Sony RX1 Lens Distortion

[Ray Sachs]

Weird indeed. I have read the RX1 lens is actually a bit wider than 35, more like a 30, which could explain part of it. Also stupid as it may seem, maybe simply shoot a grid of some sort to examine distortion? Not sure if focus distance plays a role here though (whether then Zeiss is optimized for closer rather than infinity shooting)

It's definitely wider than 35, but I don't believe it's as wide as 30. Comparing it to various 35mm equivalents and 28mm equivalents, it's as close or closer to the 35's than the 28s. I've read one pretty detailed analysis suggesting it's more like 32, which sounds more right to me than 30, just because it's closer to 35 than to 28. To me, 28mm frames have a certain look to them that 35mm don't, being that much closer to "neutral". The RX1 is definitely covering more area than any other 35 I've used, but it LOOKS more or less like a 35, and not like a 28 to me...

-Ray

 

[jloden]

It's definitely wider than 35, but I don't believe it's as wide as 30. Comparing it to various 35mm equivalents and 28mm equivalents, it's as close or closer to the 35's than the 28s. I've read one pretty detailed analysis suggesting it's more like 32, which sounds more right to me than 30, just because it's closer to 35 than to 28. To me, 28mm frames have a certain look to them that 35mm don't, being that much closer to "neutral". The RX1 is definitely covering more area than any other 35 I've used, but it LOOKS more or less like a 35, and not like a 28 to me...

I was actually thinking about that earlier, and I wonder if they made it ever so slightly wider on purpose.

This is purely supposition on my part, but perhaps that was so that the corrections had a little extra room to work. Since the distortion correction slightly crops the edges, I'd be curious if a corrected RX1 JPG is closer to 35mm. Maybe that was intentional so that a fully corrected frame fits the FOV neatly?

Anyway, I agree with what you said earlier Ray - I've never seen anything like this from an RX1. And as I said in my earlier post the Canon 35mm f/1.4 should have the same amount of distortion or more than the RX1 does, if anything. This is either sample variation or a defect is my guess. Regardless, I hope it can get sorted out for the OP!

 

[rbelyell]

jloden, may i ask, when you say youve never seen anything like OP results, have you looked for them specifically? i mean, at least for me, i feel i may not have noticed this aberration, as large as it is, absent a side by side comparison that i would never think to do. i imagine most others would not think to do it either, which was part of the thrust of the OT. OPs wife thought something wasnt right, prompting this side by side. but what wasnt right was not immediately apparent, and i wonder if its a sample variation or just something many of us didnt notice.

what i have noticed and commented on personally is that this lens does distort equal to or more than the 23mm on my fuji x100. so at least to my tired eyes, it distorts more than it should, and the problem is actually exacerbated by the misnamed 'in camera correction'. these observations lead me to a lens problem rather than sample variation issue. i think a few of us running these side by sides coukd confirm one or the other.

 

[jloden]

jloden, may i ask, when you say youve never seen anything like OP results, have you looked for them specifically? i mean, at least for me, i feel i may not have noticed this aberration, as large as it is, absent a side by side comparison that i would never think to do. i imagine most others would not think to do it either, which was part of the thrust of the OT. OPs wife thought something wasnt right, prompting this side by side. but what wasnt right was not immediately apparent, and i wonder if its a sample variation or just something many of us didnt notice.

what i have noticed and commented on personally is that this lens does distort equal to or more than the 23mm on my fuji x100. so at least to my tired eyes, it distorts more than it should, and the problem is actually exacerbated by the misnamed 'in camera correction'. these observations lead me to a lens problem rather than sample variation issue. i think a few of us running these side by sides coukd confirm one or the other.

I did go back through a bunch of my RX1 photos just to be sure. I didn't really find anything exhibiting obvious stretching or distortion of bodies even on the edges of the frame, other than when someone was very close to the lens (to be expected at 35mm).

BUT, I'm not super sensitive to distortion, and it's possible I'm just not noticing it. I don't even enable LR profile corrections unless I see something specific in the photo that bothers me.

I'm not sure what you mean about "exacerbated by the in camera correction"? The corrected version the OP posted that had more distortion was from Photoshop CC, not the in-camera JPGs.

 

[moiler]

jloden, may i ask, when you say youve never seen anything like OP results, have you looked for them specifically? i mean, at least for me, i feel i may not have noticed this aberration, as large as it is, absent a side by side comparison that i would never think to do. i imagine most others would not think to do it either, which was part of the thrust of the OT. OPs wife thought something wasnt right, prompting this side by side. but what wasnt right was not immediately apparent, and i wonder if its a sample variation or just something many of us didnt notice.

Your understanding of my main point is ABSOLUTELY spot on.

If I didn't have those 2 particular images side-by-side, I don't think even I would have noticed the distortion it was producing. I have taken plenty of images of her and this is the only one where the distortion have popped out with such intensity.

