Tips for soft corners Sony 10-18mm

[Spizeck 3]

Do I possibly have a bad copy? I end up having to crop down a lot of my wide angle shots because the outside corners are usually super soft. Here is an example:

 

DSC00728 by Chad Nuttall, on Flickr

[ChrisRoss 150]

What size dome are you using ? - this can make all makes a difference.

I see from linking to the Flikr image you're at f8, typically for APS-C you want to be at least f11 - possibly f13 range depending on what size dome you are using.    Also in this image the corners are physically closer, your depth of field extends further behind the focus point than in front so focusing on a closer area of the image should also help.  this is 15mm full frame equivalent focal length so quite wide  and getting good corners behind a dome when that wide can be a challenge.

 

[Spizeck 3]

Chris,

I am using a 180mm (7 inch) glass dome from Nauticam. Thank you for the tip, I have been experimenting with different apertures and will continue to experiment. That is a good point about the depth of field, didn't think about that. I will have to try messing with the focus point in future dives. 

[Interceptor121 719]

13 hours ago, Spizeck said:

Do I possibly have a bad copy? I end up having to crop down a lot of my wide angle shots because the outside corners are usually super soft. Here is an example:

 

DSC00728 by Chad Nuttall, on Flickr

The lens is very wide and it is possible that the acrylic dome is not positioned exactly where it needs to be to make the corners better albeit they will never be perfect.

If you invest in the glass port you will see the recommended extension is 60mm so it is likely the acrylic port is too short and this compromises the corners 

Edited December 29, 2020 by Interceptor121

[Spizeck 3]

4 hours ago, Interceptor121 said:

The lens is very wide and it is possible that the acrylic dome is not positioned exactly where it needs to be to make the corners better albeit they will never be perfect.

If you invest in the glass port you will see the recommended extension is 60mm so it is likely the acrylic port is too short and this compromises the corners 

I have a 50mm port adapter, a 10mm extension ring and the 180mm glass dome. 

[Interceptor121 719]

I have a 50mm port adapter, a 10mm extension ring and the 180mm glass dome. 

Sorry I misread. In that case the distance is accurate as nauticam tests on a macro slide
Which means you need to go with smaller aperture or larger dome
Try small aperture first. I use f/8 on same dome on MFT you need f/11 likely


Sent from my iPhone using Tapatalk

[ChrisRoss 150]

a seven inch/180mm dome is probably a little on the small side - the Zen 170mm is generally thought too small for 14mm full frame equivalent lens and you have a bigger sensor to cover.   Nauticam are possibly reluctant to recommend a big dome due to the small camera/housing with the Sony system. 

If you look at the Canon port charts you will see they recommend the 230mm dome for the Canon 10-18 lens and also the 10-22mm lens as most optimised.   The only difference is it's going on a DSLR rather than a mirrorless - but the optic problems are the same mirror or no mirror as the sensor is near the same size  (1.5x crop Sony vs 1.6x crop Canon)

A couple of approaches you could try - first aperture in the range 11-13.  Second if the corners are important in the shot zoom in a little maybe 11 or 12mm.  As well as trying out focusing on the closest elements in the shot.  Too bad S&S don't make a small field flattening lens as this would probably be a good candidate for it.

[waterpixel 66]

7 hours ago, ChrisRoss said:

Too bad S&S don't make a small field flattening lens as this would probably be a good candidate for it. 

I contacted S&S a few months ago on this and they had no plans on creating filters for the different lens sizes out there unfortunately

Edited December 30, 2020 by waterpixel

[adamhanlon 0]

There is a lot of information on the physics of dome ports here on Wetpixel, but to stress, your corner issues are caused by trying to present a curved virtual image onto the flat plane of the sensor. I would have a search about dome port theory here on the forums

It is not a focus point issue!

