Identify which photos contribute to pixel after merge

[kankamuso]

Hi, all

I have used Photoshop to merge photos for a long time. Nevertheless, I am currently using it for a technical project and would need to know which photo/photos is/are contributing to a certain pixel in the merged image. Is this even possible using Photoshop? If not, could you point me to some stitching software that can do it? Preferably for free as this is a personal and one time project.

Thanks a lot in advance,

Jose

 

[thebestcpu]

Hi Jose
The information you provided leaves a fairly incomplete picture of the problem you are trying to solve.
- Do you have access to all the original images and the final merged image?
- Do you have the origianl alignment of all the contributing images
- When the images were merged, did each one contribute to their own separate or did they each contribute to each merged pixel.
- Where they merged with a simple combination such as varying transparency values or were they merged with a particular blend mode
- If a blend mode was involved, what was the bend mode used
- Is any other action applied to the layers to achieve the final merged Layer

For any hope of finding the result that you seek, I believe one would need to know the maximum amount of information of how the images were merged. If the conditions of merging are rather limited, it might be possible to deduce which layers contributed. Yet if many Layers were involved in creating a given pixel, it could prove difficult or impossible.

Just some initial thoughts
John Wheeler

 

[kankamuso]

Hi Jose
The information you provided leaves a fairly incomplete picture of the problem you are trying to solve.
- Do you have access to all the original images and the final merged image?
- Do you have the origianl alignment of all the contributing images
- When the images were merged, did each one contribute to their own separate or did they each contribute to each merged pixel.
- Where they merged with a simple combination such as varying transparency values or were they merged with a particular blend mode
- If a blend mode was involved, what was the bend mode used
- Is any other action applied to the layers to achieve the final merged Layer

For any hope of finding the result that you seek, I believe one would need to know the maximum amount of information of how the images were merged. If the conditions of merging are rather limited, it might be possible to deduce which layers contributed. Yet if many Layers were involved in creating a given pixel, it could prove difficult or impossible.

Just some initial thoughts
John Wheeler

Thanks, John,

I have a bunch of thermal images from solar panels taken from a UAV. Due to limited resolution (640x480), you need to get the images using a rather hight overlap (80%). Therefore, each pixel in the resulting merged image is most probably resulting from two or more original images. The thing is that the merged image allows to find a problem in the panels but you need to identify which unit is failing on-site. This is only possible if you know which original image is contributing to that special pixel (or area) on the merged image, if that makes sense. So, I was thinking if PS kept trace of the original images contributing to the merge.

Bestest

 

[thebestcpu]

Hi Jose
If all you have is a final merged image (without the original image Layers), I don't think there is any embedded information that would be of help.
If you have the original Layered file (e.g. TIFF or PSD with all Layers), then that is a different matter.

From your description it did not seem that you had the Layered files yet if you do, why not share the file for folks to give it a look and provide more targeted suggestions. It might be too large to share directly on the Forum yet could be done with a file sharing site e.g. Dropbox etc.

John Wheeler

 

[admin]

Sounds like you may be using DroneDeploy or similar software to stitch multiple aerial images together after flying a grid.

If that is the case, which software exactly are you using? I'm guessing the stitching of your images is not done by Photoshop, and therefore it would be difficult for Photoshop to tell you which original image makes up a selected portion of your final composited image.

If the stitching software (other than Ps) cannot tell you which image it is, then you would need to visually compare the selected portion of your composite image to each original photo in a batch, to identify the most likely match. This is a task probably more suited to an image comparison algorithm than to Photoshop.

If I'm wrong and the stitching is being done in Photoshop, then it should be fairly straightforward to tell which image any pixel is from by keeping the originals on different layers when the composite is made, then using auto select to find the layer of the portion you need to identify.