Be like Google! Dither your astropics!

Dithering is a method that has long been used in astrophotography to get both better signal to noise ratio, but also the ability to drizzle, e.g. actually increase the resolution based on information provided by the dithering. And now, Google uses dithering in its own Pixel smartphones! In this video I go through how dither works, its effects in real life, and how to actually use it!
#astrophotography

[@Fenriswaffle]

Every video I watch proves to me how little I know about astrophotography. I knew *of* the drizzle option in many stacking softwares but couldn't understand the purpose. Now it makes so much sense that I wish I knew about it 6 months ago. Great stuff!

[@davidaylsworth8964]

Now I get it! I had been considering what was causing the banding in my stacked frames until your explanation of how dithering works. My next session will incorporate dithering. Now that I’ve mastered focus and a stable polar alignment this would seem to be the next step to getting better data. Thanks for the clear explanation.

[@AnakkiDan]

Man, you're an amazing person. Not only are the videos very informative but i also REALLY enjoy the way you give that info, very friendly and calm and happy, which is also contagious. 13:33 brought a big smile to my face and i wish u nothing but the best Also i dither every shot, even though i shoot 1s exposures. Not because of anything but i don't have a tracker yet so technically all my shots get dithered, i do tilt the camera as well ever since i only shot a small amount of pics and found out walking noise exists

[@CuivTheLazyGeek]

Thanks for that feedback Mighty Droid!

[@przemekmajewski1]

I usually like your videos, but the examples of hot pixels and blocky stars, however theoretically easy to understand, have melted my heart :D this is an outstandingly great video!

[@tchallalemou5307]

your'e a monster bruh. good on ya for putting out all this astro content.

[@CuivTheLazyGeek]

Thank you :) The confinement is helping with putting up all that content - it will likely slow down a bit next month, once confinement relaxes here.

[@ak13three]

Thanks for explaining the concept in simple English. Appreciate all your videos, very well laid out and easy to follow. Thanks again!

[@CrescentUmmi]

You area great astrophotography channel, glad I found you! No one ever explained dithering like this on their channels. Keep it up!

[@CuivTheLazyGeek]

Thank you so much for this comment! I love explaining stuff, and apparently I'm good at it (?) so I'm so glad it is being helpful! I will definitely keep it up.

[@truefreedom3662]

Very informative as always. Many thanks for your contribution 🎉

[@CuivTheLazyGeek]

Always welcome

[@luisrivera-lopez3648]

I really love your videos because you offer valuable insights that I really need to know to enhance my skills. I started astrophotography back in November 2020 and still I am learning. Thank you!

[@CuivTheLazyGeek]

Wonderful!

[@dezm0n679]

Wait, you implemented Direct Guider into NINA? You are crazy, not lazy :) Thank you very much for that feature and videos

[@CuivTheLazyGeek]

Yep! My very first NINA contribution - I also contributed a lot to the autofocus routine and to properly supporting OSC cameras :)

[@dezm0n679]

@CuivTheLazyGeek Wow that explains your extensive NINA autofocus knowledge, can you tell me then: I own very minimalistic setup right now (mount and Nikon dslr with lenses), so I found out then I can dither without guidescope, that's nice. But can I also autofocus without focuser, just by analyzing few images and varying focus? (I can control focus in other app, I tried, so it should be possible) I saw plugin for Canon, but nothing for Nikon 😞

[@dezm0n679]

Also, you know, I'm something of a programmer myself. So if not currently possible, I can try hack something myself, but don't know where to start with NINA and Nikon, never touched anything like that before 😀

[@PolanskiJoe]

Excellent video! now I understand the drizzle function and why it wasnt working for me in the past! Thank You Cuiv!

[@rudycajuso29]

Great video. I have been using dithering but this video helps to make more sense of it.

[@CuivTheLazyGeek]

Excellent to hear - dithering is one of the best weapons in our arsenal, but I've seen too many people (including my past self) skip on it entirely.

