D7 distortion at WA: not barrel, tilde !!!

hehehe, I like the 'creatively useful opportunities'! That's even better than calling it "linearity challenged".
But the interesting concern coming out of this is what happens when
you stick a .55 or even a .7 WA converor on this. I suspect that
will really give some interesting distortion (should I say
'creatively useful opportunities').

dh
 
Bingo! I wasn't expecting it 'cause it didn't seem to show up in any tests. None of the reviewers mentioned it either, that I saw anyway.

I guess I haven't been paying close enough attention to zoom lenses. I only recall seeing it on a couple of film zoom lenses before, and they were both defective (and replaced my the manufacturers).

Still might be worth a call for the Tech Support folks, guys. Guess if it were me, I would anyway - nothing to lose trying really.
I think that it is "normal" for the D7. They don't call it
anomalous dispersion glass for nothing. I've seen the gull wing
distortion on other zooms for film cameras. I don't know if it can
be corrected in software or not.
Bryan
Still curious, why wouldn't have this shown up on the barrel
distortion test target in Phil's review? Is it still possible they
don't all do it? Seems unlikely since two of you have it, but
stranger things have happened.

I don't have a digicam yet, but I've never noticed this on
my Minolta 24-85 zoom that I use on my film camera. I'm
certainly going to check it out, though. Not being digital,
however, I won't know the answer for a while.

-D
 
I think that it is "normal" for the D7. They don't call it
anomalous dispersion glass for nothing. I've seen the gull wing
distortion on other zooms for film cameras. I don't know if it can
be corrected in software or not.
Here is a copy of Bryan's siding picture that I have corrected
(more or less) for the distortion (more below):



Here is the caveat, though: I did this using a plug-in called
"Curve Bend" for Gimp, which is a UNIX based image manipulation
program similar to PS - I don't know if there is a similar filter for
Photoshop, although I suppose there is. Another caveat is that
this filter wouldn't work for vertical lines, as it only allows adjustment
along the top and bottom. Most architectural shots tend to
perspective-distort vertical lines anyway, so it might not be a big
limitation.

The "Curve Bend" filter allows one to apply an arbitrary correction
curve to both the top and bottom of an image. The filter leaves
artifacts, however, so I was only able to get good results by first
resizing the image up and then back down after applying the
distortion, YMMV. It does tend to soften the image just a bit,
as it tries to smooth the effects of the distortion that it is applying.
Also, note the white area at the top right. This is
unavoidable and would need to be cropped out (I left it in bacause
I am running late getting back to work ).

-D
 
Bingo! I wasn't expecting it 'cause it didn't seem to show up in
any tests. None of the reviewers mentioned it either, that I saw
anyway.
Hi all,
Many thanks to Bryan, now I feel less alone ;-)

I tried to shot a grid. The sheet of paper was not very flat sorry (maybe someone here has a pro set-up for this kind of test?) but obviously the seagull wing does NOT appear, while the effect was very strong and unacceptable in my previous shots of buildings.

I'd bet that the defect depends on the angle of the shot (X and/or Z axis?) and maybe the distance of the object?

Knowing that, Phil's D7 may be as defective as mine but still can pass the classic distortion tests!
The following grid shows barrel only distortion:

 
obviously the seagull wing does NOT appear, while the effect was
very strong and unacceptable in my previous shots of buildings.
I'd bet that the defect depends on the angle of the shot (X and/or
Z axis?) and maybe the distance of the object?
Make sure the grid goes to the edge of the image - that's where the effect is most evident. My camera does indeed show the "seagull wing" on straight on, close up shots.

I found that a combination of Photoshop's Pinch and Spherixe (both under the Filter/Distortion menu) can help. This is one of GBo's images corrected in Photoshop.



Scale the canvas size to 150% width and 200% height, apply the Pinch filter at 12%, and Spherize at 16%. Then crop it back to its original size. This process softens the image a bit, so I applied 1 pixel, 50% unsharp mask.

Obviously not perfect, but an improvement.
aturoff
 
Scotty wrote:
The following grid shows barrel only distortion:
It actually looks as though if your grid were to extend completely
to the edge of the image that the gull effect would appear near the
edges. Right where your grid stops on the left and right, it appears
that the lines are just beginning to "turn around" and might start
bending up again (or down, on the bottom). Would it be
possible for you to reshoot with the grid filling the entire field?
 
I found that a combination of Photoshop's Pinch and Spherixe (both
under the Filter/Distortion menu) can help.
Wow! In a few hours time we have a Photoshop workaround
and a GIMP workaround!

It makes me wonder, however, whether it would be possible
for Minota to put some software correction for this in the
camera itself. Perhaps there just isn't enough processing
power to do it in a reasonable amount of time.

