artifacts after stitching

195 views
Skip to first unread message

mike....@gmail.com

unread,
Nov 13, 2025, 2:09:11 AMNov 13
to PTGui Support
Hello,
Look at this pano, I used PTGUI 13.3, when I align the images, there is no artifacts on the
preview of the pano editor but when I run the stitching there is some artefacts at the door (middle of the pano). I already tried to disable "find optimal seams"  but others artefacts appears in the sky region 
Screenshot 2025-11-13 075908.png
ptgui_project.pts

John Houghton

unread,
Nov 13, 2025, 3:08:58 AMNov 13
to PTGui Support
Mike, This is a common problem with fine mesh screens and similar.  You can just use a green mask to select the door on one of the images.

John

Stephan Weber

unread,
Nov 14, 2025, 2:17:31 AM (14 days ago) Nov 14
to pt...@googlegroups.com
Can confirm, using masks either green or red is so easy with Ptgui. Good for many things like walking people, either get them twice or not at all. Best make your shots with more overlap than necessary.

--
You received this message because you are subscribed to the Google Groups "PTGui Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ptgui+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ptgui/dce1cdfe-1654-407b-8da3-5f1ec43e892fn%40googlegroups.com.

mike....@gmail.com

unread,
Nov 14, 2025, 2:52:57 AM (14 days ago) Nov 14
to PTGui Support
Hello, Indeed using green mask solve the issue but I use PTGUI in command line because the rendering must be performed without manual editing. The multiband blender of PTGUI 11 does not produce this king of artifacts. 

Manuel Dahmann (kubische-panoramen-de)

unread,
Nov 14, 2025, 3:21:48 AM (14 days ago) Nov 14
to PTGui Support
I switched back to PTGui 11 for my SLR panorama work. There were constant color problems, artifacts, and other issues. Mastering that is beyond my capabilities. PTGui 11 just works without any problems.
In difficult lighting conditions, PTGui 13 is better for stitching, but then I render in PTGui 11.
I also use PTGui 13 for Insta360 panoramas—an expensive purchase just for that.

PTGui Support

unread,
Nov 14, 2025, 4:09:02 AM (14 days ago) Nov 14
to pt...@googlegroups.com
Hi Manuel,

I'd be happy to take a look if you could post a problematic project.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com
> rsWDC6ZSuIeAxsqQhDS/view?usp=sharing <https://drive.google.com/
> file/d/12K1f4Lvnf0YN-rsWDC6ZSuIeAxsqQhDS/view?usp=sharing>
>
> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com
> <mailto:ptgui+un...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/
> ptgui/9f1c10ef-1c49-4fbe-a82e-df0c157cdb83n%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/9f1c10ef-1c49-4fbe-a82e-
> df0c157cdb83n%40googlegroups.com?utm_medium=email&utm_source=footer>.

John Houghton

unread,
Nov 14, 2025, 4:12:49 AM (14 days ago) Nov 14
to pt...@googlegroups.com
On Fri, Nov 14, 2025 at 7:53 AM mike....@gmail.com <mike....@gmail.com> wrote:
Hello, Indeed using green mask solve the issue but I use PTGUI in command line because the rendering must be performed without manual editing. The multiband blender of PTGUI 11 does not produce this king of artifacts.7483-60c0-4f09-ab4b-25326540b903n%40googlegroups.com.

Mike, Indeed it doesn't, but the same multiband blender is available in PTGui V13 on the flyout menu on the Panorama Editor window. It might also help if you improved the setup of your panorama head, as there is significant parallax shift affecting the alignment of the images.

John

Manuel Dahmann (kubische-panoramen-de)

unread,
Nov 15, 2025, 11:17:44 AM (12 days ago) Nov 15
to PTGui Support
Hi Joost,

I don't use PTGui 13 anymore, so I don't have any test projects. Dont remember which projects I tried.
I always try out the new releases, but I had immediate color problems also with 13.3. Since I only use .jpg and never any HDR, this is inexplicable to me. My confidence in the color accuracy of PTGui 13 is gone. When rendering these #13 projects in #11, the color shifts disappeared. I don't want to always have to check whether the colors are correct.
I also had artifacts, probably because something was set incorrectly or the color shift problem. They also disappeared when rendering in #11.
Another problem is the subsequent addition of further images to a project, which unfortunately always changes the position and angle of the anchor image. I tried long to find a solution for this.
I'm sure there are solutions for all the problems in #13. But I want to work quickly and easily. 

MichaelG

unread,
Nov 16, 2025, 4:37:05 AM (12 days ago) Nov 16
to PTGui Support
Unfortunately I can support Manuel's observation about color problems. But first I should say I have two color managed EIZO monitors connected to the PC. That means even small color differences are visible.

The following image shows two screenshots of the detail viewer side-by-side. Left with blending mode "no overlap", right with "multi band" selected. The seam goes exactly through the middle of the marked frame. On the left screenshot you'll see the left window in the frame is a bit greenish, the right more gray. On the right screenshot bot windows seem to have the same colors.

