Ugly image after STF stretch! What's the problem?
Vložit
- čas přidán 14. 10. 2024
- I've had this problem come up myself and have had others ask me about it. So in this video I'll show you why your image is getting blown out and ugly looking after you apply the STF stretch to it.
Help support the channel! If buying astro gear use these links below and I'll receive a small commission at no cost to you:
OPT (USA): bit.ly/37ZCnzv
High Point Scientific (USA): bit.ly/3CMPKEy
Ontario Telescope (Canada): bit.ly/3cncCjE
All-Star Telescope (Canada) bit.ly/3PZFyxI
New to Pixinsight image processing? Get in touch! I provide one-on-one lessons via Zoom.
Join other VisibleDark supporters on Patreon here: bit.ly/3v5dEaB
Thanks for watching! Clear skies!
Shawn, I am glad you covered this topic! Stretching methods, whether its with my pixelmath script, STF, Ez Soft Stretch, etc, these methods stretch an image to a targeted median pixel value. When you perform denoise on a linear image, you change the median value of the image so the image can be blown out when you stretch it. To prevent this, say using my pixelmath script, you just need to open my math and reduce the median background value. I would also like to mention that when using Blur Xterminator (BXT), you can denoise in non-linear, because Russ actually stretches the image internally, then denoise and then return it back to linear. I believe he still does this in his process unless he changed something but using BXT you can perfectly fine using this in non-linear mode. If you are not using BXT and say you use EZdenoise on a linear image, then you will need to reduce the median background value within my math to control the stretch. Hope that helps buddy!!
This is exactly what the STF is designed to do, it’s a feature not an issue. STF is designed to produce an aggressive stretch so you can see everything, including the noise floor. Once you use NXT you massively reduced the noise and so reapplying STF produces an even more aggressive stretch to capture that new noise floor. Of course if you aren’t aware that is what STF does it can be a shock so great to have a video out there that showing you can deal with it using appropriate conservative stretching of your image.
I save auto STF settings before NXT by dragging SFT settings to desktop from triangle button and then use that saved setting to auto STF image after NXT. Works too! :)
Same here :-) In fact this is a pretty good method since you can use the previous settings as a refrence and do some further "finetune" manually in HistogramTransromation.
Woot Woot 😅😅😅
Nice to see Bill and Paul checking in!!
Yup. Watched ‘til the end. I look at this as a ‘pearl’ of wisdom. Now I know how to remedy this if it occurs.
I just ran into this problem today! Lifesaver🤘
Thanks. Noticed this myself and came to the same conclusion.
Woot! Woot!! Had that happen before, thanks for explaining. Now that I’ve been using PI for a couple years, I’ve forced myself to learn multiple ways for stretching an image. Really like GHS!!
Definitely a Woot Woot from me Shawn. Clear skies 👍🏼
Woot Woot... thanks for the video, I've had that problem several times and wasn't sure what the issue was.
WOOT! WOOT! Great video. I have come across this several times and now I know the reason for it.
I did watch until the very end LOL. I have seen this exact scenario as well and I’ve just always resorted to using noise exterminator after stretching. Now I know why I have to do that. Thanks for the video.
I am just starting to use NXT and observed this phenomenon. Your video is a great explanation of why this happens. Before I found your video, I used the ScreenTransferFunction process to see what was happening when NXT was applied to the image. The values on the STF AutoStretch were radically different before and after NXT was applied to the image.
A super easy work around is to preserve the STF AutoStretch settings before you apply NXT. To do this open the ScreenTransferFunction process and use it to apply an AutoStretch to your image before using NXT. Assuming that you like what you see, create a new instance of the STF values by dragging the blue triangle to the desktop to create a process icon populated with the current STF values. Name the icon, “Pre_NXT_Stretch” or something that you will recognize. After you apply NXT, do not AutoStretch your image by using the native PixInsight AutoStretch function which now creates an ugly, blown out image. Instead, you can restore the AutoStretch to your image by dragging and dropping the Pre_NXT_Stretch icon onto your image. This process is a quick and easy method to restore the previous STF AutoStretch values to an image that has been processed with NXT.
I watched until the end. Woot woot. Thank you for letting us know about this HT issue. I, too, have experience this without giving it much thought. Now, I know! :)
Excellent video, ...was just encountering that issue today, ...that saves me me extra hassle to figure it out by myself. Many thanks
Why doubting that we don’t watch your videos to the end? Your videos are always to the point - no blablablabla but real interesting and helpful stuff. Keep up the good work Shawn and looking forward to your next video already!
