20 replies to this topic

[roguebantha]

I've been post-processing images with a Vaonis Vespera Pro and I've been noticing these inconsistent artifacts showing up after the stacking and background extraction process. They appear as bands, wide striations, etc. and while sometimes I can compensate during stretching, other times the image appears worthlessly unprocessable. The effect seems to vary depending on stacking software, background extraction method, smoothing parameters etc. but not fully purgeable if it exists no matter what I use, although sometimes it simply won't show up at all.

Sometimes the effect is reasonably subtle especially after post-processing and/or gathering additional data such as (see the background darkness fluctuations especially in the bottom right):

https://drive.google...iew?usp=sharing

 

Other times it's noticeable and distracting but somewhat tolerable such as:

https://drive.google...iew?usp=sharing

 

And other times it's so dramatic the image appears essentially unusable such as:

https://drive.google...iew?usp=sharing

 

I'm having difficulty tracking down what could be causing this effect. Is it because I'm not using flats and bias frames and only using lights and darks? Is it because I'm adding low-quality images to my stack? Is it some artifact of the stacking method? Why does it become so noticeable after background extraction? Are these perhaps different effects that have different causes?

Any help root-causing the issue would be very appreciated!

[gumbajoe]

I don't have any experience with the Vespera automated telescope - their web page says that you can take dark frames:

 

"...using the "Expert Mode'' within the Singularity app. This mode comes with three distinct features: control of Vespera's camera settings, the capture of dark calibration frames, and the automatic subtraction of these dark frames for reduction of thermal noise."

 

but it does not discuss flat or bias/ flat dark / dark flat frames.

 

In my experience the "band-like" artifacts that you are seeing is usually not associated with problems with the light frames alone (i.e a sensor problem): this does not look like Canon banding artifact or something similar.

 

 

In general, for astrophotography you should have the following frames:

1. Light
2. Dark
3. Flat
4. Either Bias or [Dark Flat aka Flat Dark aka Dark frames which match the exposure time of the flat frames]

I will leave it to the experts as to which sensors do not need certain calibration frames. IMO as a beginner it is a good idea to use #1-4 above to understand the process of calibration. Flats can help with gradients related to optics.

 

What program are you using for post-processing?

How are you applying gradient correction?

 

Pixinsight Dynamic Background subtraction (DBE) without ideal results will emphasize any residual gradients in your picture: it is better to try to fix them rather than just hiding them by lowering the background brightness. How good a job DBE does at removing gradients is influenced a lot by the points you select on the image (are you selecting only background?). You can always place points, run the process, and review the results: if you still see residual gradients you can place new points and try again. Automated Background Extraction is an alternative.

 

The new GradientCorrection process in PixInsight does a good job with complex gradients without having to place any points, but there is a learning curve. PixInsight has a series of educational videos you can watch.

Edited by gumbajoe, 05 September 2024 - 06:42 PM.

[bobzeq25]

In general, for astrophotography you should have the following frames:

• 
  
• Light
• Dark
• Flat
• Either Bias or [Dark Flat aka Flat Dark aka Dark frames which match the exposure time of the flat frames]

I will leave it to the experts as to which sensors do not need certain calibration frames. IMO as a beginner it is a good idea to use #1-4 above to understand the process of calibration. Flats can help with gradients related to optics.

Some general (and very good) rules.

Every setup can benefit from flats and bias. For a VERY few cameras (I don't know of any besides the older 1600s and 294s) bias do not work well, and you need dark flats as a substitute. But in any event, you should only take one of these, taking both adds NOTHING of any value, and is a frequent source of problems.

Darks are tricky. Some cameras with low thermal noise can get away with omitting them. Whether you fall into that category is best determined by experiment. It may change between broadband (short exposure times) and narrowband (long exposure times). It's somewhat a matter of how precise you want to get, how demanding you are.

Edited by bobzeq25, 05 September 2024 - 06:56 PM.

[roguebantha]

Yeah I'm aware that capturing some bias and flat frames would be useful, although it seems to me (and based on other people's results with the vespera and seestar) that I should be able to create a decent stack without those sorts of frames. My guess is that I have some sort of issue related to frame alignment. Here's an example uncropped mosaic that is perhaps the most illuminating image yet in terms of the output I'm getting:

https://drive.google...iew?usp=sharing

[gumbajoe]

The link you are posting is pretty big at 358 MB, I believe I am seeing the artifact you are describing from images on your initial post.

 

Perhaps you could provide a link to an unprocessed single frame and perhaps to an unprocessed stacked image of reasonable file size - it would help clarify if this is related to the image itself, or to the image post processing.