[@ippiu]

Avoid dithering by doing several bad pixel maps with each set of darks. APP does this very well. I'm using APP and its BPMs. But now i want to start dithering and see if there is a more clean background effects compared to BPM-APP method. Thanks for your wonderful videos

[@CuivTheLazyGeek]

That's a good point - in PixInsight, something equivalent can be done, but it's more complicated. Or it is about using Cosmetic Correction on the calibrated frames prior to stacking: www.lightvortexastronomy.com/tutorial-pre-processing-calibrating-and-stacking-images-in-pixinsight.html . Dithering does provide an advantage in any remaining fixed pattern noise, and the ability to drizzle (or drizzle and then resample, which should increase SNR). The more I hear about APP, the more I want to try it!

[@theartoframos]

Dude, Cuiv love your content! Thanks for making NINA easiER to understand AND Calibration Frames. I understand you better than the other guys, buuut... Please can you set your camera's focus to manual please. It keeps jumping around because of your hand movements and its very unnerving I have to look away. Don't mean any disrespect just happened in another as well, kinda hurt my eyes. That's it, everything else is AWESOME! Thanks again for making things easy for me! Didn't know I had to learn so much just to be lazy! LOL

[@michal_king478]

Thanks for the tutorial. My images have always had lines of noise in one direction and I never knew why

[@dankahraman354]

I suspected what walking noise was not you have confirmed it visually. Thanks.

[@felipemacauliffe1528]

I feel even more convinced about keeping on dithering thanks to your great lecture. Thanks for your great work ... it's truly useful to me. All the best, clear skies and shine on from the "Cluster One Observatory"

[@CuivTheLazyGeek]

Glad that is is being useful! Dithering really is important... Clear skies!

[@claudioarchi]

Excellent video Cuiv. Very useful and well explained. Congrats, I really appreciate the info you share with the community. Suggestion on reviews: The mini pc you are using, maybe compare it to a compu stick (I'm on the market for that)

[@CuivTheLazyGeek]

Interesting one - I don't have a compustick, but I could do a video on the imaging PC, its specs, and what are the demands of imaging setups overall, especially with NINA.

[@claudioarchi]

@@CuivTheLazyGeek that would be great.Thanks

[@mdaugher1811]

Great presentation! Thank you for sharing.

[@CuivTheLazyGeek]

Thank you for watching! Glad it's being useful

[@blueastrophotography6131]

I love all your content. Congrats on this video it was really helpful

[@ant_hart]

Gutted I bought an iPhone now😓 Great explanation mate 👍🏼

[@CuivTheLazyGeek]

Hehehe your iPhone will do just fine - I actually wonder, maybe Apple does the same type of processing, but doesn't advertise it as Google does...

[@ant_hart]

Cuiv, Lazy Geek in Japan Compared to the Pixel my iPhone 11 on night mode isn’t a patch on it. You can only shoot for a max of 10 seconds too and I think with the pixel you can do a few minutes?

[@NeilW210]

So well explained Cuiv...I understand it much better now. On an off-topic note, where can I see what's been changed in each nightly build? The "Changelog" link in the Betas section of the NINA website seems to be the full 1.10 version log. Tnanks.

[@CuivTheLazyGeek]

Glad that it is clearer! Dithering is really an amazing tool that really works great. Unfortunately, for the beta to beta or nightly to nightly releases, you need to use this to see which actual commits and pull requests have been done, so a bit cumbersome...: bitbucket.org/Isbeorn/nina/commits/

[@NeilW210]

Mmm...that's a little technical for me. Guess I'll just have to rely on you to let us know any neat new features that are added :-)

[@Quoutub]

2:55 Google's zoom only really works in low light conditions (or good light conditions with a low signal to noise ratio) because it improves the signal to noise ratio. In perfect light conditions it's not significantly better (regarding details) than an upscaled crop of a perfectly exposed single frame dng file. Occasionally it is even worse than a crop of a dng file. Google also uses machine learning for the zoom which leads to artificial results (as machine learning only guesses) and even makes dead pixels visible. Sometimes it even transforms dead pixels into white curvy lines. I tried multiple Pixel 3a and they all have this issue, but with different severity. Surprisingly, the white curvy lines and dead pixels can appear when you zoom or when you use Night Sight, but they don't seem to occur in the astro mode.

