop_11776elems",
"case59_plate_quadtri_24Mpx_2subsamp_crop_11776elems",
"case60_plate_quadquad_1Mpx_1subsamp_nocrop_5888elems",
"case61_plate_quadquad_1Mpx_2subsamp_nocrop_5888elems",
"case62_plate_quadquad_1Mpx_1subsamp_crop_5888elems",
"case63_plate_quadquad_1Mpx_2subsamp_crop_5888elems",
"case64_plate_quadquad_5Mpx_1subsamp_nocrop_5888elems",
"case65_plate_quadquad_5Mpx_2subsamp_nocrop_5888elems",
"case66_plate_quadquad_5Mpx_1subsamp_crop_5888elems",
"case67_plate_quadquad_5Mpx_2subsamp_crop_5888elems",
"case68_plate_quadquad_24Mpx_1subsamp_nocrop_5888elems",
"case69_plate_quadquad_24Mpx_2subsamp_nocrop_5888elems",
"case70_plate_quadquad_24Mpx_1subsamp_crop_5888elems",
"case71_plate_quadquad_24Mpx_2subsamp_crop_5888elems",
"case72_cube_TET4_1Mpx_1subsamp_nocrop_24elems",
"case73_cube_TET4_1Mpx_2subsamp_nocrop_24elems",
"case74_cube_TET4_1Mpx_1subsamp_crop_24elems",
"case75_cube_TET4_1Mpx_2subsamp_crop_24elems",
"case76_cube_TET4_5Mpx_1subsamp_nocrop_24elems",
"case77_cube_TET4_5Mpx_2subsamp_nocrop_24elems",
"case78_cube_TET4_5Mpx_1subsamp_crop_24elems",
"case79_cube_TET4_5Mpx_2subsamp_crop_24elems",
"case80_cube_TET4_24Mpx_1subsamp_nocrop_24elems",
"case81_cube_TET4_24Mpx_2subsamp_nocrop_24elems",
"case82_cube_TET4_24Mpx_1subsamp_crop_24elems",
"case83_cube_TET4_24Mpx_2subsamp_crop_24elems",
"case84_cube_TET10_1Mpx_1subsamp_nocrop_24elems",
"case85_cube_TET10_1Mpx_2subsamp_nocrop_24elems",
"case86_cube_TET10_1Mpx_1subsamp_crop_24elems",
"case87_cube_TET10_1Mpx_2subsamp_crop_24elems",
"case88_cube_TET10_5Mpx_1subsamp_nocrop_24elems",
"case89_cube_TET10_5Mpx_2subsamp_nocrop_24elems",
"case90_cube_TET10_5Mpx_1subsamp_crop_24elems",
"case91_cube_TET10_5Mpx_2subsamp_crop_24elems",
"case92_cube_TET10_24Mpx_1subsamp_nocrop_24elems",
"case93_cube_TET10_24Mpx_2subsamp_nocrop_24elems",
"case94_cube_TET10_24Mpx_1subsamp_crop_24elems",
"case95_cube_TET10_24Mpx_2subsamp_crop_24elems",
"case96_cube_HEX8_1Mpx_1subsamp_nocrop_24elems",
"case97_cube_HEX8_1Mpx_2subsamp_nocrop_24elems",
"case98_cube_HEX8_1Mpx_1subsamp_crop_24elems",
"case99_cube_HEX8_1Mpx_2subsamp_crop_24elems",
"case100_cube_HEX8_5Mpx_1subsamp_nocrop_24elems",
"case101_cube_HEX8_5Mpx_2subsamp_nocrop_24elems",
"case102_cube_HEX8_5Mpx_1subsamp_crop_24elems",
"case103_cube_HEX8_5Mpx_2subsamp_crop_24elems",
"case104_cube_HEX8_24Mpx_1subsamp_nocrop_24elems",
"case105_cube_HEX8_24Mpx_2subsamp_nocrop_24elems",
"case106_cube_HEX8_24Mpx_1subsamp_crop_24elems",
"case107_cube_HEX8_24Mpx_2subsamp_crop_24elems",
"case108_cube_HEX20_1Mpx_1subsamp_nocrop_24elems",
"case109_cube_HEX20_1Mpx_2subsamp_nocrop_24elems",
"case110_cube_HEX20_1Mpx_1subsamp_crop_24elems",
"case111_cube_HEX20_1Mpx_2subsamp_crop_24elems",
"case112_cube_HEX20_5Mpx_1subsamp_nocrop_24elems",
"case113_cube_HEX20_5Mpx_2subsamp_nocrop_24elems",
"case114_cube_HEX20_5Mpx_1subsamp_crop_24elems",
"case115_cube_HEX20_5Mpx_2subsamp_crop_24elems",
"case116_cube_HEX20_24Mpx_1subsamp_nocrop_24elems",
"case117_cube_HEX20_24Mpx_2subsamp_nocrop_24elems",
"case118_cube_HEX20_24Mpx_1subsamp_crop_24elems",
"case119_cube_HEX20_24Mpx_2subsamp_crop_24elems",
"case120_cube_HEX27_1Mpx_1subsamp_nocrop_24elems",
"case121_cube_HEX27_1Mpx_2subsamp_nocrop_24elems",
"case122_cube_HEX27_1Mpx_1subsamp_crop_24elems",
"case123_cube_HEX27_1Mpx_2subsamp_crop_24elems",
"case124_cube_HEX27_5Mpx_1subsamp_nocrop_24elems",
"case125_cube_HEX27_5Mpx_2subsamp_nocrop_24elems",
"case126_cube_HEX27_5Mpx_1subsamp_crop_24elems",
"case127_cube_HEX27_5Mpx_2subsamp_crop_24elems",
"case128_cube_HEX27_24Mpx_1subsamp_nocrop_24elems",
"case129_cube_HEX27_24Mpx_2subsamp_nocrop_24elems",
"case130_cube_HEX27_24Mpx_1subsamp_crop_24elems",
"case131_cube_HEX27_24Mpx_2subsamp_crop_24elems"
]