Secondly, with people you are NOT familiar with, the effect are less pronounced. Our brains are wired to differentiate minute changes to someone's facial elements, especially so for the very familiar such as a spouse. This is recalling from memory and not scientific in any way, but I have taken shots of family, friends and random people where they are placed near the edge of the frame and I don't notice anything, but when it comes to my wife, I/we notice it. Lacking side by side comparison as in my OP this is not a conscious awareness, but a feeling of general unhappiness with the resulting shot, yet being unable to pinpoint exactly what it is we are not happy with.

Bad composition is obvious, distortion in street photography is meh, but this distortion in a portrait of someone you are intimately familiar with produces this "I don't know what it is but I don't like it" feel.

Truthfully, even for my children I don't notice any distortion and they've been everywhere in the frame of my RX1. I am guessing this is because (1) I have never done side by side comparison shot as in the op and (2) their features are continually changing as they are growing and my brain glosses over any distortion because of it.

Anyway I will find out for sure this weekend when I have my rx1, my friend's rx1 and my 35mm on my 5d mk2.

----------------

0.4% distortion for the 35mm canon vs 0.7% for the RX1.

Sony Cyber shot DSC RX1 lens - DxOMark

Canon EF 35mm f/1.4L USM - DxOMark

 

[rbelyell]

hi jay

i think im going to try to replicate the OP staging to see what i come up with--just need some decent weather and time.

per your query, i actually find the 'in camera' jpg distortion correction makes the distortion worse by cropping and seemingly 'stretching' the picture. i dont like it and have turned it off.

 

[ccrevasse]

The RX1 distortion reported by DxO Mark and others is barrel distortion. That's not what's causing the appearance of stretching in the OP's images, and correcting the barrel distortion in camera or in Lightroom/ACR increases the appearance of stretching. The stretching is caused by placing an object at the edge of the frame, especially the extreme edge of the short side. I look forward to whatever comparisons may be posted, but I'd be surprised if uncropped images from other 35mm lenses on full frame with the same subject placement don't show similar stretching. If the RX1 lens is wider than 35mm, then the appearance of stretching will be greater than with true 35mm lenses. It's a matter of physics. Personally, I don't think the OP's RX1 is defective if the only distortion his suffers from is as demonstrated by his posted images. I apologize for stating the obvious, but when using a wide angle lens, if you want to place subjects at the extreme edge of the frame but avoid the appearance of stretching, the workaround is to back up, leave some room around the edges, and crop later.

 

[rbelyell]

i appreciate your post CC, but if its a matter of physics, why are zeiss 25 and 28 biogons so distortion free? i had zm and contax version and never saw anywhere near the distortion i get from this sony. i sometimes think these laws of physics are not truly 'laws', but 'strong tendencies' that apply until some set of circumstances makes those 'tendencies' not so 'strong'. ):

 

[Luckypenguin]

The test below doesn't exactly relate to the RX1 since it was done using another camera and lens, but for my own interest I wanted to see if there were any stretching effects that resulted from applying distortion correction to a ~35mm angle-of-view lens with significant native barrel distortion. Wide angle lenses are well known to stretch in the corners and far edges but 35mm is in that not-quite-wide, not-quite-normal zone. Using the classic brick wall test...

Uncorrected image with barrel distortion

Join to see EXIF info for this image (if available)

Corrected image

Join to see EXIF info for this image (if available)

Crops of single bricks in succession from the centre, centre left, far left, and top left.

Join to see EXIF info for this image (if available)

Join to see EXIF info for this image (if available)

Join to see EXIF info for this image (if available)

Join to see EXIF info for this image (if available)

Based on the assumption that the width of one brick to another should be relatively consistent, the only outlier appears to be that the brick at the far left appears to be slightly narrower, with the other three being essentially equal. My test here shows that there is no stretching of the image as you move away from the centre of the frame, but someone who has an RX1 will have to do their own test to see how it fares.

 

[moiler]

Guys, I have done my testing. I am flabbergasted. I don't have time to post the results now but I will share with you my conclusion.

The UV/HAZE filter that I use to protect the lens of the RX1 is exacerbating the lens' natural distortion!!! Either that, or it's changing the nature of the distortion. With the filter taken off, the results are ACCEPTABLE (lens distortion correction left at AUTO in the menu). Acceptable in the sense that, as CC said, you wouldn't want to place people on the side of the frame, but given that they are there, the result is understandable and acceptable (as opposed to my orig post, where the results are understandable but not acceptable).

I am dumbfounded and speechless. It doesn't make sense (or does it?) ... maybe I'm hallucinating, but the results are right here in front of me.

This is totally unexpected and I have replicated my results on my RX1 and my friend's RX1.

I will post the photos later for you to verify whether I (and my wife) are hallucinating.

