Unexplained Alteration of Horizon


Dan Margulis
 

Can anyone explain an anomaly, apparently due to differences in acquisition, in the Grand Canyon images we're studying?

In all of the versions that I posted (which I derived from the raw) the far horizon is bowed, that is, the left and right sides are slightly lower than the center. The large majority of entrants have this characteristic and register with each other perfectly.

One entrant, however, submitted one where the horizon seemed to have been straightened, it no longer bowed but rather went very slightly downhill going from left to right. Although this image has the same number of pixels as all the others it doesn't quite register with them, especially higher up in the image, as the canyon itself is affected.

I thought that this must have been someone being hyperenergetic, but then two more entrants showed up with *exactly* the same pattern. The three entrants match up precisely with each other but not with the majority. So there has to be something funky going in in acquisition. Does anybody know what it is?

The three entrants who don't match the others are JJB, NJ, and JP.

Dan Margulis


dbernaerdt
 

Dan,

I think it is likely the Lens Profile Corrections have been enabled in the "Profile" tab of the "Lens Corrections" section of Camera Raw. This could be applying a distortion correction that is straightening the horizon - especially since it is so consistent.

Darren Bernaerdt

--- In colortheory@..., Dan Margulis <DMargulis@...> wrote:

Can anyone explain an anomaly, apparently due to differences in acquisition, in the Grand Canyon images we're studying?

In all of the versions that I posted (which I derived from the raw) the far horizon is bowed, that is, the left and right sides are slightly lower than the center. The large majority of entrants have this characteristic and register with each other perfectly.

One entrant, however, submitted one where the horizon seemed to have been straightened, it no longer bowed but rather went very slightly downhill going from left to right. Although this image has the same number of pixels as all the others it doesn't quite register with them, especially higher up in the image, as the canyon itself is affected.

I thought that this must have been someone being hyperenergetic, but then two more entrants showed up with *exactly* the same pattern. The three entrants match up precisely with each other but not with the majority. So there has to be something funky going in in acquisition. Does anybody know what it is?

The three entrants who don't match the others are JJB, NJ, and JP.

Dan Margulis


Michael Demyan <mdemyan@...>
 

Hi Dan:

I noticed the bow also. I attribute it to the spherical abberation of the camera lens.

I discovered that if in the Lens Corrections module the "Enable Profile Corrections" box is checked in the RAW converter (both ACR and Lightroom 3.6 allow for this adjustment) the images will have the anomaly.

The metadata on the RAW image shows the camera as Canon 5D MKII and the lens as EF70-200 f2.8L.

Mike Demyan

----- Original Message -----
From: Dan Margulis
To: colortheory@...
Sent: Tuesday, January 10, 2012 11:07 PM
Subject: [colortheory] Unexplained Alteration of Horizon



Can anyone explain an anomaly, apparently due to differences in acquisition, in the Grand Canyon images we're studying?

In all of the versions that I posted (which I derived from the raw) the far horizon is bowed, that is, the left and right sides are slightly lower than the center. The large majority of entrants have this characteristic and register with each other perfectly.

One entrant, however, submitted one where the horizon seemed to have been straightened, it no longer bowed but rather went very slightly downhill going from left to right. Although this image has the same number of pixels as all the others it doesn't quite register with them, especially higher up in the image, as the canyon itself is affected.

I thought that this must have been someone being hyperenergetic, but then two more entrants showed up with *exactly* the same pattern. The three entrants match up precisely with each other but not with the majority. So there has to be something funky going in in acquisition. Does anybody know what it is?

The three entrants who don't match the others are JJB, NJ, and JP.

Dan Margulis


David Lawrence
 

--- In colortheory@..., Dan Margulis <DMargulis@...> wrote:

Can anyone explain an anomaly, apparently due to differences in acquisition, in the Grand Canyon images we're studying?
Dan Margulis
Could it be *process method* with different RAW converters or versions of Lightroom/ACR?

