abhishekbajpayee / openfv Goto Github PK

• parse this field in parse_reconstruction_settings() instead of parse_refocus_settings()
• add the field definition in the function that has all the fields defined

This is likely due to functionality to allow it to handle relative paths but it is not aware when an absolute path is passed and prepends the path anyway...

• update sa_reconstruct to render mask stacks
• filter folder containing mask stacks in the PIV matlab function
• output mask_dir field in the PIV yaml file
• output other commented variables for masking

Does not make sense to have OpenMP as a dependency just for timing functions.

In dividing the images during additive refocusing, CPUrefocus* functions all currently divide by 9 when actually they should be diving by num_cams_.

[x] accept different config file (parse_reconstruction_settings)
[x] auto calculate dz based on voxel size but leave option for manual dz

Hi,

I have met a matlab problem.

It seems that there is a missing folder "/tomo/lib_tomo“, which should contain include mart_parbatch.F90, etc

Best,

Problem making Open FV, here is my isssue:
[ 5%] Building NVCC (Device) object src/CMakeFiles/cuda_openfv.dir/modules/./cuda_openfv_generated_cuda_lib.cu.o
/home/anthonybisulco/Downloads/openfv/src/modules/cuda_lib.cu:48:22: fatal error: cuda_lib.h: No such file or directory
#include "cuda_lib.h"
^
compilation terminated.
CMake Error at cuda_openfv_generated_cuda_lib.cu.o.cmake:206 (message):
Error generating
/home/anthonybisulco/Downloads/openfv/bin/src/CMakeFiles/cuda_openfv.dir/modules/./cuda_openfv_generated_cuda_lib.cu.o

make[2]: *** [src/CMakeFiles/cuda_openfv.dir/modules/./cuda_openfv_generated_cuda_lib.cu.o] Error 1
make[1]: *** [src/CMakeFiles/cuda_openfv.dir/all] Error 2
make: *** [all] Error 2

refocus sample doesn't build and CMakeLists still reference sa-ptv

Add description of "shift" input for time shifts and input format

Add x y coordinate shifts as config file input variable for refocusing

• add function to write this file
• add m function to read this file and call PIV run
• add auto writing of this file to sa_reconstruct

Refocusing mtiff results in saturated images in live view

Live refocus crashes if you scroll beyond the number of frames in an image sequence or scroll to a negative frame number (i.e., holding down a key to advance a large amount of time).

• if destination dir exists, check if it is empty or not
• if not empty, write to new dir
• generate new dir in same location
• if empty then write in dir
• if dir don't exist, create and write
• pass frame number vector to class so that frame number dir names are same as numbers in dataset

Ideally with some sort of scheme to pick frames

• images_path
• save_path

This is to fix (as much as possible) some sync errors

Clean it up...

• check if number of frames in each cam folder are equal or not
• check if cam names are read to be identical
• check if folders named after cam names from calibration are missing in images path

If a config file contains both mult = 1 and minlos = 1, refocusing will proceed with mult. Should throw an error telling the user to pick which algorithm by fixing the config file.

undefined symbol: _Z20gpu_calc_refocus_mapRN2cv3gpu6GpuMatES2_fi