The post number 1000 is mine !
This makes me wish I had bionic eyes so I could see the world in those colors all the time! :)
You could without bionic eyes but I guess it would not be legal
Hello,
i have uploaded a new version of JetsonSky in Github : JetsonSky_Github_V53_11RC.
The changes :
- some Claude AI optimizations
- some personal optimizations
- some bugs fixed
- can now track some birds (or planes) while daylight capture
That’s all.
Alain
Quite clear and beautiful!
Hi linuxdev,
Very few videos and photos from Artemis 2. I expected much more !
Alain
I am guessing they have those photos and videos, but are planning commercial use or use in some sort of future marketing.
The clarity is stunning, and the colors themselves are a work of art.
I will try to find more stuff to get beautiful colours!
Hello,
last night, it was Moon hunting.
I used my Celestron C9.25 which was sleeping for more than a year.
Mirrors setup was ok, nothing moved.
The focal lens was 2.35 meters and the sensor was IMX485.
I used JetsonSky for sensor setup and capture.
Treatment was more traditional : Autostakkert and Registax.
A 6 captures mosaic :
Alain
Hello,
i founded a RAW capture of the Moon from Artemis 2 mission.
A quick treatment with JetsonSky to retrieve the colors of the soils.
I also work on a new HDR algorithm for JetsonSky. I need to test it with in real conditions.
It is quite fast and also run with a single capture. With a 4K video capture, i get quite high treatment framerate (more than 20fps).
I will post things about this ASAP.
Alain
Hello,
despite the fact we no longer get a tremendous support from NVIDIA (as i am not billionaire, i am not interesting people), i will continue to post my work here.
I talk about HDR (useful for specific targets like Orion Nebula for example) and i made a daylight test with various HDR methods and 16 bits captures.
So, i use 16 bits capture and select a specific number of LSB to convert then into 8 bits image. From a unique 16 bits capture, i can create several images with different exposure level in order to calculate HDR image.
I tested various methods :
- Mertens (very low FPS)
- Mean (high FPS)
- Percentile (high FPS)
- Inverse Percentile (high FPS)
The results :
As always, i used CUDA raw kernels (CUPY) to get fast treatments with GPU).
I am very pleased with the treatment time (high FPS) and the Percentile & Inverse Percentile methods (inverse percentile is my own personal method).
Alain
I am reminded of early imaging for CAT/CT scans and MRIs. Actual film was quite high resolution and what we might call high dynamic range, but companies were starting to work on digital versions which could be shared over a network. Those versions were not actually legal to work with for many uses, but there was an accepted method to be able to go digital: The display had a “wheel” (like a rotary encoder) and the image would shift which bits were displayed (it couldn’t display all bits due to both monitor and perhaps other reasons), and when manually shifting back and forth one could see changes or differences which no single image could display. It would be interesting to see what “edge detection” (like Laplacian) would show as you go through the different HDR settings in your video.
Yes, I am sure people have found interesting ways to brake technology limits.
I will try and edge detection filter to see the difference between HDR filters.
Daylight tests won’t really teach me if my HDR filters are interesting with deep sky. I will make tests this autumn.
Alain
Sounds like it will be a fascinating test! Just to make it even more interesting, I wonder what the change of edge detection will show among the different HDR values…more or less “the edge detection of the multiple edge detections”. This would be a direct statement on what the different dynamic ranges can do for people looking at the given ranges.
Hello,
to illustrate the difference between 8 bits capture and the same image in 16 bits & inverse percentile HDR & contrast CLAHE, an other daylight test.
In this test, we can directly compare 8 bits raw video and 16 bits & HDR & Clahe.
In my opinion, the result is very good.
This autumn (or winter), i will try to make a night test, taking a video of the road in my car to see if this method can bring significant improvements of what we see at night while driving. This test will be interesting.
Alain





