11.01
#...
·
·
2
likes
|
|---|
|
Having checked out the trial of NoiseXterminator I have to say that I will be ditching most of my other noise reduction methods. The fact that it is so quick and that you can work directly on a preview in Pixinsight to tweak it to your liking PLUS it seems to be really gentle on stars makes it a winner for me. The default denoise setting of 0.9 seemed a bit to high for me and ended up with waxy images, but at 0.7 to 0.75 it looks quite natural. I can't believe that something that is this fast and can produce what I consider to be much better results than the in-built tools exists. Anyone see any real downsides to it? |
0.90
#...
·
·
3
likes
|
|---|
Hi ive tested both and its pretty good not gonna lie! im not replacing topaz however. still payed 70€  |
|
7.46
#...
·
·
2
likes
|
|---|
Andy Wray: 0.7 is waay too much. If you need this much then you really need more exposure, you still have to reach the noise floor. 0.5 seems a resonable compromise. |
0.90
#...
·
·
1
like
|
|---|
| I've not touched Topaz or DeNoise since I bought NoiseXterminator. Really loving it, can't wait to see future improvements! |
7.14
#...
·
·
1
like
|
|---|
|
I've used 0.7 for some really weak noisy channels, but .5-.6 for everything else. But yeah, for me it's the biggest advancement in processing in recent years. |
4.67
#...
·
·
1
like
|
|---|
|
I use Affinity photo Denoise.. best $24 dollar software out there. I only use Denoise on starless images, I never use it on the stars anyway no matter what noise process I use. Take the stars out, use noise reduction, add the sharp (usually reduced) stars back in. |
8.42
#...
·
·
1
like
|
|---|
| I have to disagree here. I was just trialing it out today and NoiseXterminator had a lot of trouble on star cores that were near saturation (note: they weren't actually saturated yet, but above 0.5 normalized) and introduced pixels that had weird color values. This was on a linear image of course. Stretched images seem to fare better. These pixels weren't visible with an STF or stretched using HT, but when stretching with a star-friendly method (like Arcsinh or GHS) they were really noticeable artefacts. I think it's a part of a stretch-denoise-reverse the tool does behind the scene (the reverse ends up clipping?) in linear mode. The tool is still young of course, and might fix this soon enough, but for the image I'm working on now the Jon Rista method still rules supreme. |
7.75
#...
·
|
|---|
Andy Wray: There is a very nice German youtube channel Astrophotocologne by Frank Sackenheim. He did extensive testing and found a few occasions where Topaz does a better job. He's in direct contact with the Russel Croman, the developer of NoiseExterminator. So we can expect further improvement. To me it's a no brainer I'll buy it once me trial period comes to an end. Frank also recommends values below 0.5, I yet have to perform more testing for my own opinion on my preferred settings. I assume it depends a lot on whether you apply it on the stretched or unstreched image. Clear skies Wolfgang |
|
11.01
#...
·
·
1
like
|
|---|
andrea tasselli:Andy Wray: You are probably right ... I was testing it on my 50 mins of exposure on M106 in a suburban setting. With decent integration time I would reduce it massively. |
1.20
#...
·
|
|---|
| Still going to use Topaz Denoize for non-DSO stuff, but I have NoiseXterminator in both PI and Photoshop and I was impressed putting it at .6 with how I stretch the crap out of my images. I know it's a bad habit, but... |
6.77
#...
·
|
|---|
|
Will be trying it out for sure! Especially since you get the discount if you already have a product, nice little touch to reward loyalty! |
1.91
#...
·
·
1
like
|
|---|
I am sorry I started another discussion because I wanted to ask some impressions about NXT. In my opinion it has done an amazing job in a few cases. In the case I am showing Denoise was 0.6 (no way to increase this parameter otherwise it is too strong) and also details no more than 0.2 to avoid artifacts. NXT vs TVG. |
4.52
#...
·
·
1
like
|
|---|
|
Very noised image. Choose Yours the best. QHY695A (icx695), RedCat51, fews 900s subs. I think nXt is created for cmos matrix. In my case ccd camera this staf (nXt) not working.  |
0.00
#...
·
|
|---|
| Got, use it ( see may last 2 image posts), not gonna use anything else. Since I'd bout gradient terminator for PS a long time ago the noiseXterminator license was discounted. Win-win. |
7.53
#...
·
|
|---|
|
I decided to throw it at the M13 I shot from B8 a year ago. It killed it! In fact it did such a good job I had to inject noise back into it just to ensure that it wasn't too smooth. The most impressive thing I find is how it's able to preserve star sharpness and details even at the higher settings. This tool is really going to be a boon for people who live in light polluted skies.  M13 From Bortle 8 You'd have a hard time convincing me this was 2 hours from B8 if I didn't know any better. |
11.14
#...
·
·
1
like
|
|---|
In fact it did such a good job I had to inject noise back into it just to ensure that it wasn't too smooth. I just saw that particular image and was ready to ask you if you have added some "comfort noise"!  |
6.08
#...
·
·
1
like
|
|---|
| Neat Image... Topaz DeNoise-AI,.. these work pretty darn good for astrophotography considering they were designed for daytime photography, you just need to back way off on the settings. NoiseXterminator is designed just for artro imaging by the best, Russell Croman, I have all his other plug-ins so for $40, it's a no brainer to at least see what it can do, and so far, it looks very good. |
1.20
#...
·
|
|---|
|
Based on the comments here I felt compelled to try it, and agree with all supportive comments. Wow. So far I have only used immediately after going from linear to non-linear, so I really look forward to exploring this more in a linear state. Thanks so much for developing it and I look forward to determining best place in workflow to utilize this powerful tool. Best, Rob |
1.20
#...
·
|
|---|
|
I agree, NoiseX works very well indeed. But it should be used with caution, personally I do not use reduction values higher than 0.40 and in the details I always try to use a value that is about half the noise reduction (eg 0.20). It is important to work well on the applied and inverted luminance mask. The best results in my opinion are obtained using a tightened and curves adjusted luminance mask. In this case it is possible (regenerating the mask each time) to carry out many passes of NoiseX without losing the details that matter. Personally I make my luminance masks for noise reduction with the John Rista method but I DO NOT use TVGDENOISE but rather NoiseX which works better. I also do a light pass at the end of all the processing, in non-linear. |
