Should YOU switch? Testing the NEW Fast Batch Pre Processing vs WBPP - Broadband and Narrowband

[@CuivTheLazyGeek]

To try everything Brilliant has to offer-free-for a full 30 days, visit brilliant.org/cuivlazygeek . You’ll also get 20% off an annual premium subscription. My Patreon: www.patreon.com/cuivlazygeek My Merch Store: cuiv.myspreadshop.com/ Amazon affiliate: amzn.to/49XTx01 Agena affiliate: bit.ly/3Om0hNG High Point Scientific affiliate: bit.ly/3lReu8R First Light Optics affiliate: tinyurl.com/yxd2jkr2 All-Star Telescope affiliate: bit.ly/3SCgVbV Astroshop eu Affiliate: tinyurl.com/2vafkax8

[@stephanep1330]

Hi Cuiv, Great Video again! Another arguments for WBPP: - it has the ability to set a minimum weight for the frame to be integrated. I personnaly set it to 0.4 instead of default very low value. I find it better to have more rejected lights than adding bad data. And another argument for shooting darks every 6 months: - the automatic cosmetic correction (based on masterDark) takes way less time than standard cosmetic correction. And uses way less disk space! :)

[@old_photons]

Agreed, my number is typically 30-33 percent for rejected frames. Regardless of the number, something more selective I think is helpful vs the default.

[@snowflakeete]

Interesting I have to try that

[@CuivTheLazyGeek]

Excellent points!! Setting the minimum weight is also something I usually do (at around 0.3 for me)

[@gspinin]

Totally worth the switch. I’ve used FBPP, super fast and you can still enable some options for weighted batch PP.

[@Reverend-JT]

Meh. I stack when I'm at work, personally I have no issue waiting for WBPP to stack. We spend hours and hours capturing objects that have existed for 10,000s of years. I cab handle weighting (lol) another hour or 2.

[@bobbryant4410]

I always use WBPP. I takes way longer, but the results looked better than it did when I tested Fast Track processing.

[@old_photons]

Thanks Cuiv, I wondered what the perception was for others. I can't recall my image I tried it on but I was horribly disappointed in the result when I tried FBPP. Way worse that what you show - but might have been some settings involved that I didn't do right. Always appreciate your efforts to help the rest of us learn - whether in theory or by empirical evidence.

[@CuivTheLazyGeek]

Also to be fair FBPP has been constantly improved since first being released, so maybe it's much better now than when you tried it. Still WBPP for the win!

[@victortapia-montano8114]

Really good info and good to know. Not worried about how long it takes. I installed all the processing programs in a virtual machine and check progress every 3-4 hours. I can see using the light process if you only have one computer and it's using all the RAM or CPU, and you need to pay bills or make a Skype/Teams call.

[@CuivTheLazyGeek]

Yep exactly - many people use their computer as both their main computer and their processing computer... virtual machine sounds like a very good workaround

[@danielpetzen]

Brilliant video, as always. I love these deep dives. I only feed WBPP my finest frames. It is a bit of a primadonna. AstroPixel Processor will chew almost anything up and produce really good results in a fraction of the time it takes WBPP, so I almost never use WBPP anymore. I had 6 hours of data from two sessions with loads of clouds and really different quality frames, so based on your findings, I was keen to feed my frames to WBPP. What took APP about 15 minutes took WBPP 1h40m. I ended up comparing the non-drizzled stacked results. Despite watching you video about noise, I'm still very much a newbie when it comes to SNR etc. I ran the PI SNR script on both, unprocessed, stacked, images. The average over three channels for WBPP was 34db and 63db for APP, which means that I have (mathematically) a better SNR in the APP stacked image, right? The APP image looks sharper.

[@CuivTheLazyGeek]

I really need to try APP....

[@danielpetzen]

@@CuivTheLazyGeek It would be absolutely amazing if you would review APP!

[@Naztronomy]

Very interesting that you had so many hot pixel constellations. It's been a few months since I've done a comparison between the two but I guess I've never seen it because I still do dark libraries. I know Seestar images are automatically dark calibrated but I wonder if we'd see a difference. I'm going to do a test tonight!

[@HeavenlyBackyardAstronomy]

Hi Cuiv, I did notice the same issues on my system when comparing FBPP to WBPP with the hot pixels on the FBPP whereas the WBPP was much cleaner. Your suggestion of using darks is a great idea. I also noticed that hidden galaxy (UGC 2838) in the images around the time code 20:07. The galaxy can be seen at the very top right side. And yes, the contrast from the two was clearly seen in the KZbin video. - Patrick -

[@CuivTheLazyGeek]

Thanks Patrick! Using the galaxy as a "contrast meter" was also a great idea!