DetailViewer.jpg

Below you'll find a screenshot that shows a crop of the original images side-by-side. No visible color difference, no measurable difference with a color picker.

OrgImages.jpg

I have panoramas with even bigger (not to say extreme) color differences but don't find them right now. As far as I remember the solution was always to use multi-band blending.

Michael

MichaelG

unread,
Nov 16, 2025, 4:50:25 AM (12 days ago) Nov 16
to PTGui Support
Just in case you want to try out that on your own, you can download the project and the files here. I converted the TIFF files to JPG and have the feeling the effect is even bigger. The middle of the marked windows is redish now.

PTGui Support

unread,
Nov 16, 2025, 6:50:44 AM (11 days ago) Nov 16
to pt...@googlegroups.com
Hi Michael,

This is to be expected in this case, it's how the zero overlap blender
works.

The thin vertical red line is the seam. As you can see it runs right
through the edge of the window frame. There is some green fringing in
your images. The location and extent of the green fringing differs
slightly in the images at the overlap.

The zero overlap blender tries to equalize the colors at both sides of
the seam. If one side of the seam is greener than the other side, the
blender will create a gradual overlap, by making one side of the seam a
little more greenish.

One solution is to enable optimum seams: this will try to run the seam
across an area with little color difference.

Another is to use Multiband blending. This blender doesn't blend at the
edges of the seam; rather it uses the entire overlap area and creates a
smooth blend in that area. It may improve the blend here but it suffers
from other problems such as ghosting.

And of course if you disable blending altogther, the color of the
panorama will match the colors of the source images exactly. This may
seem silly advise but it shows the trade-off required to get a smoothly
blended panorama.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

On 11/16/25 10:37, 'MichaelG' via PTGui Support wrote:
> Unfortunately I can support Manuel's observation about color problems.
> But first I should say I have two color managed EIZO monitors connected
> to the PC. That means even small color differences are visible.
>
> The following image shows two screenshots of the detail viewer side-by-
> side. Left with blending mode "no overlap", right with "multi band"
> selected. The seam goes exactly through the middle of the marked frame.
> On the left screenshot you'll see the left window in the frame is a bit
> greenish, the right more gray. On the right screenshot bot windows seem
> to have the same colors.
>
> DetailViewer.jpg
>
> Below you'll find a screenshot that shows a crop of the original images
> side-by-side. No visible color difference, no measurable difference with
> a color picker.
>
> ab4b-25326540b903n%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/cbd47483-60c0-4f09-
> ab4b-25326540b903n%40googlegroups.com?
> utm_medium=email&utm_source=footer>.
>
>
> Mike, Indeed it doesn't, but the same multiband blender is
> available in PTGui V13 on the flyout menu on the Panorama Editor
> window. It might also help if you improved the setup of your
> panorama head, as there is significant parallax shift affecting
> the alignment of the images.
>
> John
>
> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com
> <mailto:ptgui+un...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/
> ptgui/268f0b12-c497-4280-8ecc-6ed7bd2ae9ean%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/268f0b12-
> c497-4280-8ecc-6ed7bd2ae9ean%40googlegroups.com?
> utm_medium=email&utm_source=footer>.

PTGui Support

unread,
Nov 16, 2025, 6:58:11 AM (11 days ago) Nov 16
to pt...@googlegroups.com
On 11/16/25 10:37, 'MichaelG' via PTGui Support wrote:
> I have panoramas with even bigger (not to say extreme) color differences
> but don't find them right now. As far as I remember the solution was
> always to use multi-band blending.
Sometimes that's good advice, but here's an example of the contrary:
https://ptgui.com/whatsnew.html?version=120000

Scroll down to the 'Zero overlap blending' section. As you can see, zero
overlap blending works much better here.

If there's any difference in the overlap area (for example due to moving
objects), Multiband blending will suffer from ghost images, because it
merges over the entire overlap area.

Zero overlap blending doesn't suffer from this because it only uses
pixels at either side of the seam. It doesn't merge pixels from
overlapping images.

Andrew Mumford

unread,
Nov 16, 2025, 3:45:46 PM (11 days ago) Nov 16
to pt...@googlegroups.com
Sorry for this noob question but I’m trying to follow and understand this thread …

Are the color problems not just due to the fact that the two pictures are two different colors ? Possibly due to white balance or lens flare issues because of the extreme brightness outside the window ?

Thx

___

Andrew Mumford
> --
> You received this message because you are subscribed to the Google Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to ptgui+un...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/ptgui/7b42c7da-0c2c-4a5e-8c2c-ba0cf4f8f202%40ptgui.com.

Andrew Mumford

unread,
Nov 16, 2025, 3:51:21 PM (11 days ago) Nov 16
to PTGui Support

I should have specified this is referring to the images from Michael G...

Andrew Mumford

unread,
Nov 16, 2025, 4:05:52 PM (11 days ago) Nov 16
to PTGui Support
I'm sorry please ignore my post's - I understand Joost's reply now.

