Wipe results look hideous

Questions and answers about processing in StarTools and how to accomplish certain tasks.
Post Reply
Orion_Rockstar
Posts: 3
Joined: Fri Feb 07, 2020 4:05 am

Wipe results look hideous

Post by Orion_Rockstar »

I've been working with starstools the past few weeks and I am able to get pretty decent results. But I do have concerns on my data as shown through the wipe function and addtion to the autodev. When I run autodev the 1st time, I usually get a sea of red pixels with the object of interest and bright stars being white. Is this normal? I also noticed when I run my m81/82 images through autodev, the red is replaced with pink. Is the moonlight to blame for this autodev color change ( clear nights have been a rarity as of late so I had expose under a half moon in the west sky)?
Onto wipe. When I run wipe, I get wonderful mess of colorful pixels. This seems abnormal as I think I am taking good calibration frames which should help wipe perform better. What could be causing this mess? Or is this normal? It seems as though knocking up the da filter to 5 pixels seem to make the colorful pixels be more evenly distributed. Any insight and help would be helpful.
Canon 77d
Iso at 400 and 800 (testing which one is ideal)
5min subs (with Orion being at 3 min)
Total time: Leo@2h Orion@45m 8182@1h
Attachments
iso400,5m*24,da=5
iso400,5m*24,da=5
leowipeda5.jpg (444.1 KiB) Viewed 5120 times
iso400,5m*24,da=1
iso400,5m*24,da=1
leowipeda1.jpg (451.29 KiB) Viewed 5120 times
iso400,5m*24
iso400,5m*24
autodevleo.jpg (452.81 KiB) Viewed 5120 times
Orion_Rockstar
Posts: 3
Joined: Fri Feb 07, 2020 4:05 am

Re: Wipe results look hideous

Post by Orion_Rockstar »

Extra images.
Attachments
orionsnip-iso400.jpg
orionsnip-iso400.jpg (282.8 KiB) Viewed 5119 times
8182iso800wipe.jpg
8182iso800wipe.jpg (386.06 KiB) Viewed 5119 times
8182auto-iso800.jpg
8182auto-iso800.jpg (462.41 KiB) Viewed 5119 times
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Wipe results look hideous

Post by admin »

Hi,

Would you be able to upload one of the datasets and share a link to it (Google Drive, Dropbox, One Drive, etc)?
Would you be able to provide some more details on how the dataset(s) were stacked? Were they straight CR2s you imported into your stacker? Which stacker? What settings did you use?
Though I cannot say for sure looking at the JPEG screenshots, I seem to be able to spot rather sever mottling that is not just the result of shot noise.
Regardless, I can spot some classic signs of dark anomalies (e.g. colored halos) in your dataset. You will find the dark anomaly at the center of each halo.
Try bumping up the Dark Anomaly filter and/or mask these anomalies out.

Hope this helps in the meantime,
Ivo Jager
StarTools creator and astronomy enthusiast
Orion_Rockstar
Posts: 3
Joined: Fri Feb 07, 2020 4:05 am

Re: Wipe results look hideous

Post by Orion_Rockstar »

Here is the link to somefts files.
https://drive.google.com/open?id=11XfDF ... UZ_FQcg0qt
I use the the autosave file produced by deep sky stacker. And I use their recommened settings. And yes, I use my camera's raw files when I stack.
I do bump up the dark anomaly filter to about 5 pixels. An example can be seen on what what happens in the pics above in the leo galaxies. It goes from large blotches to a more uniform multi-colored pixel set.
almcl
Posts: 263
Joined: Wed Jan 21, 2015 7:15 pm
Location: Shropshire. UK

Re: Wipe results look hideous

Post by almcl »

While we wait for Ivo's better qualified and much more expert opinion, I had a quick play with your data and think it looks very nice. Orion result below. I don't think you should worry too much about how the image looks after the first autodev or wipe. There are quite a few steps to go before the image will look 'normal' and the initial stages will help to identify what needs to be done subsequently.

Here's what I got (heavily cropped and binned to speed up processing):
orion.jpg
orion.jpg (492.97 KiB) Viewed 5062 times
Skywatcher 190MN, ASI 2600 or astro modded Canon 700d, guided by OAG, ASI120, PHD2
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: Wipe results look hideous

Post by happy-kat »

For my data I always get a similar none usable effect in Wipe but I always do a manual dev as my next step and don't see what auto dev does.
Edit: Lovely colour spread in that process above
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Wipe results look hideous

Post by admin »

Orion_Rockstar wrote:I do bump up the dark anomaly filter to about 5 pixels. An example can be seen on what what happens in the pics above in the leo galaxies. It goes from large blotches to a more uniform multi-colored pixel set.
Thank you for uploading. I have to admit, I'm not seeing anything problematic here....
Perhaps the problem lies in a misunderstanding what the Wipe module (and the in-built AutoDev) does and how the processing engine in StarTools works?

In general, you may be interested in the tutorials and videos here. In particular, you may find this older (but still relevant) video on an M42 dataset very similar to yours enlightening.