[@deep_space_dave]

Fast Integration and now FBPP is only useful if you have well over 1000 images or a slow processor. The best way to take advantage of it is to use darks as the stacking algo is very basic to increase speed so it isn't as good with removing hot pixels or even satellite trails so blinking your subs is critical. I stopped using the cosmetic correction and just use darks even with an IMX585 sensor which has no amp glow but can still have hot pixels. Also I don't use the image created by FBPP, instead I use the drizzle files to create a drizzled image to remove any interpolation artifacts even if I do a 1.0 drizzle scale. I rarely use FBPP as it is reserved for my crazy images that have over 10,000 subs LOL. Thanks for the video!

[@AstroQuest1]

Fantastic video Cuiv. Thanks for the tips on Local Normalization and switching to sigma rejection! Cheers Kurt

[@nadirteymurov1]

Interesting comparison, thank you very much !

[@CuivTheLazyGeek]

My pleasure, glad it's helpful!

[@FrancoGrimoldi]

Great video!! In my experience, shooting from an "almost white" zone, I still get severe LP gradients even with a 7nm filter. My situation might be worse being that I use a non-modified stock Sony camera. Quick note: it seems that KZbin doesn't play well with the "blinking" comparisons, each image takes some time to start showing, the video compression wants to keep showing the same picture regardless of the "subtle" change. Side to side comparisons are much easier to appreciate.

[@CuivTheLazyGeek]

Thanks for the feedback! Have you set the resolution to 1440p on the KZbin video? In theory this should help a lot... But got it on side by side!

[@Spaced_Out_Bill]

Thank you for another great video Cuiv! You've answered all of my FBPP questions in one video.

[@CuivTheLazyGeek]

My pleasure! Glad this is helpful!

[@williamswayne8090]

Excellent comparison, I am inspired to restack my Dark Shark data using WBPP. I have only done it with FBPP due to time constraints. Knowing that I can save a huge amount of processing time by switching to Windsor Clipping and skipping Local Normalization without sacrificing quality makes it worthwhile to compare. I totally agree that PI should not default to the most time-consuming settings. In WBPP you can select for "Fastest" in the Calibration panel and it simply disables Local Normalization but does not change the Rejection Algorithm. You would think if you selected "Fastest" that it would also change that. Great job as usual.

[@markice.s]

I’ve found an improvement in my images when using LN on broadband targets, even picking up faint dust around fireworks galaxy/seahorse nebula. I think that a strong local norm reference frame with appropriate number of subs (1/3 the stack) for the reference stack helps improve SNR as well as gradient removal when using graxpert

[@CuivTheLazyGeek]

That's good to know, thanks for that feedback!

[@thomasrider5852]

I have an i7 14700kf, 128gb of ram and two 4tb nvmes. I can stack 200-300 photos at 5 minutes each in maybe an hour or two if I drizzle 2x. Its really not bad. When I had a DSLR I would have like 2000 photos at 12 seconds each and that would be an overnight stack and that sucked. If I spend 20 hours collecting data I dont mind waiting 1-2 hours for best stacking possible. I dont want to cut corners due to being impatient.

[@Astro_Shed]

Great video Cuiv, I have not tried the new FBPP yet, but will give it a try, as I do use darks with my IMX571 camera…👍🏻

[@CuivTheLazyGeek]

Have fun!

[@RyanParle81]

Personally I like to use FBPP for a preview stack to see if I have enough integration time, or to see if a faint target is worth continuing with. My final processing is always done using WBPP with LN & Integration disabled, i then use Normalize Scale Gradient and it's invoked instance of Image Integration for my final stack. Typically this process takes around 8 hours on my machine with around 30hrs of 2 min subs

[@CuivTheLazyGeek]

That sounds like a great way of proceeding, makes sense to me!

[@TucsonHippy]

I have Seestar so I have to use a lot of subs. Ive found that there is a brick wall for integration at about 1024 subs. up to 2 days processing time was normal. I discovered Winsorized Sigma Clipping a long time ago and its been a lifesaver. As for the difference between WBPP and FTPP is night and day to me. Even tho I have a Bortle 1/2 site I go to I can still see the difference. WBPP all the way for me.

[@CuivTheLazyGeek]

Excellent to hear! Bortle 1/2 sounds like a dream!