MichaelG

unread,
Nov 17, 2025, 3:26:36 AM (11 days ago) Nov 17
to PTGui Support
Hi Joost,

thx for the reply. I do know all that. Lets have a slightly different look at the situation. 

I created 3 output files. The only difference is the blending mode. Lets take "no blend" as the reference for a moment. Next I created the difference between "no blend" and "Multiband". I think we can agree that we do not expect big differences across the whole image - especially at the seam. And that's exactly what the result is. Ok so far.

MultiBandNoBlend.jpg

Next, I created the difference between "no blend" and "no overlap". Here I get an unexpected sharp edge at the seam. And that's exactly the color difference I mentioned in the final image created with "no overlap".

MultiBandNoOverlap.jpg

Michael

MichaelG

unread,
Nov 17, 2025, 3:32:16 AM (11 days ago) Nov 17
to PTGui Support
Sorry, the caption of the second image is wrong. Its of course "difference "no blend" - "no overlap".

Erik Krause

unread,
Nov 17, 2025, 6:31:36 AM (10 days ago) Nov 17
to pt...@googlegroups.com
Am 17.11.25 um 09:26 schrieb 'MichaelG' via PTGui Support:

> Next, I created the difference between "no blend" and "no overlap". Here I
> get an unexpected sharp edge at the seam. And that's exactly the color
> difference I mentioned in the final image created with "no overlap".

This is to be expected. "no overlap" takes pixels either from one image
(on one side of the seam) or from the other image (on the other side).
In order to create a seamless blend, the blender changes color and
brightness of pixels on both sides of the seam such that there is no
difference at the seam. This eliminates ghosting altogether.

"no overlap" works best with "Find optimum seam", since this places the
seam where there are the least differences between the images.

The multi band blender uses a weighted average of pixels from both
images throughout the overlap. But other than a simple feather, where
only the distance from the seam determines the weight of a pixel, it
uses a pyramid approach, essentially creating a feather of variable
width depending on the image details. This reduces, but does not
eliminate, the probability of ghosting. Details see here:
https://persci.mit.edu/pub_pdfs/spline83.pdf

Your difference images show, that the two adjacent images have slightly
different colors, most likely due to a white balance difference, or due
to lighting differences.

--
Erik Krause
http://www.erik-krause.de

PTGui Support

unread,
Nov 17, 2025, 8:33:10 AM (10 days ago) Nov 17
to pt...@googlegroups.com
Hi Michael,

Creating difference images is not a good way to judge blending. The
unblended image always wins, but it looks terribly.

I'd say the blended results look so good actually because of these
adjustments.

Here, the entire image has a different color or brightness than the
other images, and it is fully enclosed by other images, and there is a
consistent color or brightness difference at the seams. Then, yes, the
blender will adjust the entire image. This is how the zero overlap
blender works, and this is why the results (in general) look so good IMO.

Think of it this way: the blender needs to make some adjustments in
order to get a smooth blend. Suppose you want to keep the center of the
image unchanged (which you seem to desire). The blender would then need
to apply gradients from the center of the image to the edge. But how
wide? If this gradient is narrow then the adjustments would be visible.
If the gradient is wide then you would post the same complaint about the
blender adjusting too much.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

On 11/17/25 09:26, 'MichaelG' via PTGui Support wrote:
> Hi Joost,
>
> thx for the reply. I do know all that. Lets have a slightly different
> look at the situation.
>
> I created 3 output files. The only difference is the blending mode. Lets
> take "no blend" as the reference for a moment. Next I created the
> difference between "no blend" and "Multiband". I think we can agree that
> we do not expect big differences across the whole image - especially at
> the seam. And that's exactly what the result is. Ok so far.
>
> MultiBandNoBlend.jpg
>
> Next, I created the difference between "no blend" and "no overlap". Here
> I get an unexpected sharp edge at the seam. And that's exactly the color
> difference I mentioned in the final image created with "no overlap".
>
> > https://ptgui.com/whatsnew.html?version=120000 <https://
> ptgui.com/whatsnew.html?version=120000>
> >
> > Scroll down to the 'Zero overlap blending' section. As
> you can see, zero overlap blending works much better here.
> >
> > If there's any difference in the overlap area (for
> example due to moving objects), Multiband blending will
> suffer from ghost images, because it merges over the entire
> overlap area.
> >
> > Zero overlap blending doesn't suffer from this because it
> only uses pixels at either side of the seam. It doesn't
> merge pixels from overlapping images.
> >
> > Kind regards,
> >
> > Joost Nieuwenhuijse
> > www.ptgui.com <http://www.ptgui.com>
> >
> > --
> > You received this message because you are subscribed to
> the Google Groups "PTGui Support" group.
> > To unsubscribe from this group and stop receiving emails
> from it, send an email to ptgui+un...@googlegroups.com.
> > To view this discussion visit https://groups.google.com/
> d/msgid/ptgui/7b42c7da-0c2c-4a5e-8c2c-
> ba0cf4f8f202%40ptgui.com <https://groups.google.com/d/msgid/
> ptgui/7b42c7da-0c2c-4a5e-8c2c-ba0cf4f8f202%40ptgui.com>.
>
> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com
> <mailto:ptgui+un...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/
> ptgui/190127c6-43e2-46e2-a7c3-d8e569e886c1n%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-a7c3-
> d8e569e886c1n%40googlegroups.com?utm_medium=email&utm_source=footer>.

