A match made in heaven: DRIZZLING & BXT! Improve your old images EVEN MORE!
HTML-код
- Опубликовано: 29 сен 2024
- Support me on Patreon!: / cuivlazygeek
BlurXterminator: www.rc-astro.c...
Under/Oversampling tool: www.rc-astro.c...
Russ Croman video : www.youtube.co...
Some references:
www.samys.com/...
www.stsci.edu/...
www.pixinsight...
-----------------------
My equipment
-----------------------
Cooled camera: s.click.aliexp...
SW Quattro 150P Budget Newt: bit.ly/3BABCxl
Askar V Modular Telescope: bit.ly/3Imoq2M
or bit.ly/3GLuyR8 or tinyurl.com/2h...
ZWO AM5 Mount: bit.ly/3BAS8x5 or bit.ly/3JeVUkn
iOptron CEM70: bit.ly/431Ii2U (successor of my mount)
MiniGuideScope: bit.ly/3MyUJhn
Celestron C6: bit.ly/3ocePVs
Hyperstar C6: starizona.com/...
Mini Computer (newer version): amzn.to/3LQTgTJ and amzn.to/3K6x2vy
Cable management: www.etsy.com/s...
Camera: aliexpress.com...
IDAS GNB filter: idas.uno/space...
Flocking paper/Felt paper: amzn.to/3zASgvu or amzn.to/3m94zwa
----------------------
Affiliate Links
----------------------
NEW Agena Astro Affiliate Link: bit.ly/3Om0hNG
General OPT Affiliate Link: bit.ly/2OIw6jH
General High Point Scientific affiliate link: bit.ly/3lReu8R
General All-Star Telescope: bit.ly/3SCgVbV
Dwarf 2 Smart Telescope: bit.ly/3SyChXu
General Amazon Link:
amzn.to/3hTB5Ne
Follow me!
My Instagram: / cuivlazygeek
My Facebook page: / cuivlazygeek
Support me on Patreon!: www.patreon.com/cuivlazygeek
BlurXterminator: www.rc-astro.com/product/bxt/
Under/Oversampling tool: www.rc-astro.com/mtf-analyzer/
Hi Cuiv. Very informative post. I have just started to use dithering and have not figured out the best settings yet and have found your comments very helpful. Thank you
Happy to help!
If you use a one shot color camera, square kernel is the only one that works well. Also 2x + 0.9 drop size is optimal as anything lower and BXT won't work properly.
Hey Cuiv! Thanks for sharing my ideas and concepts on this topic! I hope a lot of people get better results with this methodology! Thanks and Clear Skies!
Thanks so much for sharing and making the hobby a better place Dave!
I am waiting for a DrizXterminator.
I'm waiting for the multi-image BlurXterminator , where I can add more than 1 image to be deconvoluted at once. Mono camera guys know what I'm talking about..
Drizzling and dithering, we are in uncharted territory for naming conventions. Feels like quarks; charm, up, down, etc. Thanks for bringing up the subject.
Hahaha indeed! Thanks Guy!
I'm late to the party, but for those of us using ASIair instead of NINA, the options for the Dither Distance setting are 1, 2, 5, 10 or 30 pixels. Do you know if that refers to the guide camera movement or the main camera? Can you recommend which value to select for a modest under-sampled setup (350mm FL with ASI533MC-pro camera, and guide scope 120mm FL with ASI120MM-mini)?
There is one tip I will give about how I fully take advantage of this process. I dither EVERY frame by 20.5 pixels. When I blink my subs, the galaxies look like they are doing hula hoops 😂 but the end result is very low noise, no hot pixels, and excellent drizzling. Just make sure that your mount can handle the wild ride to keep the stars round! If necessary increase your settle time. Yes this eats into the integration time but well worth it! Clear Skies!
The man, the legend! Thank you!
Thank you for the hint! Justa a question: 20.5 pixels are a "one rule to fit them all" or there is some math to calculate an optimal dithering based on focal length?
