Automatic hot/dead pixel detection across source images
Marco Schwaer
Dear all,
this might be a feature request, although perhaps PTGui already offers something similar.
My camera sensor has a few hot and dead pixels, mostly outside the image center. Since these defective pixels remain at exactly the same sensor coordinates in every source image, PTGui should theoretically be able to identify them by analysing the complete set of imported images.
This detection would not depend on panorama overlap. The photographed content changes between the source images, but a defective sensor pixel remains fixed at the same position within each camera frame. PTGui could therefore generate a sensor defect or bad-pixel map.
The important part would be how this map is used during stitching:
Where another source image covers the same panorama area, PTGui should preferably use the valid pixel from that image instead of the flagged defective pixel.
Where no alternative source image is available, the flagged pixel should remain usable. Keeping the hot or dead pixel would be better than creating a hole or an area with missing image data.
In other words, the defect map should reduce the priority of defective pixels rather than exclude them unconditionally.
(Optionally, PTGui could interpolate such pixels in areas without alternative coverage, but this would be secondary to selecting valid pixels from overlapping source images.)
This could be particularly useful for large panorama sets, HDR/AEB sequences, high-ISO images, and long exposures.
Is anything like this already possible in PTGui?
John Houghton
Marco Schwaer
Thanks, John. Yes, I’m aware of in-camera pixel remapping and service-based sensor remapping.
My suggestion is slightly different: PTGui already has multiple source images available and could potentially identify defects that repeatedly occur at the same image coordinates, then treat those areas as lower-priority during blending.
This would not replace sensor remapping, but it could provide an additional safeguard, especially when processing older images or files from cameras for which remapping was not available or had not yet been performed.
The same approach might even help with recurring dust spots or other sensor-related artefacts: where overlapping source images provide clean image data, PTGui could preferentially use that data instead of the affected area.
I don’t know how complex this would be to implement internally, but conceptually it seems like a useful fit for PTGui’s existing overlap and blending logic.
Erik Krause
> This detection would not depend on panorama overlap. The photographed
> content changes between the source images, but a defective sensor pixel
> remains fixed at the same position within each camera frame.
- Identify the defective pixels in one image
- verify it repeats in other images (might be obsolete for hot/dead pixels)
- paint a tiny red mask over the defective pixels on Mask tab.
- go to Source Images tab, copy the mask (or save it to disk)
- select all images
- past the mask for all images (or load it from disk)
You can make this more convenient by creating a template. On Project
Settings tab under "Template Behaviour" you can specify what is copied
from the template (Image Masks in this case). You can even make it
permanent by choosing "Make Default" from the File Menu.
The downside is, that it might cause empty areas in your panorama, if
there are no pixels from an adjacent image to fill the gap. You work
around this if you choose "Red Masks: Use only to adjust seams" on
Blending side menu in Panorama Editor, in which case some residual
hot/dead pixels might still be in the result, or check "Fill holes" in
the same place, which would fill the holes with a color that adapts to
the holes edges.
--
Erik Krause
http://www.erik-krause.de
Marco Schwaer
PTGui Support
I had never considered this but it is a nice idea. Indeed PTGui sees
multiple images from the same camera so it can use that information I'll
add this to the wish list.
Kind regards,
Joost Nieuwenhuijse
www.ptgui.com
On 7/31/26 10:20, 'Marco Schwaer' via PTGui Support wrote:
> Dear all,
>
> this might be a feature request, although perhaps PTGui already offers
> something similar.
>
> My camera sensor has a few hot and dead pixels, mostly outside the image
> should theoretically be able to identify defective sensor pixels: they
> remain at the same sensor coordinates in every source image, while the
> actual scene content changes.
>
> Could PTGui automatically detect these persistent pixels and create a
> defect-pixel map, then use that map as a mask to interpolate or remove
> them before stitching?
> This could be particularly useful for large panorama sets, HDR/AEB
>
> Is there already a way to achieve this within PTGui?
>
>
> BR
>
> Marco
>
> --
> 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/
> ec2fd203-2b03-4067-a19d-0c850f1691dan%40googlegroups.com <https://
> groups.google.com/d/msgid/ptgui/ec2fd203-2b03-4067-
> a19d-0c850f1691dan%40googlegroups.com?utm_medium=email&utm_source=footer>.