PTGui Support

unread,
Nov 17, 2025, 8:48:56 AM (10 days ago) Nov 17
to pt...@googlegroups.com
Also, have you verified that the seam does not run entirely through an
area affected by CA or purple fringing? The seam seems to coincide with
the window frame for a large area. Edge-based blending will adjust the
fringing of one image to the fringing of another image.

If that's the case, try enabling optimum seams, the result will look
totally different.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

MichaelG

unread,
Nov 20, 2025, 3:48:42 AM (8 days ago) Nov 20
to PTGui Support
Of course I checked the seam position. Here it is marked in the source images. There is only very little CA at that position, even at magnification of 500%. 

SeamPosition.jpg

Optimum seams don't help. The position of the "color jump" is just moved to another position. In the meantime I found a second panorama that shows the same effect. To emphasize the effect I increased the color saturation in PTGui to 2005. In unblended mode, left and right of the seam are almost identical, in blended mode (Zero overlap) there is a sharp edge magenta/green.

Saturation3.jpg

For no specific reason I developed the RAW images without CA correction. When I turn on CA correction the problem disappears. Loading the RAW files and turning on CA correction in PTGui doesn't help at all.

I found out another really strange work-around with CA correction during RAW development turned off during RAW development:
1: load the images
2: goto to crop and reduce the cropping circle a little bit (below left: original, right: small reduction) at the left-hand side of the image only => the colors left and right of the seam match
3: increase the cropping circle to the original size => the colors still match, no magenta/green edge

crop.jpg

Michael

PTGui Support

unread,
Nov 20, 2025, 4:12:23 AM (8 days ago) Nov 20
to pt...@googlegroups.com
Hi Michael,

But if I look at the first image in your email (SeamPosition.jpg), at
the ceiling above Horizon and compare the left and right fisheye images,
there's an obvious white balance difference. The left image looks
slightly more yellow/greenish.

The zero overlap blender notices there's a consistent color difference
across the entire seam, so it adjusts the color of both images to
minimize the difference.

Isn't that the issue?

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

On 11/20/25 09:48, 'MichaelG' via PTGui Support wrote:
> Of course I checked the seam position. Here it is marked in the source
> images. There is only very little CA at that position, even at
> magnification of 500%.
>
> SeamPosition.jpg
>
> Optimum seams don't help. The position of the "color jump" is just moved
> to another position. In the meantime I found a second panorama that
> shows the same effect. To emphasize the effect I increased the color
> saturation in PTGui to 2005. In unblended mode, left and right of the
> seam are almost identical, in blended mode (Zero overlap) there is a
> sharp edge magenta/green.
>
> Saturation3.jpg
>
> For no specific reason I developed the RAW images without CA correction.
> When I turn on CA correction the problem disappears. Loading the RAW
> files and turning on CA correction in PTGui doesn't help at all.
>
> I found out another really strange work-around with CA correction during
> RAW development turned off during RAW development:
> 1: load the images
> 2: goto to crop and reduce the cropping circle a little bit (below left:
> original, right: small reduction) at the left-hand side of the image
> only => the colors left and right of the seam match
> 3: increase the cropping circle to the original size => the colors still
> match, no magenta/green edge
>
> crop.jpg
>
> Michael
>
> PTGui Support schrieb am Montag, 17. November 2025 um 14:48:56 UTC+1:
>
> Also, have you verified that the seam does not run entirely through an
> area affected by CA or purple fringing? The seam seems to coincide with
> the window frame for a large area. Edge-based blending will adjust the
> fringing of one image to the fringing of another image.
>
> If that's the case, try enabling optimum seams, the result will look
> totally different.
>
> Kind regards,
>
> Joost Nieuwenhuijse
> www.ptgui.com <http://www.ptgui.com>
> > www.ptgui.com <http://www.ptgui.com>
> <https://ptgui.com/whatsnew.html?version=120000> <https://
> >> ptgui.com/whatsnew.html?version=120000 <http://ptgui.com/
> whatsnew.html?version=120000>>
> >>              >
> >>              > Scroll down to the 'Zero overlap blending'
> section. As
> >>             you can see, zero overlap blending works much better
> here.
> >>              >
> >>              > If there's any difference in the overlap area (for
> >>             example due to moving objects), Multiband blending will
> >>             suffer from ghost images, because it merges over the
> entire
> >>             overlap area.
> >>              >
> >>              > Zero overlap blending doesn't suffer from this
> because it
> >>             only uses pixels at either side of the seam. It doesn't
> >>             merge pixels from overlapping images.
> >>              >
> >>              > Kind regards,
> >>              >
> >>              > Joost Nieuwenhuijse
> >>              > www.ptgui.com <http://www.ptgui.com> <http://
> www.ptgui.com <http://www.ptgui.com>>
> >>              >
> >>              > --
> >>              > You received this message because you are
> subscribed to
> >>             the Google Groups "PTGui Support" group.
> >>              > To unsubscribe from this group and stop receiving
> emails
> >>             from it, send an email to ptgui+un...@googlegroups.com.
> >>              > To view this discussion visit https://
> groups.google.com/ <https://groups.google.com/>
> >>             d/msgid/ptgui/7b42c7da-0c2c-4a5e-8c2c-
> >>             ba0cf4f8f202%40ptgui.com <http://40ptgui.com>
> <https://groups.google.com/d/msgid/ <https://groups.google.com/d/
> msgid/>
> >>             ptgui/7b42c7da-0c2c-4a5e-8c2c-
> ba0cf4f8f202%40ptgui.com <http://40ptgui.com>>.
> >>
> >> --
> >> You received this message because you are subscribed to the Google
> >> Groups "PTGui Support" group.
> >> To unsubscribe from this group and stop receiving emails from
> it, send
> >> an email to ptgui+un...@googlegroups.com
> >> <mailto:ptgui+un...@googlegroups.com>.
> >> To view this discussion visit https://groups.google.com/d/msgid/
> <https://groups.google.com/d/msgid/>
> >> ptgui/190127c6-43e2-46e2-a7c3-d8e569e886c1n%40googlegroups.com
> <http://40googlegroups.com>
> >> <https:// groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-
> a7c3- <http://groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-a7c3->
> >> d8e569e886c1n%40googlegroups.com?
> utm_medium=email&utm_source=footer <http://40googlegroups.com?
> utm_medium=email&utm_source=footer>>.
> >
>
> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com
> <mailto:ptgui+un...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/ptgui/
> ba225a88-3779-49e8-ae46-df7ac80c9015n%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/ba225a88-3779-49e8-ae46-
> df7ac80c9015n%40googlegroups.com?utm_medium=email&utm_source=footer>.