I wanted to fix it with distortion correction, but didn't want to affect the pixel count or crop. So, I just pretended the earth was round. :o)

David Lawrence
www.pixelpurfect


David Lawrence
 

--- In colortheory@..., Dan Margulis <DMargulis@...> wrote:

Can anyone explain an anomaly, apparently due to differences in acquisition, in the Grand Canyon images we're studying?
Dan Margulis
Dan,

After reviewing and responding to other related posts I went back to check my own work and discovered that I had *not* used the lens profile correction in Lightroom even though I said I did in my details submission. I had it on and must have turned it off thinking it may change the pixel count. Weird, I always use lens correcton, maybe I'll have some time later in the week to resubmit.

Lens profile does't completly straighten the horizon line, but a combined +3 horisontal correction gets pretty darn close. And of course that was prohibited in the entry instructions. :o)

David Lawrence
www.pixelpurfect.com


Davide D'Angelo <dandavide@...>
 

Hi Mike



You are right, I have just try it and if you check "Enable Profile
Corrections" you can clearly see the difference, the attempt is to correct
the spherical aberration of the lens! But we are not sure about the exact
correction. It should be interesting to view the real difference with Canon
Raw Converter “DPP”.



Davide D’Angelo





Da: colortheory@... [mailto:colortheory@...] Per
conto di Michael Demyan
Inviato: mercoledì 11 gennaio 2012 7.24
A: colortheory@...
Oggetto: Re: [colortheory] Unexplained Alteration of Horizon





Hi Dan:

I noticed the bow also. I attribute it to the spherical abberation of the
camera lens.

I discovered that if in the Lens Corrections module the "Enable Profile
Corrections" box is checked in the RAW converter (both ACR and Lightroom 3.6
allow for this adjustment) the images will have the anomaly.

The metadata on the RAW image shows the camera as Canon 5D MKII and the lens
as EF70-200 f2.8L.

Mike Demyan

._,_.___

Reply to
<mailto:mdemyan@...?subject=Re%3A%20%5Bcolortheory%5D%20Unexplained%20Al
teration%20of%20Horizon> sender | Reply to
<mailto:colortheory@...?subject=Re%3A%20%5Bcolortheory%5D%20Unex
plained%20Alteration%20of%20Horizon> group | Reply
<http://groups.yahoo.com/group/colortheory/post;_ylc=X3oDMTJxNTczbW5yBF9TAzk
3MzU5NzE0BGdycElkAzI1NjA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BG1zZ0lkAzI0Njk3BHNlYwN
mdHIEc2xrA3JwbHkEc3RpbWUDMTMyNjI2NzQ1NQ--?act=reply&messageNum=24697> via
web post |
<http://groups.yahoo.com/group/colortheory/post;_ylc=X3oDMTJlNWFiMmM2BF9TAzk
3MzU5NzE0BGdycElkAzI1NjA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BHNlYwNmdHIEc2xrA250cGM
Ec3RpbWUDMTMyNjI2NzQ1NQ--> Start a New Topic

Messages in this topic
<http://groups.yahoo.com/group/colortheory/message/24692;_ylc=X3oDMTM2NXBrOD
hyBF9TAzk3MzU5NzE0BGdycElkAzI1NjA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BG1zZ0lkAzI0Nj
k3BHNlYwNmdHIEc2xrA3Z0cGMEc3RpbWUDMTMyNjI2NzQ1NQR0cGNJZAMyNDY5Mg--> (3)

Recent Activity:

·
<http://groups.yahoo.com/group/colortheory/members;_ylc=X3oDMTJmNGYxdDgwBF9T
Azk3MzU5NzE0BGdycElkAzI1NjA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BHNlYwN2dGwEc2xrA3Zt
YnJzBHN0aW1lAzEzMjYyNjc0NTU-?o=6> New Members 4

·
<http://groups.yahoo.com/group/colortheory/spnew;_ylc=X3oDMTJmYXY3NDFsBF9TAz
k3MzU5NzE0BGdycElkAzI1NjA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BHNlYwN2dGwEc2xrA3ZwaG
90BHN0aW1lAzEzMjYyNjc0NTU-> New Photos 3

·
<http://groups.yahoo.com/group/colortheory/files;_ylc=X3oDMTJnZDlhZHM5BF9TAz
k3MzU5NzE0BGdycElkAzI1NjA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BHNlYwN2dGwEc2xrA3ZmaW
xlcwRzdGltZQMxMzI2MjY3NDU1> New Files 1


