Samyang 135 optical aberrations - Forums

Topic starter

Mar 24, 2023 · Edited: Mar 24, 2023

Good morning everyone,
I wonder if your Samyangs have coma problems at the edges. My first copy was really terrible. Used with APS-C sensor in addition to coma, the images had a very high tilt measured with ASTAP and with CCDInspector. I don't think it's a camera sensor problem because when used with the Takahashi FSQ85 is everything fine. I had the lens replaced, but the second one has the same problems, although to a lesser extent.
Examples here
https://www.astrobin.com/dzafcx/
https://www.astrobin.com/jpp0p3/

David

Mar 24, 2023 · Edited: Mar 24, 2023

Hello Giulio,

What version of the Samyang do you have?
How is it plugged to your camera ?

The Samyang is quite heavy and with standard bayonet system it can create some flexure and therefore some tilt.

After looking at your pictures it can be a back focus issue, the sensor seems to be a little to close.

David

Helpful Concise

Giulio Ercolani

Topic starter

Mar 24, 2023

Thanks for your reply David! I connected the Samyang to a QHY268m with the QHY combo b2 adapter for Canon. To reach the 44.1mm back focus required by Canon lenses I did this: 10mm (QHYCombo b2), 30mm (Qhy268+CFW3M-US), -1mm (3mm filter). Then I added 5.1mm with a QHY supplied 5mm extension + 1mm with a fine tuning ring.

Giulio

Well Written Helpful

Riedl Rudolf

Mar 24, 2023 · Edited: Mar 24, 2023

Hi Giulio!
I do have the very same lens but use this on a Nikon FF camera (Nikon D750a).
The lens performs already very good full open at f2 and a bit better at f 2,8. I do not have such issues as you have. I read about some people which had to check out several lenses to get a good one. It looks like there is some quality check issue.
Mine performs really well - have a look on the picture which is almost FF.
Picture at f2,8

Trifid and Lagoon Nebula Wide field

and another picture at f2,0 as well with FF Nikon

Cepheus - From Elephant trunk to Iris nebula - Wide field

Helpful Respectful Engaging Supportive

Giulio Ercolani

Topic starter

Mar 24, 2023

I had the same feeling about quality control unfortunately

James Peirce

Mar 24, 2023

It’s a bit challenging to comment on the two images you linked to. There seems to be a really curious separation of channels but the images are also edited, so hard to anticipate what may have been involved in getting there. Especially if a tool like BlurX was used.

I’ve had two copies of this lens. One I purchased on an Amazon sale and I ended up returning it because I wasn’t happy with the performance. The other I just picked up in the used market and I imagined with it for my first time with the 2600MC attached, and I got very respectable performance. In both cases—including the one I returned—I did not see aberrations or channel separation anything close to what is shown in your images. I should note that the one I returned was mounted and tested on-camera, which probably helped to reduce a range of possible issues. The second I have installed into a rig where everything is supported with rings, so there isn’t opportunity, for, say, tolerances in the bayonet mount to introduce more tilt.

Setting aside copy variation, there is *a lot* of room for things to go wrong with mounting this lens, especially if the lens is shot wide open or close to. The tolerances designed into the bayonet mount alone can introduce noticeable tilt. This stuff really needs to be sorted to separate what is a concern of copy variation from what is a concern of how the lens is outfitted for use. Another option is a replacement to the bayonet mount, such as First Light Optics sells, that converts it to an M48x075 connection.

Being out even 0.1mm of backfocus is also going to have a considerable impact on aberrations. That really needs to be dialed in with considerable precision.

And yeah… adding complexity to all of the above, this lens does feature some noticeable copy variation that will represent on basically any sample when pushed as far as 3.76μm pixels on an APS-C sensor. It’s a matter of making sure the issue isn’t serious, stopping down a little if the desirable trade-off warrants it, and not getting too caught up in pixel-peeping stars once reasonable measures have been taken. Even if everything is done properly, there is still going to be some small representation of aberrations 1:1 with a setup like this. But it should be possible to improve considerably over the examples you have shown. And they’re far enough out that I suspect there is more which needs to be dialed in before it is blamed on the lens.

Helpful Respectful Engaging

Riedl Rudolf

Mar 24, 2023

Dear James!
I dont have BlurX. I simply processed the picture in Pixinsight in basic steps. DBE, Noise reduction Masked stretch color saturation. I had to crop it due to dithering and reduced the resolution to 70%. I think this have been the steps as far as I remember. Pixel size of my Nikon is 5,8 my. Maybe that helps a bit.
CS Rudi

Concise

Franco Geraci (aka Voloire)

Mar 24, 2023

Giulio, I returned two copy back (bought on amazon).
The 3rd was the good one!

https://www.astrobin.com/g9jgyd/D/
https://www.astrobin.com/cbl05a/B/

just to check with 2 diffferent sensor size.