MichaelG

unread,
Nov 20, 2025, 8:11:23 AM (7 days ago) Nov 20
to PTGui Support
No, that is 100% not the issue. Why should there be no difference in unblended mode.

I can do the same mentioned trick with both panoramas. Just decrease the radius of the cropping circle by some pixel and set it back and the color difference is gone. 

PTGui Support

unread,
Nov 20, 2025, 12:07:26 PM (7 days ago) Nov 20
to pt...@googlegroups.com
I'm sorry, I'm not sure if I fully understand what you mean, so you may
have lost me. But..

What I mean to explain is this, see attached screenshot: this is the
detail viewer in unblended mode. There is a clear color difference, and
the blender just tries to equalize this by adjusting the color. This
confirms what you showed in the earlier screenshot of the diff images,
there will be a significant hard jump in color adjustment at the seam,
which perfectly neutralizes the difference. One image will turn more
yellow, the other more magenta.

You said "Here I get an unexpected sharp edge at the seam." but the
sharp edge is expected, this is exactly how the zero overlap blending
works. The sharp edge is in your source images, and the blender applies
an opposite sharp edge.

Your other point: reverting between two crop settings A->B->A should
give the same blended result, blending is fully deterministic (esp with
Optimum Seams disabled). I tried and I couldn't notice anything strange,
and I'm surprised you see something else. Have you tried rendering the
panorama twice (and also save the before and after project to make sure
they are identical)?

Don't judge blending in the detail viewer btw. This renders and blends
only a small section of the panorama and ignores objects just outside
the viewport.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