 

What are you using for image processing, and specifically what steps are you using for image processing? I am only familiar with PixInsight, other forum members can provide insight into other software.

[roguebantha]

Sorry it's so big - I could have cropped away the outer edges but I thought it'd be useful to include the whole thing so it's intuitive how it appears the stacking process is leading to the observed artifacts. Here's a 1/3 the size link that I've cropped down.

https://drive.google...iew?usp=sharing

 

I'm primarily using DeepSkyStacker for processing, although I've noticed the effect in other stacking software including Siril and Affinity photo.

 

I've been using GraxPert for background extraction.

[idclimber]

Why are you doing a mosaic?

 

You need to be integrating your data, which means one master alignment frame. You also need the other frames aligned nearly the same. Not all over the place like it seems to be in your large file. If you do this you can improve SNR. 

 

The way people get decent images with these type of scopes is they point at relatives bright targets. Dark skies also help. The smart ones also find targets in the west or east relatively close the horizon where tracking is simplest and rotation is minimized. 

[roguebantha]

Ah, I've been using mosaic mode so I can crop manually afterwards. Perhaps this is part of my problem but I find that if I just let DSS choose the reference frame based on score, my image tends to end up substantially offset from the target itself. I was under the impression that it was fine to choose mosaic and then crop later - does mosaic introduce greater potential for artifacting, even towards the center of the image?

[roguebantha]

Here's the "unusable" image from above but without denoising and a wider FOV. It seems relatively clearly to be a rotation related artifact coming out in the stacking process. I normally thought field rotation was usually a bit more of a gradual artifact but perhaps not? Maybe it's because of my stacking technique "Kagma-Sigma clipping"?

 

I've also tried restacking that image I posted with the awful looking artifact above, but this time I chose a different reference frame and got a MUCH cleaner stack, with a comparatively beautiful field rotation pattern towards the outside of the center frame. So maybe a lot of this is down to reference frame choice.

 

 

EDIT: Although deeper into processing and stretching it's clear the artifacts are still there if more subtle

Edited by roguebantha, 05 September 2024 - 11:49 PM.

[sternenhimmel]

 

Here's the "unusable" image from above but without denoising and a wider FOV. It seems relatively clearly to be a rotation related artifact coming out in the stacking process. I normally thought field rotation was usually a bit more of a gradual artifact but perhaps not? Maybe it's because of my stacking technique "Kagma-Sigma clipping"?

 

There shouldn't be any field rotation unless he's using a dob, and even then, they should all be registered and stacked accordingly. 

 

[roguebantha]

Why are you doing a mosaic?

 

You need to be integrating your data, which means one master alignment frame. You also need the other frames aligned nearly the same. Not all over the place like it seems to be in your large file. If you do this you can improve SNR. 

 

The way people get decent images with these type of scopes is they point at relatives bright targets. Dark skies also help. The smart ones also find targets in the west or east relatively close the horizon where tracking is simplest and rotation is minimized. 

And now that you're saying this, I wonder if a serious part of my issue is down to the targets I'm choosing. Looking at the three targets I'd chosen above in Stellarium, I can see that the effect is most dramatic on targets closer to the north where rotation is strongest. This could explain why the effect seems to vary so significantly...perhaps it's directly correlated to how much the object I've chosen rotates on any given night.

[idclimber]

And now that you're saying this, I wonder if a serious part of my issue is down to the targets I'm choosing. Looking at the three targets I'd chosen above in Stellarium, I can see that the effect is most dramatic on targets closer to the north where rotation is strongest. This could explain why the effect seems to vary so significantly...perhaps it's directly correlated to how much the object I've chosen rotates on any given night.

My understanding of that system is it had a rotator. If so it is not working. 

 

If it does not, it is subject to rotation, which is highly dependent on where in the sky you are pointing. Latitude also is in this math. 

[roguebantha]

I don't think it has a rotator. I think it's just a simple Goto Alt-Azimuth mount.

[sternenhimmel]

Ah I missed the part about it being an Alt-Az mount -- then indeed there will be field rotation, and because your sensor is rectangular, the only part of the field that is being consistently sampled is a circular region at the center, the edges are going to have overlap of different frames. 

[bobsdobs]

You'd be better asking on the smart telescopes forum, but I think the problem is you're using CovalENS mode to do mosaics and then trying to process the subs individually? If that's the case I don't think it's going to work. The scope does some processing to progressively fill out the mosaic in a spiral type pattern and at the same time remove the vignetting etc that you're seeing between different subs. It's not like a mosaic plan in something like Nina where you have nice tiles with a predefined overlap and easy to stitch together. CovalENS doesn't work like that. I'm not sure stacking the subs from a vespera mosaic outside the scope is going to work that well. I think it's pretty much designed to give you a somewhat preprocessed master flat where the vignetting and weird artifacts introduced in the spiral are cleaned up (although I think from what I've seen aggressive stretching will still show some patterns like what you have above but much reduced).