[@dinomonaco2665]

Cuiv, thanks for the excellent explanation. You made a statement that you do not like using bias frames with cmos sensors. Could you elaborate a bit on this. Thanks 🙏 for your help.

[@CuivTheLazyGeek]

I actually have a video coming up on calibration frames and on that topic in particular, but long story short, I was traumatized by bias frames with the 1600MM and the 294MC. Those two really needs to have a 0.1s and 1s long bias frame (respectively) apparently, and weird stuff happen if you try to go shorter. So I take dark frames (which contain the bias signal), and flat darks (which also contain the bias signal), making sure that my flats are 1-4s long, so flat darks don't end up suffering from the low exposure calibration disease...

[@dinomonaco2665]

That would be great. I am just having a difficult time with calibration frames in general with my ASI1600MC.

[@expatca4444]

Thanks ! Your videos are really clear and very well explained

[@geitemans]

I really like your videos! Keep up the good work

[@CuivTheLazyGeek]

Thank you! Will do :)

[@ASTROPLANETES]

thanks a lot for sharing clear, detailed

[@CuivTheLazyGeek]

You're welcome! I really hope it can help people - dithering is something I see too many beginners skip, when it is absolutely critical to do...!

[@mariusbatalu]

very clear explanation. I was asking if there is any link between both .... simple and clear :)

[@JoeBob79569]

I did it randomly a few times as a beginner, but I didn't realise it was so powerful. I think I was thinking that the scope would have to settle for like 10 seconds after every image, and that it would increase my time massively. But thinking about it, I suppose it's really only moving a few arcseconds at the most, which is a lot different to the scope slewing at full speed across the sky!

[@CuivTheLazyGeek]

And you can dither every x frames - so dither every 10 minutes for instance

[@woody5109]

Interesting, thanks Cuiv, another great video 👍

[@fernandoapausa4931]

Dither or die!!! Haha, very true. Then you drizzle during the stacking process, right? I will check that in DSS. Thanks for another great video.

[@CuivTheLazyGeek]

Dither or die indeed! You are correct, drizzling should be done during the stacking process, and DSS is capable of doing that. Clear Skies!

[@mysticsailor9]

great video thx!!.. looks a lot like "Hi-Res" mode in Olympus OMD E-m1/ii, iii & X cameras.. shame no Astro software supports them

[@CuivTheLazyGeek]

Yep, Pixel Shift, Hi-res, super resolution, etc. it's the same thing (but with a Bayer matrix). Would be cool indeed to be able to do dithering in body... or even guide via the IBIS! But unfortunately the camera APIs don't provide access to fine control of the IBIS (otherwise some astro software would be on it)

[@simonsinkafai]

My images really improved a lot after dithering. Really dither or die !

[@CuivTheLazyGeek]

Exactly! It is so essential, and gets you so much value, almost for free!

[@MarvelousLXVII]

Excellent video--as usual.

[@CitizenSniiiips]

Nice upgrade in lighting!

[@CuivTheLazyGeek]

Thank you! Just my balcony light, which I can't really use when actually imaging... A bit of a deer in the headlight kind of effect though :D

[@michal_king478]

3:16 I was just wondering wether its worth it when I shoot from a big city and you answered my question 😆

[@JohnMcGFrance]

Another very well explained video. I dither through PHD2 and it makes a big difference. I’m unsure on the drizzle bit though. Is this done in Pixinsight or can it be done with other programs? Secondly, can you explain why you don’t take bias frames with cmos sensors please. If I don’t need them it would save me a bit of time and disk space. Thank you.

[@CuivTheLazyGeek]

Drizzling is done in the stacking software - in PixInsight the process is a bit involved. It's also apparently available in DeepSkyStacker: www.lightvortexastronomy.com/stacking-with-drizzle.html . Your best bet will be Google until I get to it! On the bias frames, I have an upcoming video that briefly goes into it, but I have read that some people have issues with having correct bias measurement in very short exposures in the CMOS cameras. Since the bias signal is contained in the dark frame as well, I don't even bother with bias anymore, and just take many dark frames (which is easy with cooled cameras). I hope this helps!

