serbiadarksky Posted November 8, 2018 Share Posted November 8, 2018 I tried dithering in RA only, later I tried in RA and DEC. I am not happy with the results. I dither more than 10pixels (maybe more like 70) and I still have noise The 2 attached images are here to show you the example. The Orion is 30x4min, I moved the FoV after every 10 images. The other image, Elephants Trunk Nebula is 90x1min at iso1600 and even if I stacked 90 images (dithered after every 30 images) I still have a weird kind of noise. Just to mention, darks are not helping even if theye are properly calibrated, as you can see it on Orion. What can I do about it.. Link to comment Share on other sites More sharing options...
geordie85 Posted November 8, 2018 Share Posted November 8, 2018 You'll need to dither a lot more often. Only dithering twice in your imaging run is just as bad as not dithering at all. Your images seem rather over stretched too. Link to comment Share on other sites More sharing options...
Ronclarke Posted November 8, 2018 Share Posted November 8, 2018 I dither between every sub!! Link to comment Share on other sites More sharing options...
Pompey Monkey Posted November 8, 2018 Share Posted November 8, 2018 The dithering appears to be doing its job correctly as the noise is mottled not streaky. To me, this looks like normal colour noise that is very obvious because the images are overstretched. Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 8, 2018 Author Share Posted November 8, 2018 34 minutes ago, geordie85 said: You'll need to dither a lot more often. Only dithering twice in your imaging run is just as bad as not dithering at all. Your images seem rather over stretched too. If I do 30 images, how many have to be dithered? Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 8, 2018 Author Share Posted November 8, 2018 14 minutes ago, Pompey Monkey said: The dithering appears to be doing its job correctly as the noise is mottled not streaky. To me, this looks like normal colour noise that is very obvious because the images are overstretched. Its obvious even on the stacked tiff Link to comment Share on other sites More sharing options...
geordie85 Posted November 8, 2018 Share Posted November 8, 2018 1 minute ago, serbiadarksky said: If I do 30 images, how many have to be dithered? I would dither 29 times. In other words, I'd dither after every image Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 8, 2018 Author Share Posted November 8, 2018 7 minutes ago, geordie85 said: I would dither 29 times. In other words, I'd dither after every image The problem is that I use the Star Adventurer which can be dithered only in RA. If I dither with 10 pixel movement and it is possible only in RA, I am affraid I will have walking noise, which I had before I dithering. Link to comment Share on other sites More sharing options...
Whirlwind Posted November 8, 2018 Share Posted November 8, 2018 I think you might be misunderstanding what dithering is there to do. At a basic level there are two types of noise that need to be considered. 1) There is a type of fixed noise which may be better described as known errors on the camera. These are things like hot pixels, cold pixels and so forth, bad columns and so forth. These stay in the same location all the time. Darks and bias can partially correct for these but they can accrue charge in a non-linear way which means even with dark/bias correction they can still partially exist (this is particularly relevant for non temperature controlled cameras). If your tracking is so accurate that the points will always fall on the same point in your image then when you combine them it will reinforce this artefact and you will end up with little bright or dark spots over the image. 2) There is random noise. This comes from that it is statistically impossible to measure something at the highest levels of precision. If as an example the flux arriving at a pixel is 300 counts then when you image the object you would find that the you get a range of values with greater frequency towards the actual value. So over five images you may get the following values 295, 298,299, 301, 308. This is called random or gaussian noise. It is impossible to predict the exact value you will measure from image to image. However, it is random and follows a gaussian statistical variation. As such because some values will be lower and some higher you can median or mean (average) combine such data and the result will trend to the 'real' value of 300. The more images you take the more you can average out these random fluctuations. Hence this is why you take multiple images and stack them. When you process the data you exaggerate this effect as you are trying to pull out the slight variations to the signal (for example in a nebula) that might be very similar to the level of random noise you get. Hence with