[@elythomaslumber]

Thanks a lot again! A lot of people are stacking with other programms due to time reasons, mainly APP or Siril and then continue processing with PI. Would be great to see who's the winner here in comparison to FBPP or finally WBPP :-) Lot of work...

[@jesusalbertoponcedeleon7124]

Cuiv, where do you keep or store so much stuff and equipment? Japan's homes are not on average large!! admirable!!

[@actudoran]

awesome dude 😅 I got the following algorithm: 585 sensor deep sky stacker DSLR or mirror less R7 in my case, sequator and play wit light pollution alogos ... oh ... sequator processing times are lighting quick!!! didn't go pixinsight way yet ... but good to know!

[@Chorge1972]

Thanks!!! Comparison with a Siril stacking would be awesome!!!

[@CuivTheLazyGeek]

Siril, APP, WBPP, etc... Would be an interesting comparison for sure.

[@TheSfyfe]

Great breakdown Cuiv! Have you ever done a high level breakdown of WBPP and what each step does? I use WBPP, but will be the first to admit that I’m pretty ignorant as to how it actually works 😂

[@CuivTheLazyGeek]

I think the very best breakdown is done by Adam Block, I would never be able to make a better summary. But I'll keep it in mind!

[@memyselfandi906]

Siril can calibrate amd stack a weighted batch of 500 images in 5-7 minutes on a nine year old pc running only an i7-4790 , 24Gb of 1800MHz RAM, and SSD @ read /write 300Mbps. Understanding your computer 's bottlenecks is key. The same run of 500 images ran from a HDD @ read/write 60Mbps took 2.5 hours. So... 7 minutes versus 180 minutes, the only difference being the read/ write speed AND : read from the HDD and process on the SSD! Run the process on your pc, hit Ctrl+Shift+Esc. Then check the activity of the harddrive, memory, cpu and video card. The one running at 100% will be your bottle neck. Happy processing! Be it PI or Siril. ;-) Clear Skies!

[@memyselfandi906]

Hi Cuiv! Here's an addition to the above, using the Linear Fit Clipping instead of Winsorized Sigma Clipping will, ehh for me.., ehh so.. can? , get rid of the weird red clipped pixels. 😊🤗

[@abbreviateTome]

Hi Cuiv, love your content. When you’re Not feeling Lazy( pun intended🙃), can you make an update video (i.e 6 month review) of the WarpAstron Wd-20 mount. Thank you 🤩🙌

[@CuivTheLazyGeek]

Good idea, I'll think about it

[@richardshagam8608]

Until recently, I've been using Siril for my stacking, as when I tried WBPP, I was totally put off by the excruciatingly long processing time. I recently tried FBPP and was impressed at how fast it goes. My only complaint is that sometimes I forget to change my filter setting name (I switch them manually), and I find that the images with the wrong filter designation in the FITS header are then processed separately. I did find I can edit the FITS header in a batch script in PI, but it's a pain. Other than that, I think I'm now sold on FBPP--sorry, Siril!

[@CuivTheLazyGeek]

Whatever works for you is the best!

[@rawhead909]

I've been using WBPP but with the new "Fast Integration" option ON.

[@CuivTheLazyGeek]

I need to try that too at some point :)

[@lcmattern]

Question that came to mind watching this video. Does noise exterminator work with large color images? (100+mp) I have had to scale images when doing denoising with graxpert as it does a poor job imo at this point. Or is there a setting i can adjust there?

[@rickerr8481]

Hello again. Could you use your critical thinking, do an in-depth review of the "object tracking" features of your Dwarf 3, Dwarf 2, and any other third party telescope control and imaging software that you know of that has these tools. I am interested in possibility using a scope to film rocket launches. Thanks.

[@saulontheroad9605]

Hello Cuiv, I've been struggling to get a decent image of the Pleiades from Montreal (Bortle 9), with no success. The light pollution here is so intense that I haven't been able to capture any of the blue nebulosity. I noticed you're using very short exposure times (around 20 seconds), and I’m wondering if that might be the key? I typically use exposures of 180 to 300 seconds per target. Thanks in advance for any advice!

[@CuivTheLazyGeek]

From a white zone like that you are likely saturating your subs at such subframe lengths. 20s-30s would work much better!

[@saulontheroad9605]

@@CuivTheLazyGeek Thank you so much !

