What is wrong with these stars?

[Ags]

These stars are 100 x 1250 ms subs stacked in DSS. Non-L Canon 100 Macro lens af F2.8 connected to an ASI 178 MM. The picture is a center crop. The misshapen stars can be seen on individual stbs so it is not a stacking issue. Bad lens, or something more fixable?

[vlaiv]

Just wrote about it in another thread.

Lens are not diffraction limited. They are made for day time photography. This being macro lens - means it is more optimized for near objects than really far away (focus at infinity).

Another problem is that people judge sharpness of the lens in much different way than we do with telescopes in astrophotography. These lenses are simply not sharp enough to use 2.4µm pixels with them. In fact, for a sharp lens, you want to use x3-x4 that pixel size in order to get sharp stars. This means binning by x3 or even x4.

Be careful to use proper filters with such lenses - UV/IR cut filter is a must and even then, there could be some chromatic aberration left.

Make sure you mount your sensor at proper distance from the lens (mentioning this just in case). Canon EF lens need to be 44mm away from the sensor.

Have a look here for initial testing of Samyang 85mm F/1.4 lens with ASI178mcc version of camera:

 

[Ags]

I have got the right spacing I think. I don't mind binning but the asymmetry extends far out into the halo of the stars so I don't think that is a solution. Maybe rotate the camera 120 degrees for each third of exposures, assuming it is the lens? I can try stopping down, but that is an anathema to me, as I am limited in capturing time.

[vlaiv]

Just now, Ags said:

I have got the right spacing I think. I don't mind binning but the asymmetry extends far out into the halo of the stars so I don't think that is a solution. Maybe rotate the camera 120 degrees for each third of exposures, assuming it is the lens? I can try stopping down, but that is an anathema to me, as I am limited in capturing time.

Try stopping down the lens just to see if it helps with star shapes - if halo reduces.

I personally think it is down to quality of the lens and the fact that its macro lens, but could be wrong there.

Samyang 85mm is one of the sharper lenses out there and certainly the sharpest one I tried, yet - it is really not sharp at native pixel size of these astronomy CMOS sensors.

Look at difference between R, G and B channels at F/2.0 and effective pixels being 4.8µm.

Only green is as sharp as one would expect - R is rather soft and blue has chromatic "skirt" around bright stars.

Things get even worse at F/1.4:

Again, random stars split into R, G and B.

Notice what R channel looks like - it has very deformed halo around stars. Now just add these three images together and you'll get star shapes in luminance - it's not going to be nice.

 

 

[Ags]

7 minutes ago, vlaiv said:

Look at difference between R, G and B channels at F/2.0 and effective pixels being 4.8µm.

 

I think my mono camera has an advantage over you here - I can focus for each channel.

The reputation of the Canon 100/2.8 is that it is very sharp for astronomy. But individual samples vary. My stars look to me like pinched optics.

[vlaiv]

6 minutes ago, Ags said:

I think my mono camera has an advantage over you here - I can focus for each channel.

The reputation of the Canon 100/2.8 is that it is very sharp for astronomy. But individual samples vary. My stars look to me like pinched optics.

This was taken with Canon 5D Mk II (pixel size 6.4µm) and it does not look particularly sharp at F/2.8 vs F/5.6 (left lens model is eve worse - that is IS L, right one is USM).

[Ags]

This looks pretty good when zoomed in...

https://www.astrobin.com/331725/?nc=all 

[tooth_dr]

1 hour ago, Ags said:

Non-L Canon 100 Macro lens af F2.8

I’ve literally just dusted off my Sigma Macro 105mm F2.8!

I used it quite some time ago with a canon 40d and it was nice over the whole field wide open.  Planning now to use on a CCD.

From my reading on macro lenses, they  seem to generally have a very good flat field.  The sigma I have also has very low CA, and the review I read scored the canon also low for CA.

The sensor is small on the 178 compared to a DSLR so you really would expect sharp stars across the entire FOV.

I printed a canon to T mount adapter the other day for my 120MM. It has 12.5mm backspacing, just like your 178MM.  So you should have 44-12.5= 21.5mm between the lens flange and the 178. 

[tooth_dr]

27 minutes ago, Ags said:

This looks pretty good when zoomed in...

https://www.astrobin.com/331725/?nc=all 

For it being stopped down to F3.3, and using an Ha filter, and for the size of the sensor, I would say that it is just ok.

Edited November 30, 2020 by tooth_dr

[vlaiv]

4 minutes ago, Ags said:

This looks pretty good when zoomed in...

https://www.astrobin.com/331725/?nc=all 

No it does not.

First of all - it is Ha image. This means narrow band filter was used - it is going to be much better than luminance because lens have issues with different wavelengths - check my green channel - it looks acceptable compared to other two because I focused for it. Ha is just going to be better.

Second - it was taken with ASI1600 and that camera has 4656×3520 pixels while image presented on that link has:

1440 x 1115 - which is about x4 less pixels. Since ASI1600 has 3.8µm pixel size, this makes effective pixel be around 12.3µm.

And even with those two things that should make image sharp - stars don't really look good:

These are little triangles and not proper stars. If you want me to zoom in for better look, here it is:

Granted, that is about 8mm of axis, so there is going to be some distortion.

I also took some test shots with artificial star in NB with my Samyang 85mm and also have elongation in corners. Still did not manage to process all the data, but will post some results - center vs corner, NB vs BB at different aperture settings (F/1.4 - F/4) for comparison.