too few images averaged then what is real and what is noise becomes confused. This is what causes the mottling effect. The more images you take the more you can average out the background noise and the more certain you are as to what is data and what isn't. Hence when people refer to images being overdone, overcooked this is what is happening. The details have been over processed to the point that the noise is being processed not 'real' signal. So why dithering? Well dithering is a random jump of your camera close to the target of interest. It makes no difference to random noise because it is not based on a specific location. Wherever you point your camera you will get random noise. On the other hand the fixed point noise (like hot pixels) are tied to specific pixels, columns and so forth. As such they will move about your image when your slightly shift (dither) the telescope. Most software when it combines data will reject any that are hugely discrepant from the average data in your image when tied to a fixed position through star alignment. So suppose you had a hot pixel that had been dithered once at a specific sky position. You may get something in counts like this:- 300, 299, 306, 295, 301, 2000. The software will recognise the last value as discrepant and then ignore that data for that specific pixel and reject it. If you don't dither you would get something like:- 1999, 1998, 2003, 2007 2003, 2000. Hence the software won't recognise that this is a hot pixel but consider it a real value and include it on your image. What you end up with is an image that has scattered single pixel bright spots because of this type of fixed noise. As such when noted above that your image looks like dithering is working it is because there doesn't appear to be any single pixels that are overly bright for their location. The noise you are looking at is random noise which can't be fixed by dithering, only by taking more images and you can never get rid of it completely. So it becomes a balance as to how much time you want to spend on an object and how much processing you can get away with before the remaining noise becomes distracting. Link to comment Share on other sites More sharing options...
don4l Posted November 8, 2018 Share Posted November 8, 2018 This doesn't look like a dithering problem to me. The data is stretched too much. If you look at the blue colour in the Running Man, you can see that you have plenty of blue data. The Flame looks like it has good data too. The bit that is missing is the Ha stuff. In other words, the camera can barely detect the particular shade of red in Ha. Your camera probably has an IR cut filter, and this is designed to cut out the shade of red that you are trying to capture. Edited to add:- Every single photo, taken by every single astrophotographer, has the same noise in it. If you scale your image so that you can only see the nice smooth bits, then the background will become smooth and blackish. Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 8, 2018 Author Share Posted November 8, 2018 49 minutes ago, don4l said: This doesn't look like a dithering problem to me. The data is stretched too much. If you look at the blue colour in the Running Man, you can see that you have plenty of blue data. The Flame looks like it has good data too. The bit that is missing is the Ha stuff. In other words, the camera can barely detect the particular shade of red in Ha. Your camera probably has an IR cut filter, and this is designed to cut out the shade of red that you are trying to capture. Edited to add:- Every single photo, taken by every single astrophotographer, has the same noise in it. If you scale your image so that you can only see the nice smooth bits, then the background will become smooth and blackish. The IR part is interesting, my camera is modified. I will think that it is a lack of signal causing this.. Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 8, 2018 Author Share Posted November 8, 2018 49 minutes ago, don4l said: This doesn't look like a dithering problem to me. The data is stretched too much. If you look at the blue colour in the Running Man, you can see that you have plenty of blue data. The Flame looks like it has good data too. The bit that is missing is the Ha stuff. In other words, the camera can barely detect the particular shade of red in Ha. Your camera probably has an IR cut filter, and this is designed to cut out the shade of red that you are trying to capture. Edited to add:- Every single photo, taken by every single astrophotographer, has the same noise in it. If you scale your image so that you can only see the nice smooth bits, then the background will become smooth and blackish. So you are saying that i should dither more images to be able to average out the unwanted things? Link to comment Share on other sites More sharing options...