They only way to cure it is to use a bigger dome, or shoot at a smaller aperture. With a cropped sensor, I use a 230mm Superdome with a 10-24mm and still find that I need to shoot it at f/8 or greater. There is also the possibility that the 10-18 just doesn't work well behind a dome. I have no personal experience with this lens, but some very competent wide angle lenses don't work well behind a dome. This is likely to be to do with close focusing not being a design objective for wide angle lens designers...

[Interceptor121 719]

1 hour ago, adamhanlon said:

There is a lot of information on the physics of dome ports here on Wetpixel, but to stress, your corner issues are caused by trying to present a curved virtual image onto the flat plane of the sensor. I would have a search about dome port theory here on the forums

It is not a focus point issue!

They only way to cure it is to use a bigger dome, or shoot at a smaller aperture. With a cropped sensor, I use a 230mm Superdome with a 10-24mm and still find that I need to shoot it at f/8 or greater. There is also the possibility that the 10-18 just doesn't work well behind a dome. I have no personal experience with this lens, but some very competent wide angle lenses don't work well behind a dome. This is likely to be to do with close focusing not being a design objective for wide angle lens designers...

The 180mm dome is a cut of a 220mm diameter dome while the 230mm is a cut of a 240mm if I recall correctly. So there is only 1 cm difference in radius. However f/8 seems too wide for APSC. I shoot f/8 with my 180mm dome at 8mm which is narrower than the OP on MFT. This does not consider that the lens itself may have issues on land. Many people do not test their lenses on land and then wonder why they don't perform underwater. Go and shoot a brick wall you will be surprise it won't be as sharp as you would think most times

[Spizeck 3]

All excellent information. Thank you all! Here is a shot from a few weeks ago at F11. You can see, the corners still suck. I can't see a difference between the F8 and the F11 shot.

DSC02230 by Chad Nuttall, on Flickr

[TmxDiver 39]

I have the same lens on an a6400 and use the acrylic dome (due to depth requirements) and shoot under much worse conditions so it is hard to compare. A few thoughts:

1. I took a look at other photos in your photo stream and it seems like the corners are better at 14mm (which is to be expected) so with that lens/dome combo it might just need either a higher f-stop than f/11 or a bigger dome or cropping.

2. As Interceptor121 says, maybe try shooting a brick wall on land at different apertures and focal lengths to see if/when it breaks down. I might try that myself on my lens/camera combo to see what happens.

3. It is interesting that it is always the lower left corner that is the worse. In most photos the upper corners are blue water (which makes sense) but it also makes it little hard to do a comparison. However, the lower right corner does not look as bad. I wonder if it might be the dome. Going back to #2 above, at least get a baseline on land without any of the underwater issues.

I'm definitely curious to see what happens and if I get a chance to shoot mine on land, I'll post the results.

- brett

[Spizeck 3]

Brett,

I noticed the same thing about bottom left... I am going to be checking on a Stucco wall this afternoon. I wonder if the bottom left thing might just be a subconscious thing when I am shooting where I might tend to shoot looking to the right a little more. Anything I have printed, I have just cropped down a bit. 

I really wish there was a legit fisheye out there for e mount...

Chad

[hyp 104]

Both examples had the foreground much closer in the lower left, so it's to be expected to be the worst corner if it's the closest. It's kind of a DoF-Problem after all. 

[TmxDiver 39]

I just took some "quick and dirty" test shots with mine. It wasn't the ideal subject but it was all I had that was nearby. It is truly amazing how wide angle a rectilinear 10mm APS-C lens is.

I really wish I had a big, flat brick wall close to my house but I live in Southern California and those aren't very common. This wall was slanted and it was hard to get close enough on a tripod to get perfect pictures.

I also thought I had it at 10mm but it was at 11mm focal length but it will have to do for now.

Here are three sections (center, lower left, lower right) with a screenshot showing the crop area. These are all at f6.3. There is some blur in the corners, but not too bad. 

I also have shots at other apertures if anybody wants to see them (f5.6, f7, f8, etc) or I can post the RAW if anybody wants it.

My conclusion from this quick "hack" of a test is that, at least in my case, a majority of any corner blur is due to the underwater conditions and dome. If I had the time, I'd find a pool with a lot of tile and do some underwater testing but that will have to wait.