[@RobB_VK6ES]

Hey Cuiv, great video but it looks like your video camera does not speak Italian ;)

[@dankahraman354]

I have to figure out Dither Scale in PhD2 versus Dither Levels.....

[@seek.l2215]

So if you are slightly off from polar alignment and the image drifts slowly over the course, would dithering even do anything?

[@allenshafer1768]

It's the law!

[@valentinotera3244]

I dither manually. Each 15 frames I push the button on the hand controller (Z pattern). Seems to works. (finger cross)

[@filipgloria]

Really good and informative video! A question: how should I think when taking calibration frames in combination with dithering? Should I still take darks and flats but skip the bias frames? Is there any point in taking the darks (thermal noise) if that too is averaged out?

[@CuivTheLazyGeek]

Take darks and flats as usual. Sure you could get away with no darks, but should still get a better result with darks!

[@filipgloria]

@@CuivTheLazyGeek: Thank you so much for your amazing tips! I took a picture of the Leo triplet and was amazed with what you can capture with only a DSRL and no filter! I took five minute exposures (it’s cold here in Sweden you know...) in a bortle 3 (and dark in Sweden aswell!) and they were so clean! The darks and flats really helped though even though I dithered!

[@JeramieCurtice]

Thanks for doing these videos. Very informative. I have a further question however. Does dithering always require a second camera? I use Canon DSLRs and notice some options in the custom settings that sound like it may be dithering options, but don't call it that. Long exposure noise reduction, High noise reduction, etc. Are those the same? Definitely increases render time by almost double before the photo is even processed from the camera.

[@CuivTheLazyGeek]

Let me address the Canon DSLR options - by all that is good in this world, turn off ALL noise reduction settings in your camera for astrophotography. The Long exposure noise reduction closes the shutter and then takes a second exposure right after the first of the same length (e.g. a dark frame! doubling the time it takes) before subtracting it from the light exposure - you don't want to do that. You want to stack multiple dark frames and calibrate your lights that way. I will be covering all of that in my lazy guide to astrophotography series - will be lots of episodes, and coming over time. The other NR settings perform noise reduction on the subframes, which you never want to do. You want to stack the frames first, and then manually perform noise reduction as needed. You can actually dither without a second camera - in the video you see me connecting "Direct Guider", which is basically a direct mount connection in NINA- it will then dither "in the blind", directly sending pulse guide commands to the mount, without using a second camera. SGP has a similar option (Direct Mount Guider ).

[@JeramieCurtice]

@@CuivTheLazyGeek I was wondering about the guider part in your video. I'm new to a lot of this so I misunderstood and thought you had a guider cam. Thanks for elaborating. And thanks for the detailed info on the DSLR NR and LE settings. It is considerably longer to expose, which is why I shut it off quickly after trying them. I also didn't see too much difference.

[@astroberg4774]

Hi Cuiv i have 2 scopes and cameras on 1 mount. Im using 2 asiairs, so cant sync dithering. Can i not dither.? Or is it better to dither on 1 camera and just drop a few frames on the other? Thank you

[@AntoninoNicoletti]

Great video again! May be a silly question but would you recommend to do dithering also on Darks and flats (I don’t ask for bias since you say you don’t shoot them anymore...)?

[@CuivTheLazyGeek]

Nope, no need to dither on flats and darks

[@dankahraman354]

There is a new app called PhD2 Dither. Just came out recently.

[@CuivTheLazyGeek]

Good to know - for apps like MaximDL

[@Quoutub]

9:00 Doesn't look like a huge difference at all. Bicubic upscaling would look similar. In my opinion such "super resolution" techniques are extremely overrated. I tried it with raw files and the main advantage is the better signal to noise ratio, which leads to more detail. Such "super resolution" techniques might lead to more detail even in areas with a perfect signal to noise ratio, but in those cases the advantage over bicubic upscaling is extremely low, probably less than 1.1x lossless zoom (notice that a 1.2x lossless advantage would be already equivalent to the difference between a 12.0 megapixel sensor and a 17.3 megapixel sensor.)