Cheers
Bert
Just got the idea from Steve at Entering into Space. He had done it and thought it would be fun 😄
@@VisibledarkAstro 🤩
Woot woot! This (along with your pinned comment from Mr. Blanshan) clarified something i found incredibly frustrating when first started using BXT and NXT, so thanks for taking the time to suss this out and post.
Woot Woot! I like your processing skills Shawn!
Thanks so much Ray!
Woot woot! Another great video Shawn. Looking forward to seeing you on Astronomy Scotland zoom.
Thanks Shawn. Woot woot! I have encountered this issue to and just by trial and error decided to use noise reduction later in the workflow. Thanks for the vid. Dr B from Manitoba, Canada 🇨🇦
understand the processing a bit more now , happened to me a few times and as a novice i just shut it all down and started again but now i can see what i was doing wrong. thank you
Woot, Woot. Shawn, yet another brilliant video. I was experiencing exactly this problem and had worked out that probably caused by NXT. So thank you for suggesting two (!) solutions. Keep up the good work.
Still working out a good work flow in PixInsight. Tips like this are really helpful, thanks!
Woot woot. Thanks for this video, it was driving me nuts trying to figure out why stretching was blowing out my images. I ended up doing a manual stretch after BlurXterminator and NoiseXterminator. So, I guess I figured it out, but thanks for clearing it up for me.
Woot Woot, I've experienced this issue many times, thanks for the info. Tim
Hoot-hoot! Thx for clarifying this cuz I’ve been experiencing this issue and it was frustrating…
Woot Woot! This has been driving me crazy for the last few months - thanks! I only see this when using my 16 bit ASI2600 camera and not the ASI294 14 bit camera.
Woot woot! Thanks for putting this out. I have been scratching my head as to why I was having this same issue.
You're welcome!
@VisibleDark do you think running NoiseXTerminator prior to ABE could have a similar effect on how ABE functions? I find that ABE has not been working well for me lately, and just now had the thought pop in my head if these two issues could be related.
@@tannergerstner1524 it will affect it. You want to do ABE or DBE prior to NoiseXTerminator.
Thanks Shawn! Great info. Many users have seen this effect and now know the reason and cure…Your work is appreciated by all. Oh ya Woot Woot
Woot woot! That has happened to me more times than I can remember and I alway thought to was some bug in PI. Thanks for explaining.
Thanks Shawn, I myself have come across this a few times and drives me nutz. and here is your Woot Woot :)
Woot woot!! Thanks, had same problem and was having to do manual stretch!
Woot woot!😊
THANK YOU! I've actually seen this a few times the last week or so and it was driving me bonkers! I appreciate how you not only gave us two suggested fixes, but also took the time to explain why it was happening.
Bingo!! You just showed what was frustrating me and its cause. Good to know. Thanks from a Noob. (Actually resurrecting a hobby that came from my days of film astrophotography and a new scope with digital astrophotography, my how times have changed.)
Great info, Shawn! Here's a "Woot! Woot!" :-) Thanks for the tip. Very helpful.
Woot woot! 😊 I think I’ve had this issue with EZ Denoise too. Now I know why, thank you!!
Woot Woot! I had run into this recently. This really helped, thanks.
woot woot, good stuff Shawn. I have experienced this and it make one crazy trying to figure it out, so thanks for the tutorial. very useful stuff.
Whoot Whoot, thank you Shawn, I've been struggling with this issue on a dataset all week, this has given me hope! Keep it coming, love what you do, cheers.
Woot woot! 😂 I've experienced this sometimes. Never thought of this. Thanks!
Happy to help!
Thanks for this Shawn, it happened to me the other night and I couldn’t think what I’d done differently .. now I know and a simple solution is always the best solution. Cheers from NZ.
Thanks. I’ve been trying to work through this myself.
I like all your videos. You're one of my go to guys.
6:50 Thank you a Shawn for tackling this issue. I was going crazy trying to figure out what misses the image when I apply STF after NXT 😵
Thanks! this has been driving me nuts. Woot woot!
My solution to this is just keep the STF stretch parameters in HT before denoising (drag the STF to HT), then denoise the image. The previous "mild" STF stretching parameters are still in HT so I can easily use it as an reference and do some "finetune" manually with HT.
