Electronically Assisted Astronomy with a Jetson Nano

[victorleyner]

Alain,

Thank you for such quick response. I wanted to follow up a bit on the same items. Does your mount have a large lag time between sending a signal to servos and action? (this would be my main concern in general application for computer vision). Golang is an excellent language to tackle these sorts of problems in efficiently, take a look GoCV GitHub repo and the examples they provide for tracking object recognition etc.

Regarding what I called “speed booster”, I was referring to focal reducers - to for instance go way below F1.4 without spending arm-and-a-leg for a very expensive F0.90 lens. Theoretically, this can be done with a bunch of cheap lenses and with smaller sensor - as long as they are paired correctly. There is an excellent write up by skyandtelescope if you google “of-pixel-size-and-focal-reducers” (can’t post links as the bot will remove my post).

Agree with you regarding image intensifiers - these are basically night vision tubes repurposed for astronomy, they are way too expensive for modest hobbyists. I’ve seen people use these at star parties to reduce the need for even bigger dobsonians, with amazing detailed and largely noiseless in-eyepiece/camera viewing. Way out of reach for me, but was curious if you’ve seen that application.

And one last thing, have you looked into modifying your camera to cool it yourself?

Thanks,

Victor

[easybob]

Hello Victor,

i don’t have any lag time when sending signal to stepper motors. Everything is ok with motors control. I use threads to control the motors. Each motor thread runs and wait for its order to move. With threads, i don’t have lag time issue.

Ok for the focal reducers. F 0.95 lens is a bit expensive for sure. For now, i have an old Canon 50mm F1.4 and it does the job. I have no experience with focal reducer for lens. Using my telescope (Celestron C9.25 F10), i do use focal reducer. In fact, i use 2 focal reducer : 1 Celestron F6.3 reducer and 1 low coast Kepler 0.5x focal reducer. As the IMX178 is rather small, i don’t have big optical issue with the reducers. I have finfd the Sky&Telescope article and i will read it.

Concerning image intensifier, i am curious to see what they can do. Maybe i will have one day the opportunity to test on.

I do not plan to cool my camera.I am not sure cooling my camera will raise the final result, considering i make heavy treatment. I used to have a cooled camera but from what i saw, cooling was useful for long exposures. As i try to get very short exposure, i think cooling is not really important for me.

Alain

[easybob]

Back to JetsonSky.

As noise is now managed (4 different denoise filters) i can try to get improvements in signal software amplification (if noise is not managed correctely, signal software amplification means noise amplification).

I already get signal management routines but i need better signal amplification routine.

For now, i had very simple amplification routine (just amplification factor). I will soon implement selective amplification factor based on gaussian curve. It will be a kind of band pass filter tranfert function.

I hope this will help me to get better results with very small signal (better signal amplification) and in the same time to preserve high signal (no high or medium signal amplification).

The equations are ok and the routine is in my mind (2 parameters to set center of the band pass filter and width of the filter). Just have to write it.

When this will be done, i would also try to find time to put the camera in my car and drive a while at night to see if i can get significant improvements with my system in that kind of environement.

Alain

[victorleyner]

Hey Alain,

Awesome stuff! Forgive me if this question may have already surfaced, but have you tried optix-denoiser already and if so was there a reason it does not make sense here?

Also would all of this work on Nano - 4 denoisers and signal amplification? Also thank you for all your creative work!

Thanks,

Victor

[easybob]

Hello Victor,

I was not aware about optix-denoiser. It seems interesting. I will take a look at it.

For now, i use classical algorithms in order to get fast treatment (i am afraid optix-denoise needs some time to perform denoise) and more important, AI denoise filter could bring some informations which does not really exist in the sky. With classical algorithms, i can control what i do. With AI filter, it is more difficult.

My software also works with Jetson Nano. For now, it is a bit complicated for me to work seriously with the Jetson SBC. But i will try to share up to date versions of JetsonSky (still free of use for personal use only).

Alain

[easybob]

Hello,

well, my selective amplification routine is ok and works perfectly.

First test on a daylight capture :

As you can see, with linear amplification, you can’t manage high lights. With selective amplification, i can choose the signal i want to amplify and preserve high lights. That’s really great.

As the sky is horrible, i can’t make tests with the sky for now. I will put this filter in my video treatment software to see how it performs with some sky video.

With my first class denoise filter, i think i start to get a really good tool for deep sky real time survey !

Alain

[linuxdev]

I notice that the result is extremely good with respect to that white curtain. It completely saturates with linear, looks really good in raw (for that curtain), and that you get absolutely everything with quality under selective. This would be really nice in a lot of situations, not just astronomy, but I could see combining this for something like sun flares.

[easybob]

Hello linuxdev,

i think that kind of filter could be interesting in many situations. You are right.

Maybe deepsky is the worst situation for signal selective amplification because the signal is really complex and hard to manage. I will need some time to see if this filter is useful for my purpose because i maust make new deepsky videos.

Just wait to see !

Alain