@@ryther87 the 20.5 works well with sensor QHD or higher regardless of focal length. If you have a smaller sensor you may want to lower it to 10.5 but beware of walking noise when you stack. NINA now shows you the Camera Dither pixels per your Dither Pixels setting. Doesn't have to be exactly X.5, it can be some other number just not X.0 to allow the drizzle algorithm to be more effective. If you don't use NINA you can use this formula: Camera Dither Pixels = Dither Pixels * (Guide Camera Pixel Scale / Main Camera Pixel Scale). You can find your pixel scale using astronomy(.)tools. Hope that helps 🙂
@@deep_space_dave thank you very much for the explanation. When you talk about QHD you talk about resolution, right? I think that the ASI294-Mc Pro should be QHD. I'm using kstars/ekos so the formula your provided will prove useful for sure! Thank you again!
When shooting galaxies with my 60ED, 360mm focal length set-up, drizzling is an absolute must. Reason being that the galaxies are so tiny that a dramatic crop is needed to actually see something. Results are amazing and I would argue, they sometimes rival shots made with far larger scopes. Drizzling FTW!
Absolutely!!
Hey, just got that scope! Nice!
I would suggest also getting a camera that has very small pixels to get the most out of the shorter focal length. A IMX183 based cooled camera is a good start.
I found that using Image Resample just removed the 2x drizzle effect on the stars - back to more blocky like the non drizzeled.
I don't see that! Sure the stars get slightly blockier than before the resample, but still far, far better than a stack without any drizzle!
Perfectly timed video! I gave up trying to process my drizzled images due to Pixinsight taking so long to do things like noise reduction etc. I have recently been trying again after enabling GPU acceleration (I followed the steps in your video) and like the results. One thing I have seen is that BXT can sometimes introduce weird artifacts in the background noise of the drizzled data that ruins the image, although this may be down to operator error when drizzling the data.
Oh wow Pete thank you so much! Interesting on BXT, something to keep track of;
Another good technique to save time is to follow a technique that Cuiv made a video on how to crop those large frames to dramatically speed up processing.
This is now a third support donation from me to your channel, because I am going over your content and I find it a really amazing source of information, especially for a begginer like myself. Keep bringin us more content please, especially for the Seestar S50 community. It is also an opportunity for your channel to grow further, as smart scopes get more mainstream and the fast growing community of beginners needs simple guidance into the hobby. 🙏
I do have a question about drizzle: Based on the requitements that you stated, could this in theory also be useful for processin Seestar S50 files?
And sorry for so many S50 questions, but it is my first EAA scope and I am excited to start playing with it. Already got the PixInsight and the license for the trio of Blur/Star/Noise Xterminators. 👍
Oh wow Andrej, thank you so much for your support!! To answer your question, yes, the Seestar dithers fairly often, and therefore drizzling should be usable! In PixInsight stacking, I'd recommend setting Debayering to bilinear, and drizzle to 2 with square parameter 0.9!
@@CuivTheLazyGeek Thank you for the info, it means a lot! 🙏
I do dithering after every image, but after seeing your video I am wondering now whether the 5px put in my guider (MGEN-II) were enough. I really need to work out NINA.
Great work as always mate!! :-D
Thanks Luke!
My cheap mount dithers for me with its poor tracking. :D
A great explanation of the process and settings. I'm just on the threshold of being undersampled so I drizzle depending on the image. However, I was NOT aware of setting the dither in NINA so that the dither amount is a fractional value on the main camera. I'll have to verify my settings. Again an excellent video.
I'm told that feature is only in the NINA nightlies!! Otherwise you need to compute your relative pixel scales separately
Drizzle works great with the Dwarf as well, especially if you take the image at gain of 100 or more. The read noise on the Dwarf drops massively at the higher gain.
Yes! Although we're quite close to the Dawes limit for the Dwarf!
I have auto-dithering, aka shitty unguided tracking
Great content on a topic that can be confusing (there are a lot in this hobby!)
A little off topic but I see at 11:07 you are on NINA nightlies V3 and the camera dither information on the left side of the screen is new for V3 and is not shown on the released 2.2. You must be comfortable running with V3... Any major glitches/gotcha's you have run into so far that would make you want to roll back to 2.2?
Thanks Cuiv!
Oh I didn't realize it wasn't in NINA v2.22! For me the nightlies are working very well, no problem at all!