[@francoisturcotte1464]

Great channel Cuiv! Is there a way to dither when using a camera not supported by Nina or other capture software. I am using an olympus em1-iii with an intervalometer, and just got a guiding camera asi290 mini. While phd2 does the guiding can we manually tell it to dither in between shots? I will eventually get a scope and astro camera but wondering if I could do something to get dithering in the meantime.

[@shaunozs1ra92]

A question I have but not tried yet. Is it not counterproductive to add bias and dark frames in your stacking routine when using dithering. I’m thinking that it is going to introduce noise.

[@CuivTheLazyGeek]

So I have some more videos coming up about what bias and dark frames actually achieve - it is not counterproductive (with bias and dark, you are subtracting the average of the offset added by read noise and thermal noise respectively - but you are not removing the uncertainty, e.g. the random noise associated with them), it actually still useful and helps remove the offset in addition to what the dither does. Hope this makes sense!

[@Jastro-qv7bv]

Cuiv, thanks for the great videos, I really like your explanations, and you're really helping a beginner like me. I would like to start doing dithering. You mentioned in this video (and at least one other video) that it takes time away from imaging. How much time does it take to do dithering?

[@Jastro-qv7bv]

Found the answer! (at the end of your video.)

[@CuivTheLazyGeek]

Excellent! :) Thanks for the feedback by the way! Also, if you have a lot of DEC backlash, PHD2 dithering may take more time to settle because PHD2 is a bit too smart....

[@dezm0n679]

hi again, i'm wondering why DIrect Dither can take so long it says at the settings of guider that Direct guide duration is 138.2 other settings are: Dither pixels = 12 Pixel scale = 17.3 Guide rates = 1.5 Min. settle time = 5sec

[@CuivTheLazyGeek]

17.3 arcseconds per pixel is VERY large - a guide pulse is at max sidereal rate. To dither 12 pixels, you need to dither 208 arcseconds (roughly 0.06 degrees). And apparently you guide rate is super slow at 1.5 arcseconds per second. So yeah, 208/1.5 is indeed 138 seconds. What is the guide pulse rate you have setup in your mount?? I'd recommend going backwards and diminishing to something like 2 pixels and see what happens.

[@dezm0n679]

Yo, just 5 minutes ago i solved it myself and decided to write here 😮 I found guide rate settings in EQmod (i use it instead of SynScan), 0.1 guide rate by default, suuper slow, set it 0.9, and there it is, 9 times faster. 😊​@CuivTheLazyGeek

[@dezm0n679]

​​@@CuivTheLazyGeekyeah, 17 arcseconds is large, but correct, I guess. I used 50mm lens 😅 4.2 pixel size

[@CuivTheLazyGeek]

@@dezm0n679 there you go!!

[@jurgenvandemaele2896]

Hi a question what do I need to buy first? A telescope or guiding for dithering. I have the star adventurer mount, a Canon 550d not modded sigma 70-300 mm lens Laptop and getting 3 minutes exposures

[@CuivTheLazyGeek]

Depends on how well the lens works for you! The Sigma 70-300mm is a cheap lens. It's really difficult to recommend one over the other...

[@lepetitbaigneur9073]

i'd like to ask how many pixels should i dither? On a cropped DSLR(t3i) and a SW 130PDS

[@MoonbeamArcade]

Is it better to go with 3x drizzle, or is there typically diminishing returns and 2x drizzle is good enough?

[@CuivTheLazyGeek]

There are definitely diminishing returns - I've personally never done 3x drizzle, but I'm fairly sure it would be overkill on most configurations..!

[@fmrc69]

Would you bother dithering with a tracker in RA only? I find that with the Sky-Adventurer and dithering with the ASIAIR pro, it really wastes a lot of time when taking subs, almost by 30-50% so I end up with way less total integration time in the end. I'm considering not dithering with the tracker to get more subs, but will dither once I get a "real" GOTO EQ mount in the future. Thoughts?

