35 lines
1.9 KiB
Markdown
35 lines
1.9 KiB
Markdown
# video2geoframes.py
|
|
|
|
Python program to generate a collection of geotagged images from a video and a GPS track.
|
|
|
|
Designed for contribution to street-level imagery projects like Mapillary or Panoramax.
|
|
|
|
## Notes
|
|
|
|
This program was originally designed for personal use to contribute to Mapillary then Panoramax with a smartphone
|
|
(video mode) or a dashcam.
|
|
|
|
This v1 is the last version of the original code, published "as it is" with certain known bugs and limitations.
|
|
|
|
Since June 2024, a [v2](https://git.luc-geo.fr/lumathieu/video2geoframes.py/src/branch/v2) was started to develop with
|
|
the aim of :
|
|
* enhance performances
|
|
* fix known bugs
|
|
* clean code
|
|
* remove non-Python dependencies
|
|
* implement more flexible configuration with TOML.
|
|
|
|
## ⚠ Bugs and limitations
|
|
|
|
| Report | Workaround |
|
|
|------------------------------------------------------------------------------------------------------|-------------------------------------------------------|
|
|
| [bug] in non-timelapse mode, **FFmpeg** doesn't extract first video frame and shifts every timestamp | 🔃 adding chosen framesampling interval to GPS offset |
|
|
| [bug] **ExifTool** timestamp very long (~1 frame / s) | ❌ |
|
|
| [bug] uncontrolled incrementation if export directory is on a Samba share (ZFS) | ❌ |
|
|
| 9999 frames limit (naming model) | ❌ |
|
|
|
|
## License
|
|
|
|
This repository, except dependencies, is licensed under **GNU GPL v3**.
|
|
|
|
Dependencies are included in repository for development and keep their original license. |