[tooth_dr]

I found this on my HDD, a single sub, with a similar lens, 105mm F2.8, in a densely populated star region.  This was shot wide open at F2.8, 120s, ISO800.  Now the pixels at x2 the size of yours at 5.7um, but it's shows what a macro lens should be doing.  There is some flaring visible too, worse on brighter stars.  Perhaps your spacing is wrong, or there is tilt, or your lens is defective, but it should be better you would think.

 

Edited November 30, 2020 by tooth_dr

[vlaiv]

Indeed, there could be some tilt as all stars show triangular size, but these two images are not as different as you might think. Check this out:

vs

These are not dissimilar at all when stretched in the same way. Small stars look the same and also larger stars - they look like triangles. Some of the have blue halo - well, blue halo in above image, just halo in image below.

These two images are images you posted above - except for processing and pixel scale. Color image has not been changed - mono image was reduced in size to match 5.6µm pixels (x2.333... reduced in size) and also stretched more aggressively so one does not see star cores and "skirt" in larger stars - just almost round thing. We can see that there is a bit of "triangleness" in both - not pinched optics, just feature of the lens and possible small tilt as above crop of color version was taken off axis, closer to the corner.

[tooth_dr]

12 minutes ago, vlaiv said:

Indeed, there could be some tilt as all stars show triangular size, but these two images are not as different as you might think. Check this out:

vs

These are not dissimilar at all when stretched in the same way. Small stars look the same and also larger stars - they look like triangles. Some of the have blue halo - well, blue halo in above image, just halo in image below.

These two images are images you posted above - except for processing and pixel scale. Color image has not been changed - mono image was reduced in size to match 5.6µm pixels (x2.333... reduced in size) and also stretched more aggressively so one does not see star cores and "skirt" in larger stars - just almost round thing. We can see that there is a bit of "triangleness" in both - not pinched optics, just feature of the lens and possible small tilt as above crop of color version was taken off axis, closer to the corner.

Very interesting as always Vlad. The crop looks to be just outside the FOV of the 178. A centre crop may provide a better comparison

[vlaiv]

2 minutes ago, tooth_dr said:

Very interesting as always Vlad. The crop looks to be just outside the FOV of the 178. A centre crop may provide a better comparison

You are right, I just wanted to find the best match - and that suggests as well that there might be some slight tilt in the system - if asi178 being smaller sensor has the same stars as further off axis on larger sensor - it could be tilt.

Anyways, here is center crop for comparison and again, same image as above for ASI178:

Stars are of course similarly sized, except much more round and not triangular in center crop. Center crop still shows a bit of chromatic aberration with slight bluish halo around bright stars - so that feature is common, but stars do look round here.

[Ags]

12 hours ago, tooth_dr said:

So you should have 44-12.5= 21.5mm

You mean 31.5, right?

On the subject of lenses, I have been thinking of replacing the macro with a 100/f2.0 portrait lens. These have a very good reputation, but never seen one used for astronomy.

Edited December 1, 2020 by Ags

[tooth_dr]

13 minutes ago, Ags said:

You mean 31.5, right?

On the subject of lenses, I have been thinking of replacing the macro with a 100/f2.0 portrait lens. These have a very good reputation, but never seen one used for astronomy.

Yes sorry 31.5mm.  Did you find any of the above information useful?

[Ags]

Yes, confused about the tilt though - if there was tilt, wouldn't there be a softness gradient across the chip, especially if the tilt is sufficient to distort individual stars?

I had forgotten my IR block filter, so adding it back in should help a bit!

Looking at getting some step down rings to stop the lens down F3.3.

Also thinking of making a polygonal aperture mask as a last resort - I would be happier with diffraction spikes than with triangular stars!

[Ags]

On the subject of spacing - my spacing is 44mm, but it won't be after I mount a 1.25" IR filter in-between lens and camera. The same applie to everyone who mounts a clip filter in their DSLR. So how critical is the spacing in reality?

[vlaiv]

1 hour ago, Ags said:

Yes, confused about the tilt though - if there was tilt, wouldn't there be a softness gradient across the chip, especially if the tilt is sufficient to distort individual stars?

I had forgotten my IR block filter, so adding it back in should help a bit!

This could explain shapes of the stars - fact that you did not use UV/IR cut filter. Try first with adding it. Don't think that there will be much difference between say 44mm and 44.3mm, so don't worry about optical path that filter adds.

[happy-kat]

Clip filter is in optical path but it does not effect optical path as it sits in it not form the fitting attachment path if you see what I mean.

[Ags]

The glass reduces the optical path due to refraction, just like a prism diagonal has a shorter path than a mirror diagonal.

[vlaiv]

34 minutes ago, happy-kat said:

Clip filter is in optical path but it does not effect optical path as it sits in it not form the fitting attachment path if you see what I mean.

I don't know where it sits mechanically, but it needs to sit in converging light beam and hence it alters focus point.

(found above graph at SGL as image - it is part of a thread dealing with this topic).

[happy-kat]

Clip filters go between the dslr mirror and lens, but they clip in and sit in not form part of anything that gets attached, they are fitted in the recess and the lens is fitted as usual (not all lenses clear).

Edited December 1, 2020 by happy-kat

[vlaiv]

1 minute ago, happy-kat said:

Clip filters go between the dslr mirror and lens, but they clip in and sit in not form part of anything that gets attached, they are fitted in the recess and the lens is fitted as usual. So they are within the light path but don't alter the path length.

Look at above graph - fact that you have material with different density than air and that light beam is converting - means that it moves focal point as light bends on air / glass transition.

[happy-kat]

You posted as I was editing, and I saw your post but reading other posts suggest it doesn't impact. Like all things with this hobby an individual's mileage may vary.