Wipe operates on the linear data. It uses an exaggerated AutoDev stretch of the linear data to help you visualise any remaining issues. In a way, it is meant to look "hideous", as it is used here as a diagnostics tool to show all warts in your data. After running Wipe, you will need to re-stretch your dataset. That is because the previous stretch is pretty much guaranteed to be no longer be valid/desirable, as gradients have been removed and now no longer take up precious dynamic range. Dynamic range can now be allocated much more effectively to show detail, instead of artifacts and gradients. As a matter of fact, as of version 1.5, you are forced to re-stretch your image; when you close Wipe in 1.5+, it will revert back to the wiped, linear state, ready for re-stretch.
StarTools_125.jpg
StarTools_125.jpg (336.58 KiB) Viewed 5058 times
AutoDev serves two purposes; it serves as a diagnostics tool and it serves as a highly sophisticated global stretching tool. AutoDev is usually the first thing you'd do to inspect the data you're working with. It will bring out any issues that need your attention. Once you have addressed the issue that AutoDev brought out (or, at the very least are now aware of these issues), re-running AutoDev is usually the first thing you do after running Wipe.
StarTools_126.jpg
StarTools_126.jpg (186.98 KiB) Viewed 5058 times
AutoDev should already yield much better results now that you have fixed (or at least mitigated) the issues it brought to your attention first (for example, you may wish to seriously consider binning your datasets before proceeding!). If your dataset is exceptionally noisy, you may wish to make AutoDev 'blind' to fine noise grain, so that it no longer brings it out (use the 'Ignore Fine Detail <' parameter). As of 1.5+ AutoDev will re-use the Dark Anomaly setting for this value, so, chanches are you don't even have to change it.
One other important thing to do if your image contains one or multiple objects on an 'empty' background (your images do!) is to specify (click & drag) a Region of Interest for AutoDev to optimize for. This is exactly what it sounds like; a region that is a good sample of the interesting detail in your image. Without an RoI AutoDev will assume that everything in your image is equally as 'interesting' and should be brought out. However that is definitely not the case if you have lots of 'uninteresting' background in your image.
happy-kat wrote: I always do a manual dev as my next step and don't see what auto dev does.
Please note that a manual Develop is a fairly crude tool that was specifically designed to emulate photographic film. For example, it blows out (or 'enhances', depending on whether your like the effect!) stellar profiles in the exact same way film does, while dynamic range allocation is precisely as 'dumb' (or pretty! :) ) as photographic film.
AutoDev on the other hand will solve for a stretch that shows all detail in the RoI as well as possible and should/will yield tight stars.

With all the above in mind, it should be straightforward to quickly progress to results like these;
orion(1).jpg
orion(1).jpg (358.95 KiB) Viewed 5058 times
Autosave001(5).jpg
Autosave001(5).jpg (119.5 KiB) Viewed 5058 times
(only Decon, and Color applied before switching Tracking off)

As a side note, try the latest DSS version that allows white-balancing to be turned off. You can then import the dataset using the second option (when you use 'open'). This should yield a small but useful signal boost and may also help a litle with color artifacts in the highlights. (note that this separates luminance and color and you will be processing the luminance, which is of course mono, until you hit the Color module).

You may also wish to change your stacking method for the bias/flats/darks from median to something else (e.g. 'average' of some other outlier rejection algorithm). For lower amounts of light frames, median will yield good results, though if you shoot more you may also wish to try other outlier algorithms (they typically require more frames to be successful).

I hope this helps. Any further trouble, questions, do let us know!
Ivo Jager
StarTools creator and astronomy enthusiast
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: Wipe results look hideous

Post by happy-kat »

I would love to see the log for that process of m42 please Ivo.I really like your star colours/shape and background.
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Wipe results look hideous

Post by admin »

happy-kat wrote:I would love to see the log for that process of m42 please Ivo.I really like your star colours/shape and background.
I lost the log (was using a development version), but it really was a "quick and dirty" one, so I reconstructed it. The stellar profiles are kept in check perfectly by using AutoDev for the final stretch (a Region of Interest over M42 was used to give it a good sample of what's "interesting").

---
Type of Data: Linear, was not Bayered, or was Bayered + white balanced
--- Auto Develop
To see what we got.
--- Bin
Parameter [Scale] set to [(scale/noise reduction 50.00%)/(400.00%)/(+2.00 bits)]
Convert oversampling into noise reduciton
--- Crop
Framing.
Parameter [X1] set to [279 pixels]
Parameter [Y1] set to [26 pixels]
Parameter [X2] set to [2769 pixels (-242)]
Parameter [Y2] set to [1893 pixels (-117)]
Image size is 2490 x 1867
--- Wipe
Default settings, except Parameter [Dark Anomaly Filter] set to [3 pixels]
--- Auto Develop
Again, the secret here is to use 'Ignore Fine Detail <' if your background is a little noisy and specifying a RoI so that AutoDev optimises the stretch for interesting stuff, and not so much "empty" background.
Parameter [Ignore Fine Detail <] set to [2.1 pixels]
Parameter [RoI X1] set to [960 pixels]
Parameter [RoI Y1] set to [515 pixels]
Parameter [RoI X2] set to [1988 pixels (-502)]
Parameter [RoI Y2] set to [1538 pixels (-329)]
--- Deconvolution
Specific to the development version I'm running right now, but similar results should be easily achievable in any version of StarTools.
Parameter [Primary PSF] set to [Moffat Beta=4.765 (Trujillo)]
Parameter [Tracking Propagation] set to [Post-decon (Fast)]
Parameter [Atmospheric Radius] set to [2.2 pixels]
Parameter [Iterations] set to [28]
Parameter [Mask Fuzz] set to [4.0 pixels]
--- Color
Default.
--- Wavelet De-Noise
Default.

I walked into an op-shop (thrift store if you're from the US) the other day and spotted this wonderful framed photograph of the Horse Head and Flame Nebula. However, it could tell it was indeed acquired using photographic film, as the stellar profile of Alnitak took up a good quarter of the image, while blacks were much deeper, not showing much faint nebulosity. If you are using the Digital Development slider in Develop in StarTools, this is exactly the look it was meant to emulate - for good or for worse.

Hope this helps!
Ivo Jager
StarTools creator and astronomy enthusiast
Post Reply