[@timbotron4000]

Thanks Cuiv for this video. I had been using FBPP on my poor, little, old laptop and it was working pretty great, but I'm in the process of building a new PC and plan to start using WBPP again. In your opinion, with the amazing gradient correction tools like Pixinsight's GradientCorrection process and Seti Astro's Automatic DBE, is it worth it to use local normalization at all?

[@CuivTheLazyGeek]

Others have gotten good results with LocalNormalization so I would say it's probably worth it? I need to do more tests...

[@clownworlddotfart]

How many secs were the subs of the dual band images?

[@CuivTheLazyGeek]

300s

[@JakobSlaick]

Hey cuiv, do you still have the askar V? I just ordered one and want to know what your long term opinions are if you still have it.

[@tomhoskins4913]

I have been using DSS from day one in this hobby. I bought Pixinsight about a year, but I still use DSS. What are the advantages of using WBPP?

[@dadwhitsett]

Up until recently I used DSS because WBPP took 12 to 14 hours on my old PC. I finally upgraded my PC and now with dithering and highest quality rendering, I find that WBPP improves the overall quality and contrast in my images. If you can afford it update your PC and use WBPP. I have not gone back to DSS. I can now stack in 35 to 70 minutes even with dithering and highest quality rendering. IMO.

[@CuivTheLazyGeek]

The advantages can be subtle, but especially on broadband I expect WBPP to work better specifically because of the weighing of the frames... But I haven't used DSS in a looong time, maybe I'll need to do a comparison!

[@bronco_fv]

Well, afaic astrophotography is a game of patience and persistence, so I don’t mind the wait for wpbb for my traditional photography and will use fbpp for what it is made for, process lucky imaging material.

[@brokenigmatic]

On default settings, WBPP does a lot better at removing satellite trails from my final stack. There are settings in DSS that likely can be tweaked to improve that, but I'm ... lazy.

[@pompeymonkey3271]

If you're going to spend $x thousand on equipment and y hours capturing the subs, to me it is a no-brainer to have to wait a couple of extra hours for WBPP over FBPP for a much better result. It's like buying a Ferrari and then putting the cheapest tryes on it...🤣 Either that, or buy an EPYC/Threadripper computer for the task!

[@TomSupergan]

I gave up on WBPP. I tried for hours on several different targets, and never got a stack. Now that I know it's supposed to take hours for a single stack operation, and there are better parameters to use, I'll try again. I had even tried on a stack of just 4 subs to try and shorten the stacking time, but always got errors. Then I tried Siril, which is 10x faster, but got its failures to stack instead. Not sure what is wrong with my subs, they look fine to me.

[@LyngJohn205]

“Weighted Batch …”? I thought it was “Waiting for Batch …” 😊

[@CuivTheLazyGeek]

Ha! Indeed!

[@CycloKermit]

Is that the Asus Rog Ally PC? Are you using it as your Astro PC during imaging as well?

[@CuivTheLazyGeek]

Yep, that's the ROG Ally X - I only bought it recently, but it's indeed serving as my main PC (with a dock and with an external GPU) and my travel laptop (with bluetooth mouse and keyboard). I've used it for imaging sessions, but in the desktop mode effectively

[@admiraloctavio5860]

It seems that yet again, faster isn't always better.... ;)

[@CuivTheLazyGeek]

As is often the case!

[@malcolmdent241169]

Hi Cuiv, I don't understand why you don't just use dark frames. I use a zwo 2600mm and I tried it without the dark frames. I saw the hot pixels so I use them. Problem solved?

[@CuivTheLazyGeek]

I've been too lazy to take then when it's cloudy - but yes indeed, with dark frames you can also do some nice cosmetic correction :)

[@CDigata]

Fast food or Home cooked - the only fast part needed is A fast Pc

[@CuivTheLazyGeek]

That's a good analogy!

[@qx3V45p]

I have to admit I just use DSS anymore. Must be lazy!

[@nickverstand]

pixinsight on steam deck?!

[@jimwaters304]

Nope - I did my own tests on two targets and faster isn't better.

[@CuivTheLazyGeek]

Yep

[@StuartAnderton]

I can't see any difference when you switch. I think KZbin destroyed it.

[@damiengalanaud3817]

Very interesting Cuiv, even if personally I prefer processing my images « à l’ancienne », manually, mostly following Jean-Baptiste Auroux a.k.a photon millenium and light vortex astronomy’s guidelines. By the way, could you do a test of the various rejections algorithms in Pixinsight ? There are so many available now that I don’t know which one to choose and stick to the good old linear fit clipping.

[@CuivTheLazyGeek]

Aaah the good old Vortex tutorial, oldie but goodie