<http://groups.yahoo.com/group/colortheory;_ylc=X3oDMTJldDVxN2NyBF9TAzk3MzU5
NzE0BGdycElkAzI1NjA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BHNlYwN2dGwEc2xrA3ZnaHAEc3Rp
bWUDMTMyNjI2NzQ1NQ--> Visit Your Group

MARKETPLACE

Stay on top of your group activity without leaving the page you're on - Get
the Yahoo! Toolbar now.
<http://global.ard.yahoo.com/SIG=15ovehl9s/M=493064.14543979.14562481.132984
30/D=groups/S=1705115359:MKP1/Y=YAHOO/EXP=1326274655/L=20dd86a8-3c27-11e1-9a
df-bb9c419e47fb/B=5V8EAtBDRmQ-/J=1326267455088350/K=cG3KTlxKA2v0PUfXWyPvJA/A
=6060255/R=0/SIG=1194m4keh/*http:/us.toolbar.yahoo.com/?.cpdl=grpj>




<http://groups.yahoo.com/;_ylc=X3oDMTJkam8yczhqBF9TAzk3MzU5NzE0BGdycElkAzI1N
jA2NzEEZ3Jwc3BJZAMxNzA1MTE1MzU5BHNlYwNmdHIEc2xrA2dmcARzdGltZQMxMzI2MjY3NDU1>
Yahoo! Groups

Switch to:
<mailto:colortheory-traditional@...?subject=Change%20Delivery%20
Format:%20Traditional> Text-Only,
<mailto:colortheory-digest@...?subject=Email%20Delivery:%20Diges
t> Daily Digest •
<mailto:colortheory-unsubscribe@...?subject=Unsubscribe>
Unsubscribe • <http://docs.yahoo.com/info/terms/> Terms of Use

.


<http://geo.yahoo.com/serv?s=97359714/grpId=2560671/grpspId=1705115359/msgId
=24697/stime=1326267455/nc1=4025304/nc2=5191954/nc3=3848642>





[Non-text portions of this message have been removed]


Fernando Chaves
 

So there has to be something funky going in in acquisition.
Does anybody know what it is?

The three entrants who don't match the others are JJB, NJ, and JP.

Dan Margulis
Hi,
I think this is the "lens corrections" feature in ACR, Lightroom, DXO or
PTlens (after raw convertion). The auto-correction can be fine tuned and
exaggerated.
DXO was the very first raw converter product to offer this feature and I
think it does a better correction than ACR.
Best regards,
Fernando Chaves


Fernando Chaves
 

Davide D'Angelo wrote:
Hi Mike

You are right, I have just try it and if you check "Enable Profile
Corrections" you can clearly see the difference, the attempt is to correct
the spherical aberration of the lens! But we are not sure about the exact
correction. It should be interesting to view the real difference with
Canon
Raw Converter DPP.

Davide DAngelo
Hi,
This is a good paper about the math behind lens corrections:

http://www.panotools.org/dersch/barrel/barrel.html

Best regards,
Fernando


dlruckus
 

Has anyone given thought to the possibility that you are looking at the earths curvature. After all, the horizon is very far away. That doesn't say that a very wide angle lens may not have accented the view but making it exactly straight would be a profound distortion as well.
I can't comment on camera raw as I have no experience in that realm but I would be upset if I found in using it that it did things I did not intend.

Regards,
Duane Ruck

--- In colortheory@..., "Michael Demyan" <mdemyan@...> wrote:

Hi Dan:

I noticed the bow also. I attribute it to the spherical abberation of the camera lens.

I discovered that if in the Lens Corrections module the "Enable Profile Corrections" box is checked in the RAW converter (both ACR and Lightroom 3.6 allow for this adjustment) the images will have the anomaly.

The metadata on the RAW image shows the camera as Canon 5D MKII and the lens as EF70-200 f2.8L.

Mike Demyan

----- Original Message -----
From: Dan Margulis
To: colortheory@...
Sent: Tuesday, January 10, 2012 11:07 PM
Subject: [colortheory] Unexplained Alteration of Horizon



Can anyone explain an anomaly, apparently due to differences in acquisition, in the Grand Canyon images we're studying?

