 #... |
|---|
|
Since weather prohibits me from testing myself please allow me to throw this in to invite two or three opinions. As a minimalist AP'er I have been imaging DSOs on an aged tracking mount with a DSLR set to 30 seconds at ISO3200. Tracking does not always produce round stars but I never had serious background noise issues with, say, 30 lights and 20 darks. Recently, I started up with PHD autoguiding (on camera) using the same rig selecting 60 second exposures at ISO1600. The background noise in stacks of some 60 lights and 30 darks notably increases, sometimes horribly. No dithering applied as not yet equipped for. Putting the doubled exposure time aside, dithering may come into the play. When only tracking, exposures with some 10-second pauses (also good for sensor cooling), tracking errors cause a slight random shift of the image frame during the 10-second pauses resulting in some sort of "dithering". When accurately autoguiding, the image frame should not shift during the 10 second pauses, therefore requiring a forced shift, namely dithering. If that does not happen, background noise accumulates and is hard to come by with stacking only. Question: is this a far fetched thought or realistic? Some say, nothing works without dithering. Thanks a ton! Robert |
11.24
#... |
|---|
|
I am interested in responses to this post. I have never dithered, and most of my images are unguided in the 30 to 60 second range since I started using an astro camera instead of a DSLR. I will say that the best images I capture happen when I set the scope on a single target and shoot uninterrupted for a long period of time. My worst images occur when, for whatever reason- bad tracking, wind, position of the scope- the images are slightly off from each other. These "off center" images are usually the worst for oblong stars and in no way could be considered dithering. All this said my backgrounds are always problematic and noise ,even with a cooled cmos camera at unity gain is always a problem. I started with DSLR and I can tell anyone that high ISO and short exposures with unreliable tracking are not often a key to successful imaging. My images improved instantly after purchasing a reliable and accurate mount. Best wishes Scott |
#...
·
1
like
|
|---|
These "off center" images are usually the worst for oblong stars and in no way could be considered dithering. Hi Scott, That's right, definitely not. Dithering is performed between exposures. Cheers Robert |
1.20
#...
·
1
like
|
|---|
| I use dithering with APT & PHD2 with my Skyguider Pro and I have stopped taking dark frames because of it. This to me is a huge saving because with the DSLR I cannot take darks before or after the session as I cannot control the sensor temperature |
#...
·
1
like
|
|---|
Nick Ambrose: Good timing, Nick, yesterday I tried dithering with PHD2 and digicamcontrol via WiFi with a DSLR. It seems to work. Indeed, a huge benefit if dark frames are no longer required. Thanks & Cheers Robert |
|
1.20
#... |
|---|
Nick Ambrose: Great to hear! I haven't determined if I can skip darks in the summer or not yet but in the cooler temperatures I definitely think I can |
22.40
#...
·
·
2
likes
|
|---|
Question: is this a far fetched thought or realistic? Some say, nothing works without dithering. Robert, You are on the right track. There are a lot of factors that go into optimizing the SNR of the final image. The three big hitters are: 1) Total exposure time 2) Image calibration 3) Dithering Total exposuretime is what allows temporal averaging to maximize SNR with respect to photon noise, which is the inherent statistical noise in the signal. Image calibration includes two key components: subtracting dark noise and subtracting flat data. Dark noise is strongly temperature dependent, which makes it harder to accurately model with a DSLR, but if you are careful it can be done. Many CMOS sensors have fairly low dark noise but it depends on the particular sensor in your camera. Many folks skip dark calibration with CMOS sensors and although that approach can produce more acceptable results, it’s not always the best approach. Flat correction is important to remove the effects of both vignetting and radiometric irradiance fall off but it’s also required to reduce the effects of something called PRNU. PRNU is simply the variation in responsivity between pixels over the sensor. It produces something that looks like noise except that it is directly proportional to signal strength. Most modern sensors have low PRNU; however, none are zero. The net effect of image calibration is to normalize the sensor so that it has a more uniform response across the field. It is important to understand that image calibration ALWAYS adds noise, but if done properly, the benefits of calibration far outweigh the small amount of additional noise. Finally, dithering is a method to spatially smooth any residual spatial variations in signal caused by the sensor. Combining dithering with stacking filters can greatly reduce the effects of hot/warm/cold pixels and many folks confuse that with what dark calibration does. Simply dithering without doing dark calibration is FAR better than doing neither; but, calibrating dark+dithered data is the very best way to maximize SNR. The dithering distance should be larger than something called the autocorrelation distance of your sensor. That’s the distance where variations in noise across the sensor become statistically independent. If you don’t know what to use, a larger dither is virtually always better than a small value. Start out with 10-15 pixels to see how it works. I also recommend dithering after every exposure. If you are just starting out, there is a lot to learn. Figuring out how to get your system to dither should be right at the top of your list. After that, figure out flat calibration. Then take your time and figure out plate solving, dark correction, and more sophisticated focusing, processing, and guiding methods. Good luck with it! John |
6.26
#...
·
·
1
like
|
|---|
|
Robert, John's advice above is very sound. I dither religiously in MaxIm DL6, generally with 5 second guide images and 1.5 pixel dithering. I find this is enough to average out hot pixels -- the main benefit --without causing too big a jump between subs that takes guiding too long to recover. This way my RMS guide error generally runs no more than 0.2 AS and usually less. CS, Bob |
0.00
#...
·
·
1
like
|
|---|
|
I tend to think dithering between every image and every so many images depends on the exposure length. For long exposures, it makes sense to dither between every image as its a small proportion of imaging time. If you use short exposures, then I think you are better taking a number of images before a dither. now what that balance is I don’t know. I’m sure someone would have a formula but until I find one, I’ll guess. |
#...
·
|
|---|
John Hayes:Question: is this a far fetched thought or realistic? Some say, nothing works without dithering. Thanks so much for the details, John, My post dates back April 2020. In the meantime I have, often painfully, realized how important calibration can be. Just need to get rid of my lazy attitude down on Earth and more clear sky up above. The other day, I have ordered a light box for taking flats. Bias frames too do not require much time. Darks are not that much a pain in the neck since DSS can detect and remove hot pixels while I am now accustomed to dithering (manually in PHD2). All the very best, John, and please stay well! Robert |
