Tweak settings for better color on northern lights and the night sky

[gigaklatch]

So, last night, right in the middle of the Perseid meteor shower some northern lights appeared.


I am glad that my camera caught as well as it did, however, I am a bit bummed by the lack of color. I took some photos with both my phone and my DSLR and in the pictures taken with those devices the aurora had som amazingly deep colors, yellow, green, pink, red, blue.

How should I change the settings on my allsky to allow it to display colors like that as well? Should I lower the exposures and stretching, the northern lights might be clipping to white?

There are some color to the milky way and the stars, but they are quite bland.

Please help

Kind regards

[aaronwmorris]

Do you have a detection mask setup? That would be the first thing I would try. This would allow the exposure to be balanced to more areas of the sky (including the edges).

Second, you may try boosting your saturation somewhere to 1.25 to 1.5. This can really help with your star colors as well.

[gigaklatch]

So, I made a detection mask and used it during the following night. Luckily, the aurora did return. The result can be seen below:

Here is a photo, straight out of my Nikon D750, also a 30 sec exposure, depicting the exact same moment for color reference:

This camera was aimed at a patch of the sky corresponding to the bottom centre portion of the allsky image.

I did have the foresight to have save FITS enabled at the Allsky camera so here follows a link to a google drive folder containing the fits for the very same image shown first in this post:
https://drive.google.com/drive/folders/1L-tK0v5m-WW4sWBm8lPziqcONn-ONSpd?usp=sharing

Playing around with the FITS-processing tool in INDI-AllSky I can not for the life of me get the colors anywhere near the ones from my DSLR. And the portion of it closest to the horizon stays blown out.

Another thing I tried was opening the FITS-file in ASI Fits viewer, and the colors are still off, but seemingly not as overexposed:

[aaronwmorris]

Can you post a view of your detection mask?

[gigaklatch]

Certainly,

[aaronwmorris]

Thanks!

[aaronwmorris]

I was able to import the FITS file into on of my lab systems and play with the settings in the FITS Processor. There are a couple of things going on here...

1. White Balance. Your DSLR is going to do a much better job at this because its software/hardware is designed around the frequency response of the sensor it was designed around. indi-allsky does not really perform any type of White Balancing. The software option for white balance really just tries to adjust the average (the peak of the histogram) of each color to more closely align.... which is basically "make things more white" :-)
2. Even with the detection mask, the area of interest is everything within the mask. Its best that the subject occupies the majority of view, but when things are mostly at the edge.... the calculation is still done on the whole area of the mask. This is not always what you would want to happen.
3. The stretching algorithm did not really do any favors to the colors. In fact, the stretching kind of washed things out and pushed the color toward "white". Disabling stretching resulted in a better result for the aurora. Actually, doing a more aggressive stretch was a bit better, setting the deviations value below 2 was a little better. It is unfortunate, because the stretching really brings out things like the milky way.
4. Boosting the saturation helps a little bit with all sky cameras... a setting of 1.25 really helps the colors pop a bit more.

I wish I had better algorithms for white balancing. Dealing with RAW data, it is very limiting to not to be available.

[aaronwmorris]

There is another aspect here with regard to the performance of the DSLR which makes this a bit of an unfair comparison.

The DSLR probably has at least an APS-C sized sensor with a lens that has better optics, a much greater aperture (~5-10mm), and is probably a bit slower in F-stop, and a more limited FoV.

All sky lenses are usually very fast, like f/2.0-2.5, tiny aperture (~1mm), and a huge FoV. The sensors are also relatively tiny. All of these aspects severely limit the results.

The HQ camera (via libcamera) can return some impressive results due to the AWB features native to libcamera. However, there are some side effects to using the AWB (if you enable it), namely, it requires multiple internal exposures to produce a pretty picture. If you take a 30s exposure, it might take 90-120s to produce the 30s exposure, which limits the number of pictures you take at night. There is no free lunch.

[aaronwmorris]

As for the SV705c, that camera should be great. However, for color fidelity, I do not think the sensor has an IR-Cut filter which can be a challenge for white balance.

Getting colors correct with astronomy cameras has been the subject of many flame wars. :-)

I would welcome the comparison with AllSkEye.

[gigaklatch]

Yeah, the DSLR even has a full frame sensor. So the area of the sensor is around 12 times greater, the area of each pixel about 4 times that of the IMX585. The lens used was a Samyang 14 mm lens stopped down to F/4. Not fair at all. I do not expect to get colors to that degree, but at least had hoped to achieve something at least a little bit reminiscent.