In all of the versions that I posted (which I derived from the raw) the far horizon is bowed, that is, the left and right sides are slightly lower than the center. The large majority of entrants have this characteristic and register with each other perfectly.

One entrant, however, submitted one where the horizon seemed to have been straightened, it no longer bowed but rather went very slightly downhill going from left to right. Although this image has the same number of pixels as all the others it doesn't quite register with them, especially higher up in the image, as the canyon itself is affected.

I thought that this must have been someone being hyperenergetic, but then two more entrants showed up with *exactly* the same pattern. The three entrants match up precisely with each other but not with the majority. So there has to be something funky going in in acquisition. Does anybody know what it is?

The three entrants who don't match the others are JJB, NJ, and JP.

Dan Margulis





Steve Jenkins <sj9000@...>
 

--- In colortheory@..., Dan Margulis <DMargulis@...> wrote:

Can anyone explain an anomaly, apparently due to differences in acquisition, in the Grand Canyon images we're studying?
Dan Margulis
Hi current image challenge participants,

If you look at the EXIF metadata for this image, you'll see the focal length, exposure, aperture, etc.. Oh, and exact time it was taken . . . which might be a clue as to how the final image should be rendered (i.e. the atmospheric and lighting conditions).

And as for lens distortion correction, most software that does this sort of stuff relies on a database of cameras and lenses, which is static. The software reads the image EXIF metadata and applies the correction for that combination. There's also usually a manual mode to the software which allows you to "eyeball" or adjust sliders to remove the perceived distortion. The individual software companies maintain their own databases, so when you update ARC, for example, you're also updating that database. Unfortunately creating rectilinear (non distorting - keeping lines straight) lenses is extremely expensive so in most images there's usually some non-linear distortion. For example my snapshot camera only produces near rectilinear images at a zoom level between 1.5-1.8 (go figure), otherwise there's some level of barrel or pincushion distortion. And yes, actual measurements are made of the camera/lens/zoom combinations to maintain these databases.

Here's the EXIF metadata for one of the current challenge images:
=================== BEGIN EXIF METADATA ===================
Filename - gc2459-zero.jpg
ImageDescription - opened zeroed out in ACR
Make - Canon
Model - Canon EOS 5D Mark II
Orientation - Top left
XResolution - 240.00
YResolution - 240.00
ResolutionUnit - Inch
Software - Adobe Photoshop CS5 Macintosh
DateTime - 2012:01:07 18:56:00
Copyright - Copyright 2008 by Parthiv Mehta. Licensed to Dan Margulis for educational use only. May not be reposted.
ExifOffset - 372
ExposureTime - 1/100 seconds
FNumber - 16
ExposureProgram - Aperture priority
ISOSpeedRatings - 400
ExifVersion - 0221
DateTimeOriginal - 2008:08:04 16:43:18
DateTimeDigitized - 2008:08:04 16:43:18
ShutterSpeedValue - 1/100 seconds
ApertureValue - F 16.00
ExposureBiasValue - 0.33
MaxApertureValue - F 2.83
SubjectDistance - -1 m
MeteringMode - Spot
Flash - Flash not fired, compulsory flash mode
FocalLength - 70 mm
SubsecTime - 16
SubsecTimeOriginal - 16
SubsecTimeDigitized - 16
ColorSpace - sRGB
ExifImageWidth - 3072
ExifImageHeight - 2048
FocalPlaneXResolution - 3849.21
FocalPlaneYResolution - 3908.14
FocalPlaneResolutionUnit - Inch
CustomRendered - Normal process
ExposureMode - Auto
White Balance - Auto
SceneCaptureType - Standard

Thumbnail: -
Compression - 6 (JPG)
XResolution - 72
YResolution - 72
ResolutionUnit - Inch
JpegIFOffset - 930
JpegIFByteCount - 3530

=================== END EXIF METADATA ===================

Probably more than you needed to know but there it is.

Steve


Bill Theis
 

this is from a Canon EOS 5D Mark II? somehow I expected better. looks like a lot of color noise, especially visible at the tree/sky line

so for Dan's purposes, we should all print the same way, either everyone with lens correction on or everyone with it off --- making direct comparisons of the various versions easier IMHO

Bill Theis