-D
 
Scotty wrote:
The following grid shows barrel only distortion:
It actually looks as though if your grid were to extend completely
to the edge of the image that the gull effect would appear near the
edges. Right where your grid stops on the left and right, it appears
that the lines are just beginning to "turn around" and might start
bending up again (or down, on the bottom). Would it be
possible for you to reshoot with the grid filling the entire field?
David,

you may be right, my test and its conclusions are not reliable. Unfortunatly the largest size that I can print is A3 format and the D7 field of view at 50 cm and 28 mm is larger than the grid!
 
Here is my quick'n'dirty test pic:



The size of the test target was 600 x 450 mm (small squares are 50 x 50 mm). It shows quite clear barrel distrotion - except in the extreme corners. Because the distortion is symmetrical - as far as I can tell - I don't think it's an alignment problem.

Marko
 
Another good program(PS plugin) for this is Andromeda's Lens Doc, with which you can take some very complex wonk out of anything with lines, especially buildings. Repeated applications might be necessary.

dh
I think that it is "normal" for the D7. They don't call it
anomalous dispersion glass for nothing. I've seen the gull wing
distortion on other zooms for film cameras. I don't know if it can
be corrected in software or not.
Here is a copy of Bryan's siding picture that I have corrected
(more or less) for the distortion (more below):



Here is the caveat, though: I did this using a plug-in called
"Curve Bend" for Gimp, which is a UNIX based image manipulation
program similar to PS - I don't know if there is a similar filter for
Photoshop, although I suppose there is. Another caveat is that
this filter wouldn't work for vertical lines, as it only allows
adjustment
along the top and bottom. Most architectural shots tend to
perspective-distort vertical lines anyway, so it might not be a big
limitation.

The "Curve Bend" filter allows one to apply an arbitrary correction
curve to both the top and bottom of an image. The filter leaves
artifacts, however, so I was only able to get good results by first
resizing the image up and then back down after applying the
distortion, YMMV. It does tend to soften the image just a bit,
as it tries to smooth the effects of the distortion that it is
applying.
Also, note the white area at the top right. This is
unavoidable and would need to be cropped out (I left it in bacause
I am running late getting back to work ).

-D
 
I did this using a plug-in called
"Curve Bend" for Gimp, which is a UNIX based image manipulation
program similar to PS - I don't know if there is a similar filter for
Photoshop, although I suppose there is. Another caveat is that
this filter wouldn't work for vertical lines, as it only allows
adjustment
along the top and bottom. Most architectural shots tend to
perspective-distort vertical lines anyway, so it might not be a big
limitation.

The "Curve Bend" filter allows one to apply an arbitrary correction
curve to both the top and bottom of an image. The filter leaves
artifacts, however, so I was only able to get good results by first
resizing the image up and then back down after applying the
distortion, YMMV. It does tend to soften the image just a bit,
as it tries to smooth the effects of the distortion that it is
applying.
I remember an article in a photo magazine which showed how you can treat any kind of distortion by using displacement maps in Photoshop. I have to dig this article out again, but perhaps the mentioning of displacement maps does ring a bell with somebody here?

Roland
 
Another good program(PS plugin) for this is Andromeda's Lens Doc,
with which you can take some very complex wonk out of anything with
lines, especially buildings. Repeated applications might be
necessary.
Thank you David and Papatrout for proposing good software workarounds! This is a GREAT forum and I learn a lot with it! (just to compare I posted the description of the problem in a french photo forum to alert other D7 owners, but nobody seemed to care up to now).

For your info, I wrote to Minolta Europe whose support team will issue a "technical request" (I guess to the Japanese R&D?) to study my 3 pictures. They don't understand yet this non-symetric effect but suppose that the angle of shot may amplify it. They suggest also that I send my camera to the french technical support to check it in the meantime. I'll keep you informed.
 
David Rosky wrote:

I remember an article in a photo magazine which showed how you can
treat any kind of distortion by using displacement maps in
Photoshop. I have to dig this article out again, but perhaps the
mentioning of displacement maps does ring a bell with somebody here?

Roland
I've seen displacement maps both in PS and GIMP, but I didn't
know what they were. I'd be curious if you find out more.

(Of course, we're probably getting onto a topic that more
properly belongs in the "PC tools" forum, but what the heck!)

-D
 
I simplified the photoshop process a bit. Increase canvas width to 125% and height 166.67%. Pinch 7%. Increase image size to 102.5%. Change canvas size back to original.

Fixed version and original are below.

aturoff



 
The big questions.

1) Has anyone ELSE encountered this type of distorsion?
2) Has anyone got any explanation from Minolta?
3) Is GBo 's camera defective?

I would add that I like the D7 very much and this distorsion is the only problem that refrains me from buying it instead of the Nikon CP 5000. As I shoot mainly at buildings at wide angle (28mm) I am much concerned by this strange phenomenon. Thank you for your replies.