IanL Posted November 8, 2018 Share Posted November 8, 2018 There are two other issues to consider with one-shot-colour CMOS cameras (i.e. DSLRs and some astro-cameras): - Telegraph or impulse noise: This is not fixed like hot or cold pixels, but not is it random. A large subset of pixels with small defects in the sensor well will sometimes give 'normal' readings, i.e. close to the true value plus or minus a small noise component, but are prone to reading out a number of much higher values from a set of specific values for each pixel. This makes it impossible to calibrate them out like hot/cold pixels and will build up in to a pattern. Again the only real cure is a large dither in both axes between frames to randomise the pattern and aid the pixel rejection algorithm to weed these values out when stacking. - Chromatic noise - this is simply due to the fact that each pixel only detects one of red, green or blue depending on the Bayer matrix filter. Debayering estimates the missing colour data from adjacent pixels, but any noise affecting a pixel 'infects' its neighbours in that colour channel depending on the debayering algorithm. This is especially apparent when you don't dither since the noisier pixels (hot/cold/telegraph) tend to build up in to larger colour 'blobs' in a given channel, and when you overlay these noisy blobs from each of the RGB channels, you will get the delightful rainbow background you are seeing - with DSLRs it is usually most apparent in the red channel since that gets the least signal and therefore has a lower signal to noise ratio, but it affects them all. Practically: - Red is the primary component of a lot of nebulosity, and unfortunately DSLR cameras which haven't been modded to remove the IR filter are not good at picking it up. There is no point in chasing really faint Ha features with an unmodded camera as you'll end up with disappointing results but even with it's still an issue due to only 1/4 of the pixels detecting red. If you can't mod the camera then pick your targets to avoid the problem. You can see that the upper left of the second image looks pretty good, it's the rest of it that is poor. If you can, increase the focal length to get in on the brighter stuff and cut out the rest. - Dither aggressively (15-20 pixels between subs). If you can't do that due to the Star Adventurer, then manually repoint the camera in Dec as often as you can stand to, every 5 or 10 frames say. - Lots more frames - if you've got light pollution, then there is usually not much benefit in going for long exposures, 1 min or 30 seconds is probably the right ballpark as you are doing, but get as much total exposure time as you can. 90 minutes with a noisy DSLR under light polluted skies is nowhere near enough, you'd want 3 hours as a minimum, and 5 or 10 would be a good aim - I know it's tempting to hop around between targets to tick them off having been there myself, but actually spending a few nights on one target will get you much cleaner results. - I'm a lot less convinced of the benefits of dark frame subtraction on CMOS cameras due to telegraph noise, plus the lack of temperature control means you will not be able to get a matching master dark in any event. I did a lot of experiments in this area and the upshot is that the temperature readout in the EXIF data is not to be relied on for Canon DSLRs and I'd be cautious about any other make unless you verify the reliability through testing. If it's a Canon, you can pretty much forget darks anyway as there is on-camera processing that suppresses dark current in a way that makes it impossible to apply them usefully (not talking about the camera setting to subtract a second dark exposure on camera, this is in the firmware/hardware for all images and not user controllable) - Upshot is that darks may actually end up making the image more noisy, not less. Instead try using a large master bias to remove large scale fixed pattern noise, combined with a defect map process to remove hot and cold pixels. - After that, as others have said, you are over-stretching your data. The highlights are blown out which may be the stretch or just over-exposing when capturing, and the background is way too bright. You can try creating masks for the stars and nebulosity to stretch them more, and then apply a much lesser stretch to the background for example. You can also use the same mask to de-saturate the background (especially red). The background shouldn't be black, but nor should it be nearly as bright as the rest of the image. Good luck Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 8, 2018 Author Share Posted November 8, 2018 30 minutes ago, IanL said: There are two other issues to consider with one-shot-colour CMOS cameras (i.e. DSLRs and some astro-cameras): - Telegraph or impulse noise: This is not fixed like hot or cold pixels, but not is it random. A large subset of pixels with small defects in the sensor well will sometimes give 'normal' readings, i.e. close to the true value plus or minus a small noise component, but are prone to reading out a number of