Woot Woot!! Always interesting to hear your take. I learn something every time. Stretching is still a tough hill for me to climb, so more videos will always be welcome.
Woot! Woot! Thanks for this video Shawn. I have had this problem occasionally and couldn’t figure out why it was was so random. I don’t always apply noise reduction at the same point in the workflow.👍
Thanks Shawn woop woop keep up the good work always very informative
woot woot - glad you made a video on this, I was pretty perplexed for a while on this but it really made me learn GHS stretching more ;)
Woot Woot. Thank you. Was dealing with this exact problem. CS
Woot! Woot! I always watch to the end of your videos Shawn! Excellent video once again, and more food for thought. Up until now I had always performed my noise reduction on the linear image, and then stretched.Things will need to change! 🙂
Thought it was just me. Thanks Shawn. Hoot!
Woot Woot! Your hints and tutorials are always an inspiration and your dedication to enlighten un-stretched PI users is very appreciated!
Woot Woot! Nice job and very helpful as always. Thanks.
Thanks Shawn _ Was fighting with te same issue - Great viedo as always 🙂
Thanks for explaining this Shawn. I’ve seen this a few times myself. (Oh…and Woot woot!)
Thanks for the super thanks! That's so awesome. :)
Woot woot I've run into this too. Thanks for the videos
Thanks Shawn for another great video.
Woot Woot!! Hey Shawn, I'm so glad you covered this! I used NXT on separate NB images BEFORE combining them in ChannelCombination while still linear and I got kind of a blotchy background...now I know why. Thanks for showing this to us! CS
Woot woot! I ran into this issue. At first I worked around it by manual stretching via CurvesTransformation instead of STF after NoiseXterminator. More recently I have started to use numerous minor applications of ArcsinhStretch after NoiseXterminator which seems to do a very good job of preserving color and I like the results I am getting.
I don't know how ideal this is but my usual workflow has become: ABE,Background Neutralization, ColorCalibration, NoiseXterminator, mild run of BlurXterminator, ArcsinhStretch then CurvesTransformation for final adjustments.
You should always do BXT before NXT!
@@VisibledarkAstro Thank you for the tip! I should have known that. I really appreciate that you took the time to let me know. I will adjust my process accordingly. By the way, I feel so blessed. I am batch processing a bunch of aurora images from last night. Not using BXT on those so I should be good to let it roll. Clear skies. :)
Woot woot to ya! Another helpful video. Thanks
Great video how and when to apply the stretch with different tools. I personally do noise reduction after stretching.
Thank you!! Huge mystery and frustration solved for me too. I usually ended up starting over or just going with a manual stretch. I had a suspicion it was NXT but wasn’t sure why more people weren’t talking about this! 😅
Glad I could help!
Informative video, thanks. I’ve frequently found that NoiseXT doesn’t do the best job on linear images so tend to use after stretching. PS: woot woot
Hi Shawn,
Man, I sure thank you for this video! I have been pulling my hair out (really, I don't have much anyway) trying to deal with this issue of running BXT on linear stf stretched images and having a horrible background. NXT on non-linear for me from now on! Thank you!
Jim
Woot woot Better late than never! Hi Shawn!
Great video Shawn, whoop! whoop!
Hi Shawn,Thanks for the info yes this is very useful.I was wondering why this was happening but as Bill just said I just changed the median background value and it seemed ok and I watched to the end!!!
Your content is really useful and you have really improved my knowledge so I really appreciate it.Please keep up the good work
Cheers
Stu
My exact issue and saved my bacon…wasn’t an issue until it was and slight change to my workflow and problem solved
Good info. Thanks Shawn
Woot woot. I watch all your videos!
Definitely experienced this! Thank you for taking the time to deep dive, figure it out and explain it!
My pleasure!
Woot Woot and MANY THANKS, a great help
Woot woot - got some new ideas on processing.
Woot woot! Thanks, I've seen this happen as well
Great job as usual! Not experienced this myself but I can see that this can confuse people. Only use noisexterminator after stretch as that in it self can produce some noise. 😊
Woot woot! solved another of my problems thanks
Woot woot! We'll done - I hope you have some more videos on stretching images, I could really use some help there. I've been using Images Plus to do digital development or ArcSin H stretching before moving to Photoshop for final processing. PixInsight has been hard for me but I'm making progress. I really appreciate your videos, they really help! 😊
Woot woot! Watched to the end!