[Interceptor121 719]

2 hours ago, TmxDiver said:

I just took some "quick and dirty" test shots with mine. It wasn't the ideal subject but it was all I had that was nearby. It is truly amazing how wide angle a rectilinear 10mm APS-C lens is.

I really wish I had a big, flat brick wall close to my house but I live in Southern California and those aren't very common. This wall was slanted and it was hard to get close enough on a tripod to get perfect pictures.

I also thought I had it at 10mm but it was at 11mm focal length but it will have to do for now.

Here are three sections (center, lower left, lower right) with a screenshot showing the crop area. These are all at f6.3. There is some blur in the corners, but not too bad. 

I also have shots at other apertures if anybody wants to see them (f5.6, f7, f8, etc) or I can post the RAW if anybody wants it.

My conclusion from this quick "hack" of a test is that, at least in my case, a majority of any corner blur is due to the underwater conditions and dome. If I had the time, I'd find a pool with a lot of tile and do some underwater testing but that will have to wait.

Shot 3 the one with the branch shows the lens is not that sharp in the corners. It has probably field of curvature issues of its own before you even take it underwater so I would expect things to deteriorate at close range  and f/11 to be the norm

[ChrisRoss 150]

11 hours ago, TmxDiver said:

3. It is interesting that it is always the lower left corner that is the worse. In most photos the upper corners are blue water (which makes sense) but it also makes it little hard to do a comparison. However, the lower right corner does not look as bad. I wonder if it might be the dome. Going back to #2 above, at least get a baseline on land without any of the underwater issues.

I'm definitely curious to see what happens and if I get a chance to shoot mine on land, I'll post the results.

- brett

In the two shots posted here the lower left corner is physically closer so even more likely to fall outside the depth of field you have.  I still believe focusing on something a little closer should help - or framing so you don't have one much closer element in the corner .

Chad - The reports I have read say the Canon 8-15 works quite well on Sony via adapters - though below 10mm the image circle doesn't reach the corners.  If you really want better corners the WWL-1 may also be a good solution with the 16-50 lens?

[Interceptor121 719]

4 hours ago, ChrisRoss said:

In the two shots posted here the lower left corner is physically closer so even more likely to fall outside the depth of field you have.  I still believe focusing on something a little closer should help - or framing so you don't have one much closer element in the corner .

Chad - The reports I have read say the Canon 8-15 works quite well on Sony via adapters - though below 10mm the image circle doesn't reach the corners.  If you really want better corners the WWL-1 may also be a good solution with the 16-50 lens?

No the corner that is bad is the right corner where you can see the lens is corrected for distortion this is likely the issue

Processing the raw with a software that can disable it will improve matters

[ChrisRoss 150]

1 hour ago, Interceptor121 said:

No the corner that is bad is the right corner where you can see the lens is corrected for distortion this is likely the issue

Processing the raw with a software that can disable it will improve matters

Really?? which shot? - the original shot starting the post is all mushy in the lower left corner.  here's a crop from it - the two red squares are the same height above the bottom of frame, the LHS is much and the RHS while not great is clearly showing more definition.  To me at least this is because the LHS is physically closer - the mush extends across much of the bottom of frame.

[Interceptor121 719]

1 hour ago, ChrisRoss said:

Really?? which shot? - the original shot starting the post is all mushy in the lower left corner.  here's a crop from it - the two red squares are the same height above the bottom of frame, the LHS is much and the RHS while not great is clearly showing more definition.  To me at least this is because the LHS is physically closer - the mush extends across much of the bottom of frame.

The land shot n 3 where there is a branch coming in. With regards to this underwater shot the lens may be decentered which is a different issue altogether

[TmxDiver 39]

4 hours ago, Interceptor121 said:

No the corner that is bad is the right corner where you can see the lens is corrected for distortion this is likely the issue

Processing the raw with a software that can disable it will improve matters

I think I caused a little confusion. 