[roguebantha]

Ah I missed the part about it being an Alt-Az mount -- then indeed there will be field rotation, and because your sensor is rectangular, the only part of the field that is being consistently sampled is a circular region at the center, the edges are going to have overlap of different frames. 

Yeah I did expect to get field rotation of some sort - I think what I'm least sure of is why I'm ever getting field rotation like artifacts in the *middle* of my final stack. I would have expected it to have a fairly regular circular gradient of field rotation artifacting as I increase offset from the target but that does not appear to be the case.

 

You'd be better asking on the smart telescopes forum, but I think the problem is you're using CovalENS mode to do mosaics and then trying to process the subs individually? If that's the case I don't think it's going to work. The scope does some processing to progressively fill out the mosaic in a spiral type pattern and at the same time remove the vignetting etc that you're seeing between different subs. It's not like a mosaic plan in something like Nina where you have nice tiles with a predefined overlap and easy to stitch together. CovalENS doesn't work like that. I'm not sure stacking the subs from a vespera mosaic outside the scope is going to work that well. I think it's pretty much designed to give you a somewhat preprocessed master flat where the vignetting and weird artifacts introduced in the spiral are cleaned up (although I think from what I've seen aggressive stretching will still show some patterns like what you have above but much reduced).

I'm not using CovalENS mode in this case - just a normal capture and integration. That being said, clearly the tracking leaves something to be desired because I'm finding scrolling through the fits files a lot of frames where the desired object isn't remotely centered, and fairly often not in the frame at all.

Edited by roguebantha, 06 September 2024 - 10:29 PM.

[bobsdobs]

Ah ok, my mistake, sorry. I thought you said you were using mosaic mode.

[idclimber]

Yeah I did expect to get field rotation of some sort - I think what I'm least sure of is why I'm ever getting field rotation like artifacts in the *middle* of my final stack. I would have expected it to have a fairly regular circular gradient of field rotation artifacting as I increase offset from the target but that does not appear to be the case.

 

I'm not using CovalENS mode in this case - just a normal capture and integration. That being said, clearly the tracking leaves something to be desired because I'm finding scrolling through the fits files a lot of frames where the desired object isn't remotely centered, and fairly often not in the frame at all.

An Alt Az mount is tracking the sky with two motors. It does this using an internal model of the sky developed by a three point alignment or some other sampling of the sky. There are always errors in the sampling and the tracking is not perfect. 

 

When I used one of these type of mounts how accurate my model was could be seen in how long a target would remain in the center of the field. Even when it was above average targets would slowly drift off the center in 20 or 30 minutes. 

 

You can see this drift in your integration. 

 

There is also a second motion, caused by the rotation of the sky. As I stated earlier how much depends on where in the sky you are pointed, as well as a few other variables. I suggest you visit some of the web sites devoted to explaining this to better understand the issue. If you are due north or south the rotation will quickly approach 60 degrees per hour. That is one degree in a minute. If you are doing standard 60" exposures in this region of the sky each sub exposure will be rotated one degreee from the next. 

 

Here is one website:

https://calgary.rasc...ld_rotation.htm

 

What you see in your integration is both errors. A equatorial mount has the exact same issue if it is not perfectly polar aligned. It is simply minimized to a great extent. We also implement guiding or additional sky modeling to correct this error. 

 

An Alt Az mount can be setup with guiding and then the center of rotation will be exactly at the point of the guide star. 

 

Without guiding the center of rotation is somewhat random based on the imperfect sky model and pointing. It is likely not in the center of the frame, at least not precisely. 

[roguebantha]

Alright so from all this I'm trying to employ the following mitigations:

Capture objects more to the East and West rather than objects in the North where field rotation is the worst.

Avoid stacking frames where the object is significantly offset or not visible to minimize stacking artifacts.

Use clone stamping to try and minimize whatever those two measures miss.

 

But I suppose other than that, there's perhaps no secret thing I'm doing wrong that's leading to those artifacts most likely?

 

I achieved a fairly good outcome on that Andromeda data from before using those three measures. Clone stamping was probably the most impactful and is fairly easy to do after using Starnet.

[bobzeq25]

You'd be better asking on the smart telescopes forum.

THIS. And it's not a close call. Not REMOTELY a close call.

[roguebantha]

I think I've figured out the issue. It has to do with a drift over time away from the target itself, which I'm quite certain is tracking/guiding/whatever you call it issue rather than an issue with my stacking process. Time to talk to Vaonis!