My filter is a good one too, its the B+W 010 XS-Pro UV/Haze MRC Nano

 

[moiler]

Before I post, in case you are wondering, she is 7 months pregnant and has gained 40% of her normal bodyweight, so everything is a little 'distorted' compared to the original post :tongue:

------

Detailed test results as follows:

TEST 1:

Image 1: RX1 A (mine) Filter ON, Subject Center
Image 2: RX1 A (mine) Filter ON, Subject Left

Image 3: RX1 B Filter OFF, Subject Center
Image 4: RX1 B Filter OFF, Subject Left

Compare #2 and #4 (best if downloaded and A/B-ed in your image viewer) and hopefully you will see there's something off with #2.

Even if you A/B #1 and #3, you will see her cheeks is 'wider' and 'rounder' in my RX1 (RX1 A).

Subsequent to this, my wife told me to sell our RX1 and be done with it. But I noticed that there is one element which is different from the 2 tests: I had a filter on mine, the other RX1 did not. So for the sake of thoroughness (thanks to mythbusters for this habit), I took the filter off mine, not expecting anything different.

EDIT (23/3): Photos deleted. If any want to see test results, pm me.

 

[moiler]

TEST 2:

So, with the filter off my RX1 (RX1 A):

5. RX1A F-OFF CENTER
6. RX1A F-OFF LEFT

Compare #6 (F-OFF) and #2 (F-ON), both on my RX1 (RX1 A), where both subjects are on the left and you will see #2 comes off as more distorted.

Edit (23/3): Images deleted.

 

[moiler]

TEST 3

By this time, all I could think of was....WTF??

So I placed my filter on the other RX1 (RX1 B).

7. RX1B F-ON CENTER
8. RX1B F-ON LEFT

If you compare #8 (RX1 B F-ON) and #6 (RX1 A F-OFF), you will now hopefully see that my RX1 A has the better result, with both subjects being on the LEFT.

Edit (23/3): Images deleted

 

[moiler]

TEST 4:

I took one last test to reconfirm my findings. I took 2 shots, with F-ON and F-OFF with the subject in the center, both times with my RX1. The difference is slight, but even here I can see her face is more rounded with the F-ON, even in the center position.

9. RX1A F-ON CENTER
10. RX1A F-OFF CENTER

Edit (23/3): Images deleted.

 

[moiler]

Did you also conclude that it was the filter that was causing the additional (or a different kind of) distortion to the images?

If you concluded this, why do you think that is?

I for one am disappointed that my (relatively) expensive B+W filter does this. Admittedly it was bought off ebay, so it may be fake. But if it is fake, it's a very good one, the box came with a hologram (and the seller has numerous high ratings). And it is hefty for a 49mm.


Note: I was not in my house when I conducted this test, so I did not have my tripod handy. And I was in a hurry so I left the dial set to P, whereas ideally I should have gone manual and picked a constant Aperture, Speed and ISO for the test to be doubly sure everything is equal.

 

[rbelyell]

like spike lee said many years ago 'gotta be the shoes'. great test, thanks. dont be upset at anything, your $3000 camera works great. get a cheap hood instead, like i did. an expensive lens like this deserves to perform in the raw anyway. tbh, i really never understood having an expensive, top engineered piece of glass and shooting it every day through some other piece of glass. the hood is very small and will protect it just as well.

 

[moiler]

like spike lee said many years ago 'gotta be the shoes'. great test, thanks. dont be upset at anything, your $3000 camera works great. get a cheap hood instead, like i did. an expensive lens like this deserves to perform in the raw anyway. tbh, i really never understood having an expensive, top engineered piece of glass and shooting it every day through some other piece of glass. the hood is very small and will protect it just as well.

So you confirm what I'm seeing? ...I wasn't hallucinating?


I hate to mislead people, if you and a few others reconfirm my findings, I will either:

1. Delete the thread so people aren't mislead about the RX1
2. Modify my original entry to reflect this result
3. Delete this thread and create a filter distortion thread (perhaps in a more general forum). Now I'm piqued by this... I can't be the only one out of God knows how many photographers our there who has stumbled on to this.

What do you suggest...


I have attached the offending party. Tomorrow I may get a hoya or heliopan clear filter or a hood (this time from a reputable brick and mortar shop). If I get a filter, I will test it. Stay tuned.

 

[rbelyell]

why not just use a hood? why shoot through a window with this beautiful zeiss optic?

 

[serhan]

Per FM posts, you can use a 49-37mm step down ring instead of a hood, similar to FE 35mm hood.

 

[jloden]

Ouch, certainly could be a culprit... I see the effect you're talking about switching between the shots. Since I can see it in multiple sets of photos it does seem plausible it's the filter and not a subtle movement of the camera or subject.

This kind of thing is why I just don't trust filters in front of my expensive lens glass. I use an ND or polarizer filter only when I need to, and the rest of the time use hoods and lens caps rather than risk something wonky happening with an extra piece of glass in front of my lens. I totally understand why people do it, but I've seen more than one "oops, it was the filter" threads on forums to use one unless I really needed the protection from spray or something.

Anyway, glad to hear you're happier with the results sans filter, if that solves the problem for you then I'm more than glad for you!