@@CuivTheLazyGeek Opps I might have asked that same question above. I will switch to 3.0
"What is the best balance? I don't know?" That seems to be a common concern in this hobby. 😂😂😂
Exactly!
Good video! I’ve got a couple of datasets that are likely candidates for drizzling. Thanks Sir!
Awesome, have fun!
It seems that every time I watch one of your videos I end up firing up NINA to check my settings. Great video Cuiv, I have learned so much from you!
My pleasure :-)
I am in the boat with those, who were not dithering enough times and not enough pixel distance. My mount will have the final word on this, it just hates dithering, it has a tough time recovering after each dither.
Hopefully your mount doesn't protest too much...
as usual...amazing video!
I just recently started drizzling....difference is striking. Now I have to go back a reprocess my images. >
Hahaha have fun!
Great video as usual. Looks like the Camera Ditter Scale and Camera Dither Scale are part of the 3.0 version of NINA? Have you found 3.0 to be relatively stable? If I am not using 3.0 what is the calculation that running to get that result from the Camera sensor size and Ditter Pixels setting?
Hi Cuiv! In your opinion, would drizzling make sense for images coming from a Dwarf2 telescope? Would the natural shifting of the image count as dithering, what do you think?
IntegerResample does not do any interpolation; it only averages (or median, max or min, if you want). In theory it should be better for this type of downsampling (basically binning). But I wonder how different any of the interpolation algorithms used by Resample would be as you're only doing resampling by integers.
Another great video. I was always undersampled with my REDCAT so always did 1X drizzle and got good results, but I will try the 2X to see how it workes
Great video as usual recently I started applying your technique to first run WBPP for calibration only and then running it again on cropped images with integration and normalization to save disk space. Works great. However, at which stage should I run the drizzle? Should I do it on just the calibration phase of WBPP or should I do it in the second phase when I’m doing the integration in normalization?
If you're one shot color, you need to do the cropping on the calibrated Bayered images! So you need to be very careful about the Bayer matrix and make sure that your cropping offsets from the edge are even numbers! The image in the video was actually made with this technique!
Thx i use mono camera 2600mm ZWO. under this scenario I just wanted to know should I drizzle when I do the calibrations prior to the batch crop or should I drizzle when I do the integration of the cropped images?
@@schandler52 drizzle on the second step, with the integration
I have been doing this and you are so right. It is like all the extra pixels from drizzling help BlurX fill in more data in those details.
interesting video from Russel regarding deconvolution AI: ruclips.net/video/5_Dg7bYu-E8/видео.html. Kind regards. Stephane.
Thank you for the link!!
@@CuivTheLazyGeek Avec plaisir ;) Tu as pu regarder a l'adaptateur pour réduire le tirage de caméra stp? (de 17.5 à 12.5mm) Merciiiiiiiis
@@IamArtimon-ui4co il sera bientôt inclus gratuitement avec les caméras achetées - et tu pourras aussi demander aux dealers de l'obtenir séparément si tu as déjà acheté la caméra
@@CuivTheLazyGeek Merciiiis . Bonne semaine à toi. Stéphane.
I enjoyed the video and found it very informative. Thank you for sharing!
Thanks Cuiv,
Great video. How do you get the main camera data up on the guide page of NINA? I have the latest stable version 2.2 but not the nightly builds version. Is it only on the nightly build version?
Thanks
Sorry it seems indeed it's only in the nightlies!
That downsample-trick is great
Great video Cuiv as always, I don't think it will help me with the current rig : 1000mm 200 pds with a 2.5x barlow and an asi 294 mcpro one shot? :) what do you think?
Yeah that's not going to help! Also a Barlow?? That's a brave step :p
To my understanding a drizzle of 1x1 (sounds strange) is beneficial to overcome a bit of the drawbackof debayering algoritms. When pixel peeping you can definitely tell the difference between no drizzle and a 1x1. Btw, I dither every 5 minutes normally, but I have to redo the amount with which I dither.
You're absolutely right! Bilinear debayer + 1x1 drizzle with drop 1 helps apparently to some extent
Good video again Cuiv 👍. A couple of things: the primary mask doesn't just hide the primary clips, it also hides the rough edge of the mirror, reducing light scatter- I think it more than makes up for the lost mirror area. Also, my scope was shipped in March in the UK and has the better focuser. But I definitely agree that SW should include this stuff as standard.