On 11/20/25 14:11, 'MichaelG' via PTGui Support wrote:
> No, that is 100% not the issue. Why should there be no difference in
> unblended mode.
>
> I can do the same mentioned trick with both panoramas. Just decrease the
> radius of the cropping circle by some pixel and set it back and the
> color difference is gone.
>
> PTGui Support schrieb am Donnerstag, 20. November 2025 um 10:12:23 UTC+1:
>
> Hi Michael,
>
> But if I look at the first image in your email (SeamPosition.jpg), at
> the ceiling above Horizon and compare the left and right fisheye
> images,
> there's an obvious white balance difference. The left image looks
> slightly more yellow/greenish.
>
> The zero overlap blender notices there's a consistent color difference
> across the entire seam, so it adjusts the color of both images to
> minimize the difference.
>
> Isn't that the issue?
>
> Kind regards,
>
> Joost Nieuwenhuijse
> www.ptgui.com <http://www.ptgui.com>
> > www.ptgui.com <http://www.ptgui.com> <http://www.ptgui.com
> > > www.ptgui.com <http://www.ptgui.com> <http://www.ptgui.com
> > <https://ptgui.com/whatsnew.html?version=120000 <https://
> whatsnew.html?version=120000> <http://ptgui.com/ <http://ptgui.com/>
> > whatsnew.html?version=120000>>
> > >>              >
> > >>              > Scroll down to the 'Zero overlap blending'
> > section. As
> > >>             you can see, zero overlap blending works much better
> > here.
> > >>              >
> > >>              > If there's any difference in the overlap area (for
> > >>             example due to moving objects), Multiband blending
> will
> > >>             suffer from ghost images, because it merges over the
> > entire
> > >>             overlap area.
> > >>              >
> > >>              > Zero overlap blending doesn't suffer from this
> > because it
> > >>             only uses pixels at either side of the seam. It
> doesn't
> > >>             merge pixels from overlapping images.
> > >>              >
> > >>              > Kind regards,
> > >>              >
> > >>              > Joost Nieuwenhuijse
> > >>              > www.ptgui.com <http://www.ptgui.com> <http://
> www.ptgui.com <http://www.ptgui.com>> <http://
> > www.ptgui.com <http://www.ptgui.com> <http://www.ptgui.com
> <http://www.ptgui.com>>>
> > >>              >
> > >>              > --
> > >>              > You received this message because you are
> > subscribed to
> > >>             the Google Groups "PTGui Support" group.
> > >>              > To unsubscribe from this group and stop receiving
> > emails
> > >>             from it, send an email to
> ptgui+un...@googlegroups.com.
> > >>              > To view this discussion visit https://
> > groups.google.com/ <http://groups.google.com/> <https://
> groups.google.com/ <https://groups.google.com/>>
> > >>             d/msgid/ptgui/7b42c7da-0c2c-4a5e-8c2c-
> > >>             ba0cf4f8f202%40ptgui.com <http://40ptgui.com>
> <http://40ptgui.com <http://40ptgui.com>>
> > <https://groups.google.com/d/msgid/ <https://groups.google.com/d/
> msgid/> <https://groups.google.com/d/ <https://groups.google.com/d/>
> > msgid/>
> > >>             ptgui/7b42c7da-0c2c-4a5e-8c2c-
> > ba0cf4f8f202%40ptgui.com <http://40ptgui.com> <http://40ptgui.com
> <http://40ptgui.com>>>.
> > >>
> > >> --
> > >> You received this message because you are subscribed to the
> Google
> > >> Groups "PTGui Support" group.
> > >> To unsubscribe from this group and stop receiving emails from
> > it, send
> > >> an email to ptgui+un...@googlegroups.com
> > >> <mailto:ptgui+un...@googlegroups.com>.
> > >> To view this discussion visit https://groups.google.com/d/
> msgid/ <https://groups.google.com/d/msgid/>
> > >> ptgui/190127c6-43e2-46e2-a7c3-d8e569e886c1n%40googlegroups.com
> <http://40googlegroups.com>
> > <http://40googlegroups.com <http://40googlegroups.com>>
> > >> <https:// groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-
> <http://groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2->
> > a7c3- <http://groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-
> a7c3- <http://groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-
> a7c3->>
> > >> d8e569e886c1n%40googlegroups.com <http://40googlegroups.com>?
> > utm_medium=email&utm_source=footer <http://40googlegroups.com
> <http://40googlegroups.com>?
> > utm_medium=email&utm_source=footer>>.
> > >
> >
> > --
> > You received this message because you are subscribed to the Google
> > Groups "PTGui Support" group.
> > To unsubscribe from this group and stop receiving emails from it,
> send
> > an email to ptgui+un...@googlegroups.com
> > <mailto:ptgui+un...@googlegroups.com>.
> > To view this discussion visit https://groups.google.com/d/msgid/
> ptgui/ <https://groups.google.com/d/msgid/ptgui/>
> > ba225a88-3779-49e8-ae46-df7ac80c9015n%40googlegroups.com
> <http://40googlegroups.com> <https://
> > groups.google.com/d/msgid/ptgui/ba225a88-3779-49e8-ae46- <http://
> groups.google.com/d/msgid/ptgui/ba225a88-3779-49e8-ae46->
> > df7ac80c9015n%40googlegroups.com?
> utm_medium=email&utm_source=footer <http://40googlegroups.com?
> utm_medium=email&utm_source=footer>>.
>
> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com
> <mailto:ptgui+un...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/
> ptgui/500601bf-284c-4001-8904-d1377a29c8c9n%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/500601bf-284c-4001-8904-
> d1377a29c8c9n%40googlegroups.com?utm_medium=email&utm_source=footer>.
Screenshot From 2025-11-20 17-36-27.png

MichaelG

unread,
Nov 21, 2025, 3:14:33 AM (7 days ago) Nov 21
to PTGui Support
Oh, no its not at all about the slightly visible seam that is sometimes visible by the zero-overlap-blender. That is totally ok, part of the idea of zero overlap and well understood.  In other words, it is not about the seam as a principle of the blender. It is about the fact that the blender changes in rare cases the colors significantly.

