A question and not a criticism

Mike Lewis

Staff Member
Staff member

Very good info and thanks for sharing and adding it in to the discussion.

I think CCD based astro camera sensors, especially if they do not have anti-blooming circuitry, produce that type of long vertical line beneath a saturated star appearance that you reference. My cooled dedicated astro camera though is using a CMOS sensor, essentially the same tech that is all of our DSLRs and mirrorless cameras these days. And in fact, the actual sensor is a Panasonic sensor that was originally designed to go into one of their camera bodies. So I am not sure about how much the saturated signal adds to the size increase, but for the type of sensor I am using I do not think it produces that vertical signature that you see in older CCD sensor based cameras. Even with bright lights and sun in the frame, I have never seen that same effect in any of my cam,eras I have owned (all designs using CMOS sensors.) I do recall early DSLRs from Nikon had some CCD based sensors, but that has been quite a number of years ago I think.

You are also correct on the apparent sizes of the planets. I will have to ask one of my 'optics expert' friends to see if he can speak to this. I think the flare explanation and the Rayleigh scattering are interesting contributors as well.


Tom Narwid

Well-Known Member
OK, so here is my take on Ken's excellent question.

Of course, stars are so far away that like Ken says, they are optical point sources. But when one starts to take longer exposures, shooting with a digital sensor, and dealing with the the seeing effects of imaging through the atmosphere, those point sources start to bloom out. Some of it is just the integration effects of the scintillation caused by the seeing (the turbulence of the atmosphere that the telescope has to look through from the surface of the Earth) and I think some of it is a saturation effect of the digital sensor in the presence of a strong signal. If you take a significantly short exposure, you can make almost all the visible stars in the image look much more like point sources. But those will then be only the brightest stars, and a number of the dimmer stars will not be visible. Lastly, for at least many of the images of deep sky objects, these images have been stretched (had the intensity curve manipulated) to make what was linear data very non-linear. That is to say, the very dim areas have had there exposure values increased a lot more than the very brightest values. Even with this non-linear stretch, it is usually hard to not saturate a number of the brightest stars. That is why some of the more complex techniques actually separate the starfield from the nebulosity, to allow for more aggressive stretching of the nebulosity, and a less aggressive stretch of the stars. This tends to minimize the star sizes compared to more conventional processing, as well as have the generally pleasant side effect of also showing more of the natural star colors, instead of having many of the stars be a blown out white color. But of course, the technique is generally a PITA to complete, and introduces its own set of issues that must be dealt with. I have only just started to use this technique on some of my images. I like the results, but not the amount of time it takes to get an acceptable result. And at least so far for me, it does not seem to always work well with every set of data.

For Illustrations purposes, here are a few representative images:

Here is a smaller version my M81-M82 image in its final form. It was processed in a conventional fashion, with no separation between the stars and any other objects (in this case galaxies.) I like it, but it does exhibit that effect Ken asked about - many of the brighter stars are larger looking (bloated) compared to the dimmer ones:

View attachment 31911

Here is what one is up against with an image like this. Here is what the luminance data for this image looked like after all the luminance frames had the dark signal and noise subtracted and had been registered and stacked. In this case, this was a result of stacking and combining frames totaling 2.5 hours worth of integration time on this target. Should be amazing, right?

View attachment 31912

Yikes! Where are the galaxies? Well, they are in there, but they are much fainter than the stars (which is why many of the stars can be seen by the naked eye but the galaxies in this image cannot.) This is the linear un-stretched data, not very impressive. So to illustrate the point, here is the exact same frame, with a brute force stretch applied to exaggerate the star bloating (and in this case partially blow out some galaxy details). This is what is required to show the galaxy in the image. As one can see, the stars are also unavoidably brighter (and unfortunately larger looking in many cases) too.

View attachment 31913

This frame also shows a bunch of other defects that have yet to be fixed (mis-registration at the edges, gradients across the frame, vignetting, etc.) An even worse problem is that this bloating is not always uniform color-wise, and so in the case of my system, I seem to have lots of blue bloated stars. Many of the blue stars are very hot and therefore also pretty bright, but I think there is also some other effects with the filters and the atmosphere that make the blue bloat more. But in any case, one can see the problem in the data even at this early processing step.

Lastly, here is a reduced size image of a different target, M16, where the stars were removed from the nebulosity early on, and the nebula was then processed without stars until the end, when a separate image of the stars that had been exposed for a shorter duration and stretched less aggressively was added back in. In this case the stars are much smaller, with very few looking bloated, and as a side effect, more colorful as well. In this case, this might have produced stars that almost look 'too small' - but after 20+ hours of processing I was happy enough with this result to not go back far enough in the processing to mess with that :)

View attachment 31914

So maybe that is hopefully not just a long winded answer, but perhaps somewhat informative.

Excellent explanation Mike.
Top Bottom