This is a simple Numpy implementation of the Mertens Fusion Alghoritm.
https://link.raise-uav.com
7a92698dfc
(Doubles processing time, can be switched off) |
||
---|---|---|
cvHDR-Fusion.py | ||
experimental.py | ||
file_utility.py | ||
license.txt | ||
makehdr.py | ||
numpyHDR.py | ||
picamburst.py | ||
readme.md | ||
requirements.txt |
numpyHDR
- Micro Library for HDR image creation using the Mertens Fusion Alghoritm
- Optional Raspicam integration with picamera2 lib
- CLI Application "makehdr.py"
- Processing time on raspi zero ca 15 min with full res 12MP files.
Intention
- numpy and PIL, picamera2 for Raspicam funtionality.
- Whenever dependencies of other bigger image librarys aren´t satisfyable.
- 32-bit armv6 or oudated Kernels etc.
Function
Use makehdr.py CLI app for testing before implementation
- Captures an exposure bracket stack from the raspicam as fast as possible
- Alternatively from image sequences form file
- Either direct processing or save files in between
- Uses a conversion of the Mertens Fusion alghoritm
- Stretches Information to the full spectrum like Contrast or compression
- Lifts the shadows softly with an envelope.
- Clips to Image range and saves via PIL
Setup
- Download and install dependencies from requirements.txt into env
pip install -r requirements.txt
- Install picamera2 and libcamera
pip install picamera2
pip install libcamera
Usage
-
Start makehdr.py
-
python3 makehdr.py
-
Follow commandline instructions
-
Be patient on Raspi zero
-
Example on how to use the library can be found in makehdr.py
Using the library
Import numpyHDR modules
import file_utility as file
import numpyHDR as hdr
import picamburst as pcb
Get image arrays from raspicam
stack = pcb.get_exposure_stack()
OR
Get image arrays from files
stack = file.openImageList(path_list, True)
Process HDR with mertens fusion and post effects
result = hdr.process(stack, 1, 1, 1, True)
Save Result to File
- The filename gets extended by ".jpg" and "_hdr" automatically
file = file.saveResultToFile(result, 'hdr/result', 75)