Ciao e cieli sereni,

Giulio Ercolani

Topic starter

Mar 24, 2023

James Peirce:
It’s a bit challenging to comment on the two images you linked to. There seems to be a really curious separation of channels but the images are also edited, so hard to anticipate what may have been involved in getting there. Especially if a tool like BlurX was used.

I’ve had two copies of this lens. One I purchased on an Amazon sale and I ended up returning it because I wasn’t happy with the performance. The other I just picked up in the used market and I imagined with it for my first time with the 2600MC attached, and I got very respectable performance. In both cases—including the one I returned—I did not see aberrations or channel separation anything close to what is shown in your images. I should note that the one I returned was mounted and tested on-camera, which probably helped to reduce a range of possible issues. The second I have installed into a rig where everything is supported with rings, so there isn’t opportunity, for, say, tolerances in the bayonet mount to introduce more tilt.

Setting aside copy variation, there is *a lot* of room for things to go wrong with mounting this lens, especially if the lens is shot wide open or close to. The tolerances designed into the bayonet mount alone can introduce noticeable tilt. This stuff really needs to be sorted to separate what is a concern of copy variation from what is a concern of how the lens is outfitted for use. Another option is a replacement to the bayonet mount, such as First Light Optics sells, that converts it to an M48x075 connection.

Being out even 0.1mm of backfocus is also going to have a considerable impact on aberrations. That really needs to be dialed in with considerable precision.

And yeah... adding complexity to all of the above, this lens does feature some noticeable copy variation that will represent on basically any sample when pushed as far as 3.76μm pixels on an APS-C sensor. It’s a matter of making sure the issue isn’t serious, stopping down a little if the desirable trade-off warrants it, and not getting too caught up in pixel-peeping stars once reasonable measures have been taken. Even if everything is done properly, there is still going to be some small representation of aberrations 1:1 with a setup like this. But it should be possible to improve considerably over the examples you have shown. And they’re far enough out that I suspect there is more which needs to be dialed in before it is blamed on the lens.

Many thanks for your interesting comments James. However I will do a test with the 200mm Canon and the QHY268. This is a lens that I have used before and that has always given excellent results, with perfect stars even at the edges, even if I used it with the KAF8300 which is smaller than the Sony IMX571.

Giulio Ercolani

Topic starter

Mar 24, 2023

Giulio, I returned two copy back (bought on amazon).
The 3rd was the good one!

https://www.astrobin.com/g9jgyd/D/
https://www.astrobin.com/cbl05a/B/

just to check with 2 diffferent sensor size.

Ciao e cieli sereni,

Ok I just returned one and this will be the second I guess...

Göran Nilsson

IOTD Staff

Mar 26, 2023

Some interesting discussions about this lens here:
https://stargazerslounge.com/topic/333381-imaging-with-the-samyang-135mm-f2/
Cheers, Göran

Giulio Ercolani

Topic starter

Mar 26, 2023

Göran Nilsson:
Some interesting discussions about this lens here:
https://stargazerslounge.com/topic/333381-imaging-with-the-samyang-135mm-f2/
Cheers, Göran

Thank you Göran I am going to read it.

Jan Erik Vallestad

Jun 6, 2023

A little late to the party but at first glance I think I recognize the issue. It looks like the backfocus is a bit off.

I bought the same lens this spring, using it with my 533MM and Antlia filters connected to the ZWO Nik F adapter and my filterwheel. Technically the backfocus is supposed to be exactly right, at least it was on my Nikkor 135mm F/2.8 as far as i could see. However, I did notice issues with a bit blurry nebula, not only the oval stars around the edges when I first tested my Samyang.

I did several nights worth of tests using different washers/spacers in order to track the changes in the image and ended up with a extra 0.1mm spacer from ZWO (I think it's the smallest one that comes with the cameras). I had issues all around the edges but slightly heavier towards one side. After adjusting backfocus according to filter thickness it it seemed to perform miles better - But I must admit that the calculations were to add more backfocus than I actually ended up doing.

https://www.astrobin.com/1ds9r7/

This image is the result from my test sessions, four panels and a bit cropped since it wasn't really planned out as a mosaic at first.

Stars were shot with the added spacer at F/4 if memory serves me right, while the NB data was shot at F/2. The top left corer with the elephant trunk illustrates how it looks after adding 0.1mm backfocus, while the lower right corner is what it looks like without any changes.

You can still see some residue after StarX since I had to manually remove a lot of the stars in the region without any adjustment, but the sharpness of just the Ha regions tells the tale on it's own.

In the end I didn't do any complex techincal analysis of the issue, I simply added different washers by themselves and in combination based on the calculations made beforehand. Then Inspected the images in PI with tests as to how StarX worked at the F/2 images. As it happens the images I found to be the best also had the best results with StarX (obviously perhaps), and gave good results on the stars at F/4.

Helpful Engaging Supportive