You are correct that the SV705C does not have any IR-CUT filter. I believe, also, that the IMX585 sensor is very sensitive to IR-light:

The lens used on the SV705C is found here:
https://www.aliexpress.com/item/4000905611143.html?spm=a2g0o.productlist.main.11.e9dfim5lim5l5j&algo_pvid=4c8297ac-30c7-493a-9c21-878d429ee863&algo_exp_id=4c8297ac-30c7-493a-9c21-878d429ee863-5&pdp_npi=4%40dis%21SEK%21710.51%21710.51%21%21%2166.11%2166.11%21%40211b81b117237416327808284eb37b%2110000010484019276%21sea%21SE%214220988501%21X&curPageLogUid=6I6U3ggTjRje&utparam-url=scene%3Asearch%7Cquery_from%3A#nav-specification

It is stated that the lens has IR-correction, not entirely sure what that means. It is unlikely that I can get any answer about that from the seller, I have not had much luck with sellers on Aliexpress, they seem to know nothing about the things that they have on offer.

Whatever that IR-correction does maybe I could look into having an IR-cut filter placed between the back of the lens and the protective window of the camera. I do not believe that there is a lot of space between those two surfaces, so it would have to be quite a thin filter.

One of the cheaper lenses I tried for my Pi HQ camera had a square piece of IR-cut filter crudely glued to the back of it.

[aaronwmorris]

According to this page: https://www.svbony.com/blog/parameters-comparison-of-sv305-to-sv705c-cameras/

The SV705c just has the AR Anti-Reflective window.

[gigaklatch]

Then we all agree on that point.

[gigaklatch]

One thing I did note during the Aurora was that. I have a couple of Wyze Cam V3 pointed where the Aurora did show up. Using the color mode on those one could clearly see there was something there, a bit of pink and green cloudiness. When switching the cameras over to their IR-modes, where the IR-cut filter is removed, the Aurora was even more clearly visible, though no longer with any color to them, as the IR mode switchess to black and white. The Aurora showing up more clearly suggests that there is a lot of IR light coming off of it. And with all of the pixels on the SV705C being very sensitive to IR it would mean that all the detects those photons, R, G as well as B, with no discrimination towards any particular color, hence the extreme brightness on my all sky images.

A somewhat cryptic and difficult to understand reply from the seller of my lens on Aliexpress suggests that the IR-correction helps in bringing the IR-wavelengths into focus together with the visible spectrum light. The lens seem to be meant for surveillance application, so it would not make much sense to have IR blocked.

Given all of the above, it seems to me that it would make sense to try and add an IR-cut filter between the camera and lens somehow. Like I mentioned earlier, there is not a lot of room to do so, so it might not be possible, but maybe it can be done.

[aaronwmorris]

Consider the option of replacing the window glass. ZWO sells replacement windows for their cameras, so it gives you the option to swap between AR and IR options. There might be a window option that could work for your camera.

I removed the IR filter in my SV305 a few years ago. Unfortunately, it was glued in and removing required breaking the filter.

There are a lot of variations in IR filters. DSLR IR filters usually cut off just before 650nm, just before Hydrogen Alpha, which is why people have to astromod their DSLRs. Astronomy IR filters usually go up to ~700nm so HA is not removed.

Also, newer camera sensors are considerably more sensitive to IR than even 5 years ago.

[gigaklatch]

Oh, I had not even considered that. Sound like a cool route to go down.

ZWO have have IR cut protective windows with 21, 25 and 60 mm diameter. There is a thread on cloudy nights, where a guy opens up a SV705C to remove dust on the inside of the protective window. Doing some measurements in GIMP on the images of the inside of the camera posted in that thread, it looks like the window is more like 36 mm. So maybe I could get a regular unmounted 36mm luminance filter? They usualle cut away away wavelengths outside of a 400 to 700 nm range. Quite costly, but might be worth it if it does the trick. MIght be a bit thick though? Baader's have 2 mm thickness, but perhaps there is room for that.

I would try and contact SVBony, but I have horrible experiences from dealing with them in the past, so it feels unlikely they could be of any help at all.

[gigaklatch]

At Aliexpress there seem to be available some round 36 mm(also 32.5mm if that would be a better fit)IR cut filters, they are stated to cut at 650 nm. Seems like the blue and green pixels of the IMX585 starts to increase in sensitivity again at about that wavelength, so might be good to cut there.

The thickness of these are not clearly known

HOw to know if these are of any kind of decent quality?

[aaronwmorris]

You will probably have to remove the lid from your SV705c to see what shape and size you need.

To be fair, you probably do not need to worry about quality so much. In traditional astrophotography, you want a "flat" filter (eg 1/10 wave or better) to reduce distortions, but with all sky cameras, the lens causes so much distortion, you would probably never notice a bad filter anyway.