The problem happens not only with NoiseXterminator. It happens also in almost the same way if you apply the MultiscaleLinearTransform to reduce chrominance noise. After that the auto stretch produces a bad result. To be honest, I usually work manually with HistogramTramsformation but the mess is not 100% resolvable. Some sort of chromatic noise still remains. I’ve sent the issue to noiseXterminator developer with some images on how histograms are altered (with a sort of “peaky” shape of color channels). Same with MSLT even if in minor measure.
Woot Woot! Very helpful video! Thanks!
Glad it was helpful!
Great video as usual! What would your workflow be for mono images? Some people think noise reduction on each RGB or SHO image and then combining later is the way to go what are your thoughts?
My workflow I apply noise reduction on the combined RGB or SHO image. Not individual channels. I've not seen a convincing difference. Especially now with NoiseXterminator.
Great work Shawn. I was wondering what in the pixel Gods was going on as well.
Nice demo Shawn, the problem is not NXT as you mention, the problem is the default setting of the STF (that you can change by the way). I almost stretch manually all my image (by not transferring my STF to HT), except in a few cases, Have a nice day.
Hi Christian - I did say the problem wasn't NXT in the video. And yes the STF parameters can be changed but most won't want to get in to that. Bill's pixelmath stretch parameters can also be changed and is easier to do. But IMO just avoiding all that is the simplest approach for most which was the focus of the video. Thanks!
Woot woot! I am never sure where in my workflow to apply NXT and everyone does it differently. Some in linear, some almost at the end of processing. What I don’t understand is if you are leaving it till late in your processing, aren’t you basically processing the noise and it could hinder what you are trying to do?
Shawn, thanks for this. BTW, I'm seeing the same phenomenon with the stars image produced with StarXterminator. Interestingly, blend seems to produce a working image of my stars image that is properly stretched and can then be used with combine()
Woot Woot, thanks for the video!
woot woot.. thanks for the info!!
Woot woot. Great video 😊
Thanks, I had see not realised what was causing it, really useful thanks. Oh and whoop whoop 😂
Hello and THX for the video! I ran into the same issue a few times and thought it was a problem with a bad DBE or ABE. I figured out for my workflow to use a manual stretch AFTER the NXT or do an auto stretch BEFORE the NXT. Both works well, but I wonder if it's better to denoise the image before or after stretching?! Or is it personal preference?
Woot woot, Shawn! Thanks for this. I’ve been running into the auto stretch issue after running NXT and have simply not autostretched. Have you tried to do any further NR on an image after NXT? I’ve lately worked on two dim nebulae (Dolphin and Abell 35) and have been trying to tamp down chrominance noise. When I try to run MLT in linear after NXT, I cannot get the preview of the Linear Mask to show anything-the screen is just white. If I wait until after stretching, I find a lot of color in my background. So far I’ve managed that with a mask on the target and clipping the dark side of the histogram, but there must be a better way. Have you run into this other post-NXT issue?
Woot woot....thanks Shawn! I’ve seen this too but I maneuvered my way around it by using either GHS or HT to keep the stretch under control. It really freaked me out the first time I saw this and now I know the reason. Just wondering if you’d consider this a flaw in the STF code that might need an update to accommodate some of the newer processes we have now?
Not a flaw - the STF parameters can be adjusted. But most won't want to get into doing that. Easiest is to just avoid the problem, IMO. Cheers.
Thank you
Great video! Do you have a link to download the scripts by Bill Blanshan that you mentioned?
There is a way around. Apply Blur XT then Noise XT. Now critical moment. Push Enable/Disable STF (don’t use Reset STF). Then nuke with STF. No blow up will happen. If you push Reset STF, then blow up will happen. Move triangle to HT, rest is as usual.
Will try it. Thanks!
@@VisibledarkAstro It is better, but it still not on manual stretching level. I think your idea with manual stretching is better than other approaches.
Woot woot! Unrelated question to the purpose of the video, but why does your noiseXTerminate run so fast!? I only have a cpu, but it’s a 13700k. My it runs slow for me due to the lack of gpu?
I have the Nvidia Cuda hack installed so NXT and such runs faster!
I have to say, I've not come across this issue. I use Blur and noise X pre stretch. The only difference is I don't reset the image and use the STF button again.
Woot woot thank you