Paste this in Notepad++, this editor has good feature for dublicate selected strings: Ctrl+D
MKDIR “FRAMES_01”C:\python27\python.exe C:\python27\Scripts\mapillary_tools video_process --import_path "FRAMES_01" --video_file "S1780001.MP4" --user_name "trolleway" --advanced --video_sample_interval 0.3 --geotag_source "gpx" --use_gps_start_time --geotag_source_path "20180617-133016.gpx"
In Total Commander make the folder with video active, and type “cmd” in lower textbox. A terminal window will opened with current working directory.
Copy commands from Notepad++ and paste these commands in terminal.
In case of error with gpx file reading: try open and resave it in JOSM. It may be when incorrect stopping of gpx logger application.
A folder will be created and polulated with frames.
Now you sholud check and adjust coordinates correlation beetwen video and gpx track. A trying to figure value of offset_time key is hard, so i use JOSM.
Start JOSM, install plugins photo_geotagging and photoadjust.
Open gpx track, add OpenStreetMap Carto TMS layer
Drag-and-drop all video frames into JOSM. Open context menu of layer with photos, click to “Position adjust”
Using marked controls, move frames to exact location.
Select some frame with landmark, like corner of big building, or road crossings, and tune time offset to presice location.
Click to contex menu of picture layer, and search for “Write coordinates to picture headers”. If you miss some commands, you prorably missed to install photo* plugins.
Now we have frames with coordinates in EXIF tags. Now we need to set picture direction. Copy these command from Notepad++ to terminal
C:\python27\python.exe C:\python27\Scripts\mapillary_tools process --import_path "FRAMES_01" --user_name "trolleway" --advanced --rerun --interpolate_directions --offset_angle 270I use offset_angle 270 when i shot from left side of moving vehicle, 90 for right side.
Now pictures ready for upload, you can use a upload script, or web interface at mapillary side. That is all.
