Problem with vignetting correction

[Tom]

I am using PTGUI Pro (version 9.1.4) for stitching Gigapixel
Panoramas. My latest panoramas almost always suffer from vignetting,
even though I am applying the vignetting correction in PTGui Pro -
this is one example: http://www.gigapan.com/gigapans/117457
I didn't have significant vignetting problems before, when I used a
powershot G9 with a 2x teleconverter.However, with my new camera, a
Lumix FZ200 (camera params for this panorama f=600mm , fnumber=4.0,
jpg images) the vignetting correction doesn't provide good results.
However, the vignetting is relatively moderate, the intensity falls
off by about 5 percent at the edge.
Could the lens hood be responsible (with the FZ200 I always made the
photos with the provided lens hood)? Maybe the vignetting is not
radially symmetric?

Any suggestions what I could do to improve the vignetting correction?

Tom

[Matthew Ward]

On 30 Oct 2012, at 20:48, Tom wrote:

camera params for this panorama f=600mm

600mm?

Best

Matthew

[PTGui Support]

Hi Tom,

The exposure and vignetting correction in PTGui indeed assumes that the
light fall off is radially symmetric. If it is symmetric PTGui should be
able to correct it quite well.

Could you make a small section of the panorama available (project file +
source images)?

Joost

[Erik Krause]

Am 30.10.2012 21:48, schrieb Tom:
> My latest panoramas almost always suffer from vignetting,
> even though I am applying the vignetting correction in PTGui Pro -
> this is one example:http://www.gigapan.com/gigapans/117457

This doesn't look like the result of ordinary vignetting. A light
falloff from one side to the other could be the cause. A close look at
an image with uniform sky could reveal this. If you make a small section
of your panorama project available like Joost suggested, please include
at least one image with uniform sky.

--
Erik Krause

[Tom]

The lumix fz200 is a superzoomer with a focal length of 600mm
(equival.), a really nice camera.

On Oct 31, 7:22 am, Matthew Ward <matt...@matthewwardphotography.com>
wrote:

[Tom]

Hi Joost,

I've have just uploaded a small part of the panorama (images + project
file) to dropbox
https://www.dropbox.com/sh/0tarkrbtuhfq0j3/bBy4GaYe72

Best regards,

Tom

[PTGui Support]

Hi Tom,

It looks like there is a linear brightness gradient from left to right
(as seen in portrait orientation). This should be due to a difference
between the opening and closing speed of the shutter. Does the fz200
have a physical curtain shutter?

This kind of exposure correction is currently not supported, but it's on
the wish list.

Joost

[UtahBob]

On Thursday, November 1, 2012 8:05:42 AM UTC-4, PTGui Support wrote:

It looks like there is a linear brightness gradient from left to right
(as seen in portrait orientation). This should be due to a difference
between the opening and closing speed of the shutter. Does the fz200
have a physical curtain shutter?

This kind of exposure correction is currently not supported, but it's on
the wish list.

Joost

Joost,  

I'm seeing somewhat the same issue as well as exposure change resulting from changes in light levels as the sun hides behind a cloud.  I'm at a loss as to why the blenders have issues blending out these brightness changes from frame to frame even with Enblend and Smartblend.  Just for kicks I tried Enblend on Tom's pano section and the results were almost identical to the PTGui blender along with no appreciable difference when changing the Enblend "levels" option.  It seems to me that the problem impacts smaller panos less such as sphericals which seems to blend well when using 6 shots around.  I recall in the past that enblend did a great job hiding exposure differences from frame to frame especially when using the -a command and those where differences substantially in excess of what we are talking about here.  I don't see that great blending anymore and I just don't understand what has changed?

I guess I have a lot of panos that I have sitting fully unprocessed waiting for a better blender and perhaps I need to focus on preprocessing the individual images for consistent light levels within and among images.  I'm thinking that's a difficult process for multigigapixel images and probably why I have not tackled it.

Also, I've been wanting to confirm - do the blend seams shown in the Pano Editor carry over for use by Enblend and Smartblend?  If not, does the masking somehow carry over and impact the Enblend and Smartblend processing?

Thanks,

Bob

[Tom]

I've just checked, the fz200 has a physical (round) iris shutter.
Tom

[John Houghton]

Tom, The image overlaps are not very large, so computing the
vignetting effect is not likely to work very well. It would be very
helpful if you could take a shot of an evenly illuminated flat sheet
(like a grey card), using the same camera settings as per the
panorama. De-focus a little to avoid capturing any grainy detail in
the sheet. We could then identify the exposure variations with a lot
more certainty and maybe develop an adjustment mask that could be used
to correct all the images in a batch process before processing in
PTGui.