Regards,
John
Another good program(PS plugin) for this is Andromeda's Lens Doc,
with which you can take some very complex wonk out of anything with
lines, especially buildings. Repeated applications might be
necessary.
Thank you David and Papatrout for proposing good software
workarounds! This is a GREAT forum and I learn a lot with it! (just
to compare I posted the description of the problem in a french
photo forum to alert other D7 owners, but nobody seemed to care up
to now).
For your info, I wrote to Minolta Europe whose support team will
issue a "technical request" (I guess to the Japanese R&D?) to study
my 3 pictures. They don't understand yet this non-symetric effect
but suppose that the angle of shot may amplify it. They suggest
also that I send my camera to the french technical support to check
it in the meantime. I'll keep you informed.
 
The big questions.

1) Has anyone ELSE encountered this type of distorsion?
2) Has anyone got any explanation from Minolta?
3) Is GBo 's camera defective?

I would add that I like the D7 very much and this distorsion is
the only problem that refrains me from buying it instead of the
Nikon CP 5000. As I shoot mainly at buildings at wide angle (28mm)
I am much concerned by this strange phenomenon. Thank you for your
replies.


Regards,
John
John --

This thread is from last August, and nobody else has, as far as I can recall, posted a similar thread since then, nor have any of the people I know who own D7s ever complained of this. In addition, I've never seen it in any of the D7 pictures anyone else has posted anywhere on the Web.

This doesn't mean it doesn't happen -- somebody somewhere might have the problem, I suppose, but if it were common I'd have expected to see it mentioned here on many occasions (everything else people perceive to be a problem with the camera is announced and reannounced by every new wave of owners who find this site).

If this is the only thing keeping you from buying the camera, I'd suggest getting one from a dealer who will allow you to return it without penalty and try it for yourself.

Sam in SF
 
The big questions.

1) Has anyone ELSE encountered this type of distorsion?
2) Has anyone got any explanation from Minolta?
3) Is GBo 's camera defective?

I would add that I like the D7 very much and this distorsion is
the only problem that refrains me from buying it instead of the
Nikon CP 5000. As I shoot mainly at buildings at wide angle (28mm)
I am much concerned by this strange phenomenon. Thank you for your
replies.


Regards,
John
John, I have noticed the problem, too. Although I was able to realise it was there at only about 3 photos out of the many thousands I have shot with my camera.

I guess it really depends on the angle between the CCD and the 'wall edge' and also the edge and the lens.

I can not say that it has ruinned any of my shots. For all I can say, it may still be there at all my shots and maybe I am not able to see it. It may be due to the corrections of the lens, in order to avoid blurred edges.

In my opinion, I'd rather have 3 shots out of several thousands like the one submitted, rather than having most of my shots not sharp edge-to-edge.--Pablettowww.pbase.com/pabletto(samples taken with the MINOLTA DiMAGE 7)Last update 26 February 2002 (MORE food photos)
 
Scotty wrote:
With all due respect to those who suggest this is normal, I'd be
contacting the Minolta service center about this if it were mine.

Regards,
Scotty
Seems me that those shots are only a perfect example of the worst combination of the barrel distortion and the normal perspective distortion inhaerent to wideangle lenses.

From short distance and not square to the surface of a very big structure: There you have it.
Hans.
 
Hi Pabletto & others
Thanks for your reply,

You said 3 out of thousands of shots only 3 have the distorsion. I guess that most others are not at 28 mm, or shot at horizontal lines with the camera pointing at an angle to the building plane. So I would be very grateful if anyone of you here can do me a favour: Post some pictures taken in the same context as GBo (with & without distorsion).

Regards
John
The big questions.

1) Has anyone ELSE encountered this type of distorsion?
2) Has anyone got any explanation from Minolta?
3) Is GBo 's camera defective?

I would add that I like the D7 very much and this distorsion is
the only problem that refrains me from buying it instead of the
Nikon CP 5000. As I shoot mainly at buildings at wide angle (28mm)
I am much concerned by this strange phenomenon. Thank you for your
replies.


Regards,
John
John, I have noticed the problem, too. Although I was able to
realise it was there at only about 3 photos out of the many
thousands I have shot with my camera.
I guess it really depends on the angle between the CCD and the
'wall edge' and also the edge and the lens.
I can not say that it has ruinned any of my shots. For all I can
say, it may still be there at all my shots and maybe I am not able
to see it. It may be due to the corrections of the lens, in order
to avoid blurred edges.
In my opinion, I'd rather have 3 shots out of several thousands
like the one submitted, rather than having most of my shots not
sharp edge-to-edge.
--
Pabletto

http://www.pbase.com/pabletto
(samples taken with the MINOLTA DiMAGE 7)
Last update 26 February 2002 (MORE food photos)
 
I'm sorry ... but wouldn't any extreme angle shot at anything other then 1X magnification do exactly this?

I'm no expert, but I thought this was a simple matter of physics of optics (concave/convex lenses, etc).
 

Keyboard shortcuts

Back
Top