[@CuivTheLazyGeek]

I would still dither in RA only - however dithering shouldn't take that long unless you have tons of backlash. Doesn't the ASIAIR Pro not allow dithering every n frames?

[@fmrc69]

@@CuivTheLazyGeek Yes you're right it does allow n-frame dithering! I wasn't aware of that. Ok will try with it dithering every 3rd or so frames maybe? Thanks!

[@technocore1591]

Hey thanks for the vids. One quick question, I use APT and PHD and believe I have the dithering set up right (PHD notes it's dithering between captures) but I was wondering when processing afterwards in PixInsight, is there something specific I should do to let PI know my frames are dithered? Is there a checkbox I should be ticking or is it handled as part of the normal process of stacking? Thanks!

[@MarvelousLXVII]

Cuiv you and I have about the same setup (HEQ5, Sharpstar 61, 120mm guide camera.) I have been dithering but for the life of me, most of my frames after the dither look like dumbells--like the star gets imaged and then the mount moves slightly and then the imaging continues. Think of two stars (well, the same star) with a line between them. I am at wits-end. Is there a way to actually pause the imaging for say 20 seconds after a dither to give PHD2 time to move back to the right spot? I have messed with the settings to no avail. Love your channel hope you have good advice for me lol.

[@CuivTheLazyGeek]

Multiple options available in NINA under the guider equipment setup. The settle pixel tolerance (the lower the longer to settle, but the more stable it is afterwards) and the minimum settle time (which you could set to 20s).

[@MarvelousLXVII]

@@CuivTheLazyGeek Thanks a million!!!

[@randallpatrickc]

is dithering useful in EAA?

[@CuivTheLazyGeek]

It can help, EAA can also have walking noise - but it's less critical!

[@dankahraman354]

Cuiv, how to you put in the settling time into your subs (lights) plan? If I am doing 2 minute subs and settling time is 15 seconds and I am dithering with each sub do I add the interval of 30 seconds between each sub?

[@darrenwarne9005]

Did you ever revist dither off warning in NINA?

[@AndrewBirkett]

I thought Drizzle was only useful on under sampled images?

[@CuivTheLazyGeek]

You are correct, drizzle is typically for under-sampled images. However, some users testing drizzle with different VariableShape kernel and found that drizzling and resampling can provide an SNR enhancement (square kernel is apparently not great however). I haven't studied the maths behind drizzle yet, so can't give an intelligent answer besides what I heard for now...

[@AndrewBirkett]

That is interesting, I'll do some research. Many thanks.

[@dankahraman354]

Cuiv, doesn't guiding in PhD2 do dithering?

[@CuivTheLazyGeek]

PHD2 has a dithering function indeed - it relies on the capture software to tell it to dither (in between exposures, as PHD2 has no idea about what the main imaging camera is doing). So PHD2 on its own will not trigger the dithers, but it is the one that does the dithers proper once triggered by the capture software.

[@dankahraman354]

Does this mean you can get rid of dark and bias frames?

[@CuivTheLazyGeek]

Well, I never do bias, but I still do darks. Doesn't hurt...

[@dankahraman354]

You are getting more signal from a low res camera by dithering you aren't getting something that wasn't there to begin with.

[@eplumer]

where are you doing your drizzling?

[@CuivTheLazyGeek]

PixInsight!

[@Quoutub]

2:44 That's wrong. Google doesn't use machine learning in order to align the frames. Also, they don't seem to dither in order to remove hot pixels. As far as I know, they have an algorithm that recognises the hot pixels. They even said that they can recognise the hot pixels because they don't move.

[@dankahraman354]

I prefer "Don't be Evil Google!"..

[@CuivTheLazyGeek]

Hehehe, good one...! Agreed, it won't help!...

[@johnybaby2]

Don't get why you need to dither without guiding. Surely the fact you need to dither is a side effect of guiding? Eh?

[@canjosh]

Because tracking errors inherent to the mount are not random-they are predictable and recurring-this is the basis for periodic error correction. Dithering is random.

[@johnybaby2]

@@canjosh thanks for your reply.

[@user-wp5dz4xt6p]

My shitty mount always dither)