I didn't mention that I found in the meantime a second panorama that shows the effect even better. Below you'll 2 crops of the final created panoramas. On top "no blend", below "zero overlap" selected. Without any interaction from me the seam goes exactly through to pole. "no blend" creates almost the same colors (some shade of gray) left and right from the seam (the pole). "zero overlap" instead creates an intense magenta gradient starting at the seam, going to left and a green gradient also starting at the seam going to right. Instead of aligning the colors left and right of the seam, magenta and green collide at the seam.

comparison.jpg

The funny think is when I add the following very small mask both gradients (magenta AND green) are almost gone (same as reducing the crop circle). The masked part (and even a bigger portion of the area) isn't used in the final panorama.
mask.jpg

Should we continue this topic via email?
Michael

PTGui Support

unread,
Nov 21, 2025, 3:49:01 AM (7 days ago) Nov 21
to pt...@googlegroups.com
Hi Michael,

There's a clear CA / purple fringing visible along the edge of the black
pole. The gradient after blending is caused by the seam coinciding
exactly with the fringing.

The fringing has a different color in the two overlapping images. Zero
Overlap blending exclusively looks at the colors at either edge of the
seam, it doesn't use the overlap region. If the fringing is green in one
picture and purple in the other, and the seam coincides with the
fringing for a significant part, then the blender will apply this
gradient to correct this.

This obviously is a weakness of the algorithm, but blending is a trade
off. Multiband doesn't suffer from this but it has other weaknesses.

I highly recommend enabling Optimum Seams, this should fix the issue.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