John

[PTGui Support]

Hi Bob,

The reason that a blender cannot elimited the vignetting is because the
blender only works in the overlap area. The parts of the images that are
not overlapped by other images are not affected. Vignetting should
therefore be removed prior to blending.

When shooting panoramas in rapidly changing light you should use a lot
of overlap.

On 2-11-2012 4:00, UtahBob wrote:
> Also, I've been wanting to confirm - do the blend seams shown in the
> Pano Editor carry over for use by Enblend and Smartblend? If not, does
> the masking somehow carry over and impact the Enblend and Smartblend
> processing?

The red seam lines are specific to the PTGui blender. For spherical
panoramas they will differ quite a bit from the seams chosen by Enblend
and Smartblend. For non-spherical panoramas taken with a longer lens the
other blenders will use approximately the same seams as PTGui.

Red masks are carried over to the blending plugins. Green masks are only
picked up by the PTGui blender.

Joost

[Erik Krause]

Am 31.10.2012 21:13, schrieb Tom:
> The lumix fz200 is a superzoomer with a focal length of 600mm
> (equival.), a really nice camera.

Yes, but 600mm equivalent is still very high for this kind of camera. I
guess the lens is slightly decentered and this gives a heavily
decentered image at full focal length resulting in not radially
symmetric vignetting.

--
Erik Krause

[UtahBob]

On Friday, November 2, 2012 5:48:35 AM UTC-4, PTGui Support wrote:

The reason that a blender cannot elimited the vignetting is because the
blender only works in the overlap area. 

When shooting panoramas in rapidly changing light you should use a lot
of overlap.

Red masks are carried over to the blending plugins. Green masks are only
picked up by the PTGui blender.

Joost

Thanks Joost,  

Excellent insight.  I have more overlap in my sphericals so the better blending makes sense to me now.   I find that PS blends very well but for gigapixel images that's not an option so we'll have to preprocess for the vignetting, etc.  I've started using HDR to try and get around the changing light problem and that works well but creates more files and processing time.  I will increase my overlap and see how that works.  I'll note that I used to shoot row down but switched to column down like Tom shot his pano to try and reduce lighting change effects that can happen over the course of an hour or two in shooting a single pano. 

Exclusive red mask carryover now makes sense also from what I was getting using the plugins - I couldn't pin that one down.

Bob

[UtahBob]

On Friday, November 2, 2012 5:37:51 AM UTC-4, John Houghton wrote:

Tom, The image overlaps are not very large, so computing the
vignetting effect is not likely to work very well.  It would be very
helpful if you could take a shot of an evenly illuminated flat sheet
(like a grey card), using the same camera settings as per the
panorama.  De-focus a little to avoid capturing any grainy detail in
the sheet.  We could then identify the exposure variations with a lot
more certainty and maybe develop an adjustment mask that could be used
to correct all the images in a batch process before processing in
PTGui.

John

John,

 

Do you have any recommendations for an efficient method to execute an exposure adjustment for varying light conditions (post the vignettting adjustment) that could be implemented in PS or LR or ?? for panos that have a few hundred or a thousand images. I think my alternative might be to extract three images from each raw (adding a +2 and -2) to create an artificial HDR and then tonemap that either in PTGui or elsewhere.  This really is my largest hurdle right now to generate larger images properly.

Thanks,  Bob

[Tom]

Possibly I have found the culprit, it might be the image stabilizer
of the camera.
Today I made a series of images of the same part of the blue sky and
using ImageJ I enhanced the contrast. The result is that the images(at
max focal length of the cam) have an intensity minimum at the center
(ca. 6 percent lower than in the maximum)) and another minimum at the
corners. However, the center position (of the minimum) is moving from
shot to shot. Then I realized that this is only the case if the image
stabilizer is active, otherwise it rests in the center.
Images without stabilizer:
https://www.dropbox.com/sh/j572aem6jzxfa5i/24Lzi8sH5j/WithoutStabi
Images with stabilizer:
https://www.dropbox.com/sh/j572aem6jzxfa5i/EzrGH2Zyax/WithStabi
(might take a while until the upload has finished...)

Well, I hope its possible to make good gigapixel panoramas without
the stabilizer ...

Tom

On Oct 31, 9:00 am, PTGui Support <supp...@ptgui.com> wrote:

[Hambagahle]

With respect to the observation that "exposure change resulting from changes in light levels as the sun hides behind a cloud", I also find this occasionally.