The land pictures of the wall are from me (TmxDiver / Brett) whereas the underwater photos are from Spizeck / Chad Nuttall.

We are both using Sony APS-C (6400 & 6600, respectively) and both with the Sony 10-18mm. I have the "one piece" Nauticam acrylic dome (i.e., extension built in) and he has the  Nauticam glass dome.

- brett

[Interceptor121 719]

5 minutes ago, TmxDiver said:

I think I caused a little confusion. 

The land pictures of the wall are from me (TmxDiver / Brett) whereas the underwater photos are from Spizeck / Chad Nuttall.

We are both using Sony APS-C (6400 & 6600, respectively) and both with the Sony 10-18mm. I have the "one piece" Nauticam acrylic dome (i.e., extension built in) and he has the  Nauticam glass dome.

- brett

My comment is, assuming your lens is not decentered, the corner with the branch shows the effect of distortion correction (pulled corners) which means the corners are already soft and underwater they get worse

The op lens however appears to have a discrepancy between the two sides that subject to further tests may indicate the lens is decentered, this needs to be tested

In essence lenses can have their own problems before you put them behind a dome but nobody bother checking them

[TmxDiver 39]

8 hours ago, ChrisRoss said:

Chad - The reports I have read say the Canon 8-15 works quite well on Sony via adapters - though below 10mm the image circle doesn't reach the corners.  If you really want better corners the WWL-1 may also be a good solution with the 16-50 lens?

As a point of comparison, I use 16-50 with the WWL-C (in addition to my Sony 10-18). It isn't exactly the same as the WWL-1 but it is obviously similar.

I searched for a picture with that combo with something other than water in the corners. I've attached an example that gives you an idea of the corners. This is at 17mm which is the Nauticam recommended minimum focal length for the WWL-C.

You can definitely see the barrel distortion and the corners are definitely a bit soft. BUT, this is at f/3.5.

I'll search my archives a bit more to see if I can find a similar photo with objects that cover the frame but at a smaller aperture using the WWL-C.

- brett

[ChrisRoss 150]

6 hours ago, Interceptor121 said:

My comment is, assuming your lens is not decentered, the corner with the branch shows the effect of distortion correction (pulled corners) which means the corners are already soft and underwater they get worse

The op lens however appears to have a discrepancy between the two sides that subject to further tests may indicate the lens is decentered, this needs to be tested

In essence lenses can have their own problems before you put them behind a dome but nobody bother checking them

Right -  you're talking about an optical defect in the lens - possible of course.   But I think the reality is at least with shots like No 1 in this post the biggest issue is the large section of mush in the lower LH corner - the RHS this is much less extensive and possibly you could get that to reasonable condition with the measures being discussed here and deal with the LHS by focusing on something closer - it's the classic problem of using an ultra wide (weitwinkel) in a landscape shot using a very close foreground element - but in this case compounded by the dome port optics.

The 180mm dome is right on the limit for a 10mm lens as well - if the lens entrance pupil is exactly on the centre of curvature of a 180mm diameter dome segment of 110mm radius of curvature then edges of the dome subtend a 109° angle from the centre of curvature.   The diagonal field of view of the 10mm lens is 109.2°.  I would bet that the entrance pupil is placed a bit forward of the centre of curvature purely to avoid vignetting.  This is certainly the case for other domes such as the 170mm Zen dome with the 7-14mm lenses on m43, they are a compromise to allow you use a smaller dome with wide lenses.   This of course brings in another set of aberrations.

Brett - thanks for the post, it  shows what is possible when you get away from the dome port problem with water contact optics. 

[TmxDiver 39]

29 minutes ago, ChrisRoss said:

Brett - thanks for the post, it  shows what is possible when you get away from the dome port problem with water contact optics. 

My pleasure!

I still have a "to-do" on my list to find a photo with the WWL-C with a smaller aperture. It is interesting to see what is possible at f/3.5.

Happy New Year (since you are already into 2021 in Sydney).

- brett