much higher values from a set of specific values for each pixel. This makes it impossible to calibrate them out like hot/cold pixels and will build up in to a pattern. Again the only real cure is a large dither in both axes between frames to randomise the pattern and aid the pixel rejection algorithm to weed these values out when stacking. - Chromatic noise - this is simply due to the fact that each pixel only detects one of red, green or blue depending on the Bayer matrix filter. Debayering estimates the missing colour data from adjacent pixels, but any noise affecting a pixel 'infects' its neighbours in that colour channel depending on the debayering algorithm. This is especially apparent when you don't dither since the noisier pixels (hot/cold/telegraph) tend to build up in to larger colour 'blobs' in a given channel, and when you overlay these noisy blobs from each of the RGB channels, you will get the delightful rainbow background you are seeing - with DSLRs it is usually most apparent in the red channel since that gets the least signal and therefore has a lower signal to noise ratio, but it affects them all. Practically: - Red is the primary component of a lot of nebulosity, and unfortunately DSLR cameras which haven't been modded to remove the IR filter are not good at picking it up. There is no point in chasing really faint Ha features with an unmodded camera as you'll end up with disappointing results but even with it's still an issue due to only 1/4 of the pixels detecting red. If you can't mod the camera then pick your targets to avoid the problem. You can see that the upper left of the second image looks pretty good, it's the rest of it that is poor. If you can, increase the focal length to get in on the brighter stuff and cut out the rest. - Dither aggressively (15-20 pixels between subs). If you can't do that due to the Star Adventurer, then manually repoint the camera in Dec as often as you can stand to, every 5 or 10 frames say. - Lots more frames - if you've got light pollution, then there is usually not much benefit in going for long exposures, 1 min or 30 seconds is probably the right ballpark as you are doing, but get as much total exposure time as you can. 90 minutes with a noisy DSLR under light polluted skies is nowhere near enough, you'd want 3 hours as a minimum, and 5 or 10 would be a good aim - I know it's tempting to hop around between targets to tick them off having been there myself, but actually spending a few nights on one target will get you much cleaner results. - I'm a lot less convinced of the benefits of dark frame subtraction on CMOS cameras due to telegraph noise, plus the lack of temperature control means you will not be able to get a matching master dark in any event. I did a lot of experiments in this area and the upshot is that the temperature readout in the EXIF data is not to be relied on for Canon DSLRs and I'd be cautious about any other make unless you verify the reliability through testing. If it's a Canon, you can pretty much forget darks anyway as there is on-camera processing that suppresses dark current in a way that makes it impossible to apply them usefully (not talking about the camera setting to subtract a second dark exposure on camera, this is in the firmware/hardware for all images and not user controllable) - Upshot is that darks may actually end up making the image more noisy, not less. Instead try using a large master bias to remove large scale fixed pattern noise, combined with a defect map process to remove hot and cold pixels. - After that, as others have said, you are over-stretching your data. The highlights are blown out which may be the stretch or just over-exposing when capturing, and the background is way too bright. You can try creating masks for the stars and nebulosity to stretch them more, and then apply a much lesser stretch to the background for example. You can also use the same mask to de-saturate the background (especially red). The background shouldn't be black, but nor should it be nearly as bright as the rest of the image. Good luck The LP is not a problem, its a bortle 4 class The camera is not 'very noisy'. I will try to dither more frames, and I am waiting for a cable to arrive which will let me connect the Star Adventurer with DSLR so it will make a shot, dither in RA by the amount I enter, continue to shot and it goes in this circle. Link to comment Share on other sites More sharing options...
IanL Posted November 8, 2018 Share Posted November 8, 2018 6 minutes ago, serbiadarksky said: The camera is not 'very noisy'. What is the camera? You didn't say in the original post? Nowhere have I used the words 'very noisy', but it's a fair assessment of the situation. It's not a sleight on you or your equipment, just a fact that DLSRs are noisy for all the reasons I and other posters have explained. Even high-end CCD cameras have limitations and compromises in their design. One needs to understand the limitations and the correct techniques to get the best out of the equipment. Link to comment Share on other sites More sharing options...