Thanks!
That is just amazing. Thanks Cuiv!
Just tried drizzle. 1. Took a long time in PI. 2. Really improved the image. ❤
That's it in a nutshell :-)
Great video, Cuiv! Thanks. Would there be any advantage to masking your target(s) when using resampling to reduce background noise?
I don't think resampling can work with masking - it's an all or nothing operation!
@@CuivTheLazyGeek Of course! I should have realized that couldn't be applied to just a part of the image. Duh.
Dither or die! Wait.. that's drizzle or Die! No.. not quite right.. Dither, drizzle and bxt or die!
Hahaha exactly!
Hi Cuiv,
Personally I apply successively R. Croman's BlurExterminator then NoisExterminaor with excellent results. My image scale is one arcsec/pix.
That works, and typically at 1"/pixel you don't need to drizzle - but you could try
@@CuivTheLazyGeek yes I shall. Thank you for the advice.
can this be done with D2 images and processed in Siril or other free software?
Not sure about drizzling in Siril - but in principle yes it should work for D2 (although somewhat less well due to Dawes limit)
Thanks for this !!
Hey Cuiv I completed my first drizzle yesterday and the image came out really well however I noticed you set it to two times drizzle mine is only set to one the default. Have I missed a trick here should I redo again at 2x? Thanks
Should try at 2x!
@@CuivTheLazyGeek 🤦♂️ need to free up some space and plug this laptop in then 😂
According to the code of this new feature, the formula to get value at 12:22 is "(GuiderPixelScale*GuiderDitherPixel)/MainCameraPixelScale"
Using your data and assuming FL=300mm and px=3.76um, this should equal to (4.98*7)/2.5865=13.477 and not 23.26.
I wonder where that number comes from then...
The example here was with my Newt!
@@CuivTheLazyGeek thanks Cuiv :) I'm trying to find the math behind it, so I can apply it to my case. But the data doesn't match what NINA shows on your screen.
MainCameraPixelScale = (3.76 / 600) * 206.2648 = 1.2926
(4.98*7)/1.2926 = 26.9689
what am I missing?
@@PaoloStivanin f3.45 :-) So FL is around 517mm
Cuiv,
every time is a journey to listining yuor comment. Thank your
Thank you Giovanni, this means a lot!!
Questions are do you need drizzling for normal (non-undersampled) images? Do you need it for mono cameras? I am not sure whether you do.
If you are properly sampled, or oversampled, drizzling won't do anything for you!
If you have a monochrome camera and you are undersampled, drizzling will still help!
@@CuivTheLazyGeekThanks. I didn’t know about mono cameras. I thought it never works. I dither all the time. So called, dithering without guiding.
I don't see dither as significantly affecting imaging time. Really doesn't take all that much time to settle the mount
When you take 30s to 60s exposures, dither + like 10s of settling is a looong time :)
@@CuivTheLazyGeek might be because I use the ec version of the mount, I did 120 frames of 60 second exposures last night and dithered every single one
Hmmm, where are the link to the other videos ?
Added to the description, apologies about that!
@@CuivTheLazyGeek Merci Cuiv !!
ممتاز
Thank you!
Blur Exterminartor is NOT a deconvolution algorithm! It is A.I. image reconstruction It is nolonger connected to the reality (truth) of your photograph..
That's incorrect - I would suggest watching Adam Block's interview with Russ!
Great video! Fully agree! Just processed the Hercules Cluster, shot with my FRA 400 and I was amazed how much detail I still gout out of this tiny object thanks to drizzling and BXT!
Right?? It's absolutely amazing!
Thanks Cuiv!
Wouldn't too much dithering would affect your total guiding error for the worse?
My pleasure! No dithering is typically excluded from RMS figures, and of course the next exposure starts only once the dithering is complete and guiding has settled. So it shouldn't affect either figure!
The video I actually just needed. THX
Glad I can help!
Je suis en plein apprentissage avec BlurXterminator et NozeExterminater. Assez impressionnant
Ce sont vraiment des outils incroyables !
The other time my stars had a dark centre after drizzling. How can I avoid this?
That I don't know, I haven't seen that...