Thoughts from Starlight Cascade

Some decent aurora showed last evening from about 21:00 edt to 01:00 this morning.
An allnight video is here:

The startrail is here:

and the keogram here:

and the best still image from 23:19 EDT

Lucky imaging, Firecapture, ROI and cutouts

Planetary Imaging took a Quantum Leap with the advent of Lucky Imaging and digital cameras.
Lucky imaging is the process of taking thousands of very short (millisecond) exposures, analyzing them, taking the best x% and stacking them to get a great signal to noise ration and images that surpass film based, ground based images from only a decade or two ago.

The shorter the exposure, the better chance of capturing a still image freer of distortion caused by atmospheric bad seeing. So the higher frames per second is the measure to work for.

I recently submitted an image of Mars to the BAA along with some information like… 2ms exposure and I was getting 100 frames per second (fps). 1000/2=500 fps theoretical and I was getting only 20% of that. The recommendation to me was to improve! Get a faster FPS! And use the ROI (Region Of Interest) and Cutout boxes in firecapture.

With the full resolution of the ZWO ASI585MC camera (3840×2160 pixels) and a 3.5ms exposure, I got 47 fps, or about 16% efficiency. The ROI reduces, at the camera, the amount of data coming down the usb 3.0 pipeline and results in increased frame rates.

I had been using a ROI of 1200×1200. That increased the fps to 83 from 47 (full resolution) and 29% efficiency. My best test result was using an ROI of 400×400 that achieved 220 fps or 77% efficiency.

There are issues with tracking and keeping the image in the Field of View (FOV), especially in bad seeing, where the image jumps up down left right and in and out) when the ROI is small.. that is why I had a bigger 1200×1200 FOV to start with.

The next feature I started using recently was the cutout box. This centres the image in the FOV and does it faster than the autoalign feature in Firecapture. It also reduces the amount of data saved to file.
Initially I started using a cutout box of 600×600 and am now testing it down to 400×400.

One other new feature that I was unaware of was the ability to have the ROI centre itself. In the past, I would set the ROI and in the 120 seconds of imaging run, the target would stay in the FOV, but maybe not the next or the next runs, without manually adjusting the tracking. This new feature of ROI autocentre means that as the target drifts out to the edge of FOV, it would recentre itself.

Amazing! This would work for at least 10-15 runs that I have tested so far. The limiting factor is when you run out of chip completely.
So.. the summary is:
initial starting performance with my setup for Jupiter imaging:
Celestron C9.25, x1.5 barlow, ZWO ASI585MC
usb cable3m,
HP Mini elitedesk 800 G3 computer
Intel Core I5-6500T; 16GB DDR4; Intel HD Graphics 530; Windows 10 pro 64 bit;
480GB nvme ssd; gigabit ethernet; USB 3.0
.SER file format
firecapture buffer enabled

Two extreme examples, image ID #138 and 136:
Case #1: Full resolution 120second run at 3.5ms exposure hit 47 fps got 5582 frames at 16% efficiency. Using the best 10% in autostakkert! Gave me 558 frames which is a pretty low signal to noise ration giving a notsogreat image quality.
Case #2: Last nights test of ROI 400×400, autocentering, cutout box of 400×400 hit 220fps giving 26091 frames at 77% efficiency. Using the best 10% to stack gives me 2609 frames, 4 times more signal to noise than before.
Looking at the images themselves

Case#1

copyright StarlightCascade Observatory

Case#2

copyright StarlightCascade Observatory

Looking at the images zoomed in a lot, Case#2 image is superior with much less noise resulting from the increased frame rate and more frames to stack as signal. Smaller ROI and cutouts are in fact better!

These were taken in poor seeing and poor transparency and a lot of wind and wind gusts. I really hope to test this again under much better seeing conditions!
There is still more work to do and maybe gain a little more efficiency.

There are potential bottlenecks still:
Usb3.0 data rate vs camera output
usb cable quality and length
usb port on computer
usb bus inside computer
data rate to drive (nvme faster than sata)
computer cpu and ram

USB cables are the bane of imagers… having spares and testing/swapping them out may have large effects.

Ensure that the usb port you are using for the camera is used by nothing else (ie a hub with a usb2 keyboard or mouse connected as well). Those will slow down the bus rate and the camera will be affected.

There are few planetary images in our Centre… so most of this article may not be of interest to you. Now. It will make a great reference in the future if you do enter into planetary imaging!

This is the camera setup we used last night for the total lunar eclipse.

Skywatcher Star Adventurer on a tripod

Ball joint head

canon T7i DSLR with a 100-300mm zoom lens at 300mm f5.6

a usb heater on the lens, a dew shield on the lens

an intervalometer programmed for a 60 second interval

a plastic “rain coat” for the camera.

a small 5ah lithium battery pack with 5vdc output (meant to recharge smartphones) to power the heater

I had big problems with focusing.. digital zoom all the way in, focused, but on release of the focus ring, it changed/moved. then also it seemed to change by itself when nothing touched the setup. and then because the star adventurer was only loosely polar aligned and not levelled at all, the moon drifted a little during the hours and I released the ball joint to reposition it… and the focus changed again, even though I was careful not to go near it.

The total eclipse image shows the top edge is much lighter and some stars did show as well. even at 2 sec at iso 800 the image was overexposed.

All that said, I think it is still the best total lunar eclipse image I have ever taken!

whens the next one????

🙂

Whilst waiting for the double moon shadow transit on Jupiter Tuesday evening, I did a few runs of Mars imaging.

The wind was brutal.. gusting and then not. It did not get a lot better for Jupiter later on.. in any event…

This was the best 10% of 13K frames of approx 1.5ms each. Seeing was poor, transparency was poor and the Moon was nearing full phase.

There is some blur off to the right of the planet.. almost certainly an artefact of processing the wobbly wind blown image. Mars was less than 10 arcseconds in diameter

Greetings!

Another night that looked promising, what with -7C before 7pm, and skies that looked clear.

One hour, 16 imaging runs of Jupiter and a final temp of -15C later I shut everything down and tried out a new processing method.

*instead* of running all 16 imaging runs (abut 70GB) through autostakkert at 5%, 10%, and 25%, then bringing in each of the 16*3=48 images into Registax and doing wavelets on them, and then bringing in one of the 5,10, or 25% runs into GIMP to do three operations of black point, colour saturation and unsharp mask, and then onto ImageMagick for annotation, I tried the new method. It came to me from one of the so many online video tutorials that I have watched in the last month that I can’t even remember where it came from! The above method can take hours (sometimes a few, sometimes 10), especially running through autostakkert.

This time I took the 16 runs, put them in autostakkert at 10% and turned on the “sharpened” option. This does a much poorer job then Registax, but is batchable and automatic. This took mere minutes. I then looked at the sharpened “draft” results and chose only the best few. Those “chosen ones” went back into autostakkert with sharpened turned off, and the best 5%, 10% and 25% turned on. Then onto Registax, Gimp and Imagemagick as before.

This best result shows that seeing was poor, if not very poor. The raw video shows Jupiter bouncing around left, right, up and down. Firecapture ROI and Cutout Box options can minimize those motions but cannot compensate for the “Z-axis” of in and out movement. The result is a fairly poor contrast, low resolution image of Jupiter.

I do however, love the new, shortened processing method. Instead of results tomorrow, they are available today!