don4l Posted November 8, 2018 Share Posted November 8, 2018 1 hour ago, serbiadarksky said: So you are saying that i should dither more images to be able to average out the unwanted things? I don't know enough about CMOS sensors to comment on the dithering. If this was a CCD, then I would say to forget about dithering, but I've read some of the other posts and I realise that I shouldn't comment on this. The image looks like it is over processed. The background noise should be hidden by raising the black point - (in my opinion). I also think that the colour saturation is a bit too high. I would try to gather some more data. I would also try different ISO settings. If you tell us what camera and lens that you are using then someone might be able to advise on the best settings. Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 8, 2018 Author Share Posted November 8, 2018 11 minutes ago, don4l said: I don't know enough about CMOS sensors to comment on the dithering. If this was a CCD, then I would say to forget about dithering, but I've read some of the other posts and I realise that I shouldn't comment on this. The image looks like it is over processed. The background noise should be hidden by raising the black point - (in my opinion). I also think that the colour saturation is a bit too high. I would try to gather some more data. I would also try different ISO settings. If you tell us what camera and lens that you are using then someone might be able to advise on the best settings. Canon 1300Da and cosinon 135mm f2.8 at f5.6 Link to comment Share on other sites More sharing options...
Atreta Posted November 8, 2018 Share Posted November 8, 2018 14 hours ago, serbiadarksky said: I tried dithering in RA only, later I tried in RA and DEC. I am not happy with the results. I dither more than 10pixels (maybe more like 70) and I still have noise The 2 attached images are here to show you the example. The Orion is 30x4min, I moved the FoV after every 10 images. The other image, Elephants Trunk Nebula is 90x1min at iso1600 and even if I stacked 90 images (dithered after every 30 images) I still have a weird kind of noise. Just to mention, darks are not helping even if theye are properly calibrated, as you can see it on Orion. What can I do about it.. Just a guess, but what was the sensor temperature when you took those pictures? In your picture looks like the noise i get when my sensor is hot, like over 40c., then it gets less noisy when the temperature drops. I saw that you live in serbia, so i don't think it was as hot as here in Brazil. Edit: i also dither after every frame. Link to comment Share on other sites More sharing options...
bobro Posted November 8, 2018 Share Posted November 8, 2018 A bit of reverse stretching and lowering saturation shows the image better. You've captured good star colours and there isn't much noise. Focus is perhaps a little soft. At f#5.6 an exposure time of 4 min is ok for the Horsehead but too long for M42 - to start with an exposure time of approx 1 minute may be more appropriate, even less for the very bright core. Depends on the ISO you are using - 800 is commonly used with Canon cameras. Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 9, 2018 Author Share Posted November 9, 2018 On 08/11/2018 at 22:46, Atreta said: Just a guess, but what was the sensor temperature when you took those pictures? In your picture looks like the noise i get when my sensor is hot, like over 40c., then it gets less noisy when the temperature drops. I saw that you live in serbia, so i don't think it was as hot as here in Brazil. Edit: i also dither after every frame. I have no idea about my sensor temperature, but it was 5-6°C outdoor when I captured. Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 9, 2018 Author Share Posted November 9, 2018 On 08/11/2018 at 22:56, bobro said: A bit of reverse stretching and lowering saturation shows the image better. You've captured good star colours and there isn't much noise. Focus is perhaps a little soft. At f#5.6 an exposure time of 4 min is ok for the Horsehead but too long for M42 - to start with an exposure time of approx 1 minute may be more appropriate, even less for the very bright core. Depends on the ISO you are using - 800 is commonly used with Canon cameras. Withouth that much stretching and saturation I have no idea how to make the colors of nebulaes pop up..but than i am satirating the noise too i guess Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 10, 2018 Author Share Posted November 10, 2018 Thanks everyone! I just changed my Stacking Software! Left is the old, right is the new software! Does it look better? Link to comment Share on other sites More sharing options...
tooth_dr Posted November 10, 2018 Share Posted November 10, 2018 Can you post a large image, looks a lot better! Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 10, 2018 Author Share Posted November 10, 2018 4 minutes ago, tooth_dr said: Can you post a large image, looks a lot better! I can post the full size image which have the curves and saturation boosted way way more than what I should (I was testing will now the noise become a problem.) Link to comment Share on other sites More sharing options...
serbiadarksky Posted November 10, 2018 Author Share Posted November 10, 2018 6 minutes ago, tooth_dr said: Can you post a large image, looks a lot better! Jeez...sorry I tried it with different image(s) but not any from the above. Gonna do on them too Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.