For context: I shoot D300 RAW, convert RAW to 16 bit .tiff in ACR (Photoshop Extended CS4), and pull the .tiff into PTGUI.  All processing is done on an oldish MacBook Pro.  My overlaps are usually 25% or more, can go as high as 50% under operator error.

Here's what I do to "correct" exposure differences when they propagate through the PTGUI blender and compromise the resulting panorama:
1. Review image in PTGUI Panorama Editor in the "Preview without Blending".  This makes the tonal variations between overlapping images visible, especially when they're large.
2. Working one source image at a time, identify images that exhibit large tonal shifts.  Keep the project open in PTGUI, and stand by for some PTGUI magic: delete the offending .tiff, pull the source RAW into ACR, and adjust exposure up or down as appropriate (this is a guessing game), and create a fresh .tiff directly from ACR to replace the one you just deleted.  Hurry back to PTGUI, and you will see PTGUI refresh the panorama when the new .tiff is available (magic!!!).  Repeat as needed.

I delete offending .tiff files because ACR will use a different name for the new .tiff if the old one is still around.  My ACR does not overwrite the pre-existing .tiff.  If you prefer to keep the offending .tiff, you have two options: a) move it out of the current directory, or b) go to the "Source Files" tab in PTGUI and replace the old .tiff with the new one.

You guys probably know a thousand variations on this theme, hope you find this useful.

Joost, does this PTGUI magic also work under Windows?

Regards

[PTGui Support]

On 2-11-2012 14:59, Hambagahle wrote:
> Joost, does this PTGUI magic also work under Windows?

Yes it does!

But there's an even easier way: in PTGui Pro, switch to Advanced mode
and go to the Image Parameters tab. Enter a correction value in the Exp.
offset column. For example, -0.5 decreases the exposure for an image by
half a step.

Joost

[PTGui Support]

Still it's weird that the intensity minimum is at the center. For real
vignetting the center would have the maximum intensity.

Maybe the camera has a built in vignetting correction? Which seems to be
fooled by the lens shift caused by the stabilizer.

Joost

[Hambagahle]

Thanks for the feedback and the information.

Regards

[UtahBob]

Thanks guys, Wow! That does a great job cleaning up the sky.  I'm finding .1 or .2 ev is enough to conform images.  Now I'm seeing only the vignetting creating a problem.  Bob

[matthew ward]

The Lumix FZ200 claims to have a 600mm f2.8 lens. ie the aperture is 21.4cm/8" in diameter. This is clearly not the case, either Panasonic have found a way around the laws of physics and optics, or their marketing dept are being slightly enthusiastic with the use of the term aperture.

 

Best

Matthew Ward

-- You received this message because you are subscribed to the Google Groups "PTGui" group.
To post to this group, send email to pt...@googlegroups.com
To unsubscribe from this group, send email to ptgui+unsub...@googlegroups.com
Please do not add attachments to your posts; instead upload your files at a file sharing site (for example http://ge.tt/ ) and include a link in your message.
For more options, visit this group at http://groups.google.com/group/ptgui

[Joost Nieuwenhuijse - PTGui Support]

The camera probably has a relatively small sensor, and the stated 600mm actually is the 35mm-equivalent focal length.

Joost

> To unsubscribe from this group, send email to ptgui+un...@googlegroups.com

> Please do not add attachments to your posts; instead upload your files at a file sharing site (for example http://ge.tt/ ) and include a link in your message.
> For more options, visit this group at http://groups.google.com/group/ptgui
>
>
>

> --
> You received this message because you are subscribed to the Google Groups "PTGui" group.
> To post to this group, send email to pt...@googlegroups.com

> To unsubscribe from this group, send email to ptgui+un...@googlegroups.com

[matthew ward]

Sorry yes, it appears to be a 108mm

Best

Matthew Ward

> To unsubscribe from this group, send email to ptgui+unsub...@googlegroups.com

> Please do not add attachments to your posts; instead upload your files at a file sharing site (for example http://ge.tt/ ) and include a link in your message.
> For more options, visit this group at http://groups.google.com/group/ptgui
>
>
>
> --
> You received this message because you are subscribed to the Google Groups "PTGui" group.
> To post to this group, send email to pt...@googlegroups.com

> To unsubscribe from this group, send email to ptgui+unsub...@googlegroups.com

> Please do not add attachments to your posts; instead upload your files at a file sharing site (for example http://ge.tt/ ) and include a link in your message.
> For more options, visit this group at http://groups.google.com/group/ptgui

--
You received this message because you are subscribed to the Google Groups "PTGui" group.
To post to this group, send email to pt...@googlegroups.com

To unsubscribe from this group, send email to ptgui+unsub...@googlegroups.com