On 11/21/25 09:14, 'MichaelG' via PTGui Support wrote:
> Oh, no its not at all about the slightly visible seam that is sometimes
> visible by the zero-overlap-blender. That is totally ok, part of the
> idea of zero overlap and well understood. In other words, it is not
> about the seam as a principle of the blender. It is about the fact that
> the blender changes in rare cases the colors significantly.
>
> I didn't mention that I found in the meantime a second panorama that
> shows the effect even better. Below you'll 2 crops of the final created
> panoramas. On top "no blend", below "zero overlap" selected. Without any
> interaction from me the seam goes exactly through to pole. "no blend"
> creates almost the same colors (some shade of gray) left and right from
> the seam (the pole). "zero overlap" instead creates an intense magenta
> gradient starting at the seam, going to left and a green gradient also
> starting at the seam going to right. Instead of aligning the colors left
> and right of the seam, magenta and green collide at the seam.
>
> comparison.jpg
>
> The funny think is when I add the following very small mask both
> gradients (magenta AND green) are almost gone (same as reducing the crop
> circle). The masked part (and even a bigger portion of the area) isn't
> used in the final panorama.
> www.ptgui.com <http://www.ptgui.com>
>
> On 11/20/25 14:11, 'MichaelG' via PTGui Support wrote:
> > No, that is 100% not the issue. Why should there be no difference in
> > unblended mode.
> >
> > I can do the same mentioned trick with both panoramas. Just
> decrease the
> > radius of the cropping circle by some pixel and set it back and the
> > color difference is gone.
> >
> > PTGui Support schrieb am Donnerstag, 20. November 2025 um
> 10:12:23 UTC+1:
> >
> > Hi Michael,
> >
> > But if I look at the first image in your email
> (SeamPosition.jpg), at
> > the ceiling above Horizon and compare the left and right fisheye
> > images,
> > there's an obvious white balance difference. The left image looks
> > slightly more yellow/greenish.
> >
> > The zero overlap blender notices there's a consistent color
> difference
> > across the entire seam, so it adjusts the color of both images to
> > minimize the difference.
> >
> > Isn't that the issue?
> >
> > Kind regards,
> >
> > Joost Nieuwenhuijse
> <http://www.ptgui.com>> <http://www.ptgui.com <http://www.ptgui.com>
> <http://www.ptgui.com>> <http://www.ptgui.com <http://www.ptgui.com>
> > whatsnew.html?version=120000> <http://ptgui.com/ <http://
> ptgui.com/> <http://ptgui.com/ <http://ptgui.com/>>
> <http://www.ptgui.com>> <http://www.ptgui.com <http://www.ptgui.com>
> > <http://www.ptgui.com <http://www.ptgui.com>>>>
> > > >>              >
> > > >>              > --
> > > >>              > You received this message because you are
> > > subscribed to
> > > >>             the Google Groups "PTGui Support" group.
> > > >>              > To unsubscribe from this group and stop
> receiving
> > > emails
> > > >>             from it, send an email to
> > ptgui+un...@googlegroups.com.
> > > >>              > To view this discussion visit https://
> > > groups.google.com/ <http://groups.google.com/> <http://
> d/msgid/> <https://groups.google.com/d/ <https://groups.google.com/d/>
> > msgid/> <https://groups.google.com/d/ <https://groups.google.com/
> d/> <https://groups.google.com/d/ <https://groups.google.com/d/>>
> > > msgid/>
> > > >>             ptgui/7b42c7da-0c2c-4a5e-8c2c-
> > > ba0cf4f8f202%40ptgui.com <http://40ptgui.com>
> <http://40ptgui.com <http://40ptgui.com>> <http://40ptgui.com
> <http://40ptgui.com>
> > <http://40ptgui.com <http://40ptgui.com>>>>.
> > > >>
> > > >> --
> > > >> You received this message because you are subscribed to the
> > Google
> > > >> Groups "PTGui Support" group.
> > > >> To unsubscribe from this group and stop receiving emails from
> > > it, send
> > > >> an email to ptgui+un...@googlegroups.com
> > > >> <mailto:ptgui+un...@googlegroups.com>.
> > > >> To view this discussion visit https://groups.google.com/d/
> <https://groups.google.com/d/>
> > msgid/ <https://groups.google.com/d/msgid/ <https://
> > > >> ptgui/190127c6-43e2-46e2-a7c3-
> <http://40googlegroups.com <http://40googlegroups.com>>>
> > > >> <https:// groups.google.com/d/msgid/
> ptgui/190127c6-43e2-46e2- <http://groups.google.com/d/msgid/
> ptgui/190127c6-43e2-46e2->
> > <http://groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-
> <http://groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2->>
> > > a7c3- <http://groups.google.com/d/msgid/
> ptgui/190127c6-43e2-46e2- <http://groups.google.com/d/msgid/
> ptgui/190127c6-43e2-46e2->
> > a7c3- <http://groups.google.com/d/msgid/ptgui/190127c6-43e2-46e2-
> > > utm_medium=email&utm_source=footer <http://40googlegroups.com
> <http://40googlegroups.com>
> > <http://40googlegroups.com <http://40googlegroups.com>>?
> > > utm_medium=email&utm_source=footer>>.
> > > >
> > >
> > > --
> > > You received this message because you are subscribed to the Google
> > > Groups "PTGui Support" group.
> > > To unsubscribe from this group and stop receiving emails from it,
> > send
> > > an email to ptgui+un...@googlegroups.com
> > > <mailto:ptgui+un...@googlegroups.com>.
> > > To view this discussion visit https://groups.google.com/d/
> msgid/ <https://groups.google.com/d/msgid/>
> > ptgui/ <https://groups.google.com/d/msgid/ptgui/ <https://
> groups.google.com/d/msgid/ptgui/>>
> > > ba225a88-3779-49e8-ae46-df7ac80c9015n%40googlegroups.com
> > > groups.google.com/d/msgid/ptgui/ba225a88-3779-49e8-ae46-
> <http://groups.google.com/d/msgid/ptgui/ba225a88-3779-49e8-ae46->
> <http://
> > groups.google.com/d/msgid/ptgui/ba225a88-3779-49e8-ae46- <http://
> groups.google.com/d/msgid/ptgui/ba225a88-3779-49e8-ae46->>
> > > df7ac80c9015n%40googlegroups.com <http://40googlegroups.com>?
> > utm_medium=email&utm_source=footer <http://40googlegroups.com
> <http://40googlegroups.com>?
> > utm_medium=email&utm_source=footer>>.
> >
> > --
> > You received this message because you are subscribed to the Google
> > Groups "PTGui Support" group.
> > To unsubscribe from this group and stop receiving emails from it,
> send
> > an email to ptgui+un...@googlegroups.com
> > <mailto:ptgui+un...@googlegroups.com>.
> > To view this discussion visit https://groups.google.com/d/msgid/
> <https://groups.google.com/d/msgid/>
> > ptgui/500601bf-284c-4001-8904-d1377a29c8c9n%40googlegroups.com
> <http://40googlegroups.com> <https://
> > groups.google.com/d/msgid/ptgui/500601bf-284c-4001-8904- <http://
> groups.google.com/d/msgid/ptgui/500601bf-284c-4001-8904->
> > d1377a29c8c9n%40googlegroups.com?
> utm_medium=email&utm_source=footer <http://40googlegroups.com?
> utm_medium=email&utm_source=footer>>.
>
> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com
> <mailto:ptgui+un...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/ptgui/
> faa74fbb-8368-4e39-9e66-6b8600bd987bn%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/
> faa74fbb-8368-4e39-9e66-6b8600bd987bn%40googlegroups.com?
> utm_medium=email&utm_source=footer>.

MichaelG

unread,
Nov 23, 2025, 2:25:36 AM (5 days ago) Nov 23
to PTGui Support
Hi Joost,

got it! You are right, the pole shows a good amount of color fringing, here at 500% magnification:

Pole.jpg

These panoramas don't require best quality, that's why I didn't apply any noise reduction (and most likely no CA correction). My bad luck was that the seam goes exactly through the pole. I never thought about that and I learned a lot.

The images also suffer from (severe) ISO noise. Once knowing the blender better it was no surprise anymore that a good ISO noise reduction further improves the result of the blender (reducing the chance of picking up the wrong color).

Side note: The Laowa 4mm is a good M43 lens. But it isn't sharp enough to use the entire frame. That's why "optimal seems" isn't the first choice. But I can handle that with masks very well.

Thanks a lot for your detailed explanations and your patience!

Michael
Reply all
Reply to author
Forward
0 new messages