Page 1 of 1

Adding control points manually

PostPosted: Tue Jan 22, 2013 12:06 am
by APG-Unit
After many hours wasted and an exhaustive search on "Hard Linking" images, "Manually adding CPs", "Local Geometry Analysis, "Local optimization" etc. with no luck I've decided to ask the community for help. What has been covered in other posts has only grazed the surface and I have had no luck whatsoever.

I'm assembling some ~200 images and have only 1 problem. I have two subpanos and need to link two images together. Instead of providing the whole pano I will just provide the two images which need to be linked. Perhaps some one can give it a shot and if they manage to assemble them could provide steps and settings to show how.


All the best of luck and many thanks to anyone willing to give this a shot.

P.S. The overlapping area with "prominent" features Is located on the bottom of each photograph

PostPosted: Tue Jan 22, 2013 12:55 am
by klausesser
APG-Unit wrote:P.S. The overlapping area with "prominent" features Is located on the bottom of each photograph

??? What do you mean by that? They don´t look ajacent at all, right, and they´re of very different sharpness.

best, Klaus

PostPosted: Tue Jan 22, 2013 1:43 am
by APG-Unit
I have highlighted the prominent features showing two features which are "clearly" in both of the images (A+B)

I had to crop these images due to the size requirements of this forum. I know the images have different sharpness etc. but this is what I have to work with unfortunately.

I have tried hard linking the images, as well as manually adding CPs and then locally optimizing to no avail.

PostPosted: Tue Jan 22, 2013 10:02 am
by mediavets
APG-Unit wrote:I have highlighted the prominent features showing two features which are "clearly" in both of the images (A+B)

I had to crop these images due to the size requirements of this forum. I know the images have different sharpness etc. but this is what I have to work with unfortunately.

I have tried hard linking the images, as well as manually adding CPs and then locally optimizing to no avail.

The images are not the same size (zoom level) is that how they are in the originals?

Plus the features you outline are not identical and would not be recognised as identical by APP/APG although you may 'know' that they are the 'same'.

PostPosted: Tue Jan 22, 2013 11:10 am
by klausesser
APG-Unit wrote:I have highlighted the prominent features showing two features which are "clearly" in both of the images (A+B)

I had to crop these images due to the size requirements of this forum. I know the images have different sharpness etc. but this is what I have to work with unfortunately.

I have tried hard linking the images, as well as manually adding CPs and then locally optimizing to no avail.

Don´t crop them - shrink them. They need to be of identical size.

best, Klaus

PostPosted: Tue Jan 22, 2013 5:41 pm
by marzipano
It is possible to add 1 pair of manual control points to an overlap area where there is a recognizable point of similarity in the two images followed by optimization and rendering to give a single stitched image.

The blur in one image is going to be a problem whatever you do

One of the 2 images could be resized to be the same size as the other one to get better results on the stitch

This is what you get if you just stitch "as is"

PostPosted: Tue Jan 22, 2013 8:07 pm
by APG-Unit
Yes the zoom level is often different from one photograph to another but in the majority of other cases as long as there has been enough CPs (with an acceptable RMS) APG has had not much of an issue with stitching. The fact that one image is blurry isn't a big deal; I'm not worried about visual clarity but I am about accuracy (I.e. nothing being left out of the mosaic)

Klaus: I assume you mean using another program to shrink one image so the zoom level becomes more similar to the rest of the photos. Is there a way to tell what zoom level is or is it trial and error thing (sorry, I'm not very photo savvy)?

Marzipano: So when you added CPs manually you mention it was "1 pair" do you mean ONLY one CP in each photo? Is it not better to add multiple CPS manually, preferably spread out to give APG a better idea of how to optimize? My second question is do you use the quick optimize, full optimize or right-click and local optimize (to N neighbour)?

PostPosted: Tue Jan 22, 2013 8:45 pm
by klausesser
APG-Unit wrote:Klaus: I assume you mean using another program to shrink one image so the zoom level becomes more similar to the rest of the photos. Is there a way to tell what zoom level is or is it trial and error thing (sorry, I'm not very photo savvy)?

I mean that the images should have the same size. You can´t stitch THAT big differences in size.

Use the image´s original resolution for alll images. Preferablyuse RAW and process the RAW in a dedicated RAW processor first. The save them as TIF or JPG and import them into APG.

By processing the RAW you can correct some issues like CAs or you can apply a bit unsharp-mask, density or color-temperature. That way you can put optimized images into the stitcher - and it will definitely give you better results.

best, Klaus

PostPosted: Tue Jan 22, 2013 9:41 pm
by marzipano
APG-Unit wrote:Yes the zoom level is often different from one photograph to another but in the majority of other cases as long as there has been enough CPs (with an acceptable RMS) APG has had not much of an issue with stitching. The fact that one image is blurry isn't a big deal; I'm not worried about visual clarity but I am about accuracy (I.e. nothing being left out of the mosaic)

Klaus: I assume you mean using another program to shrink one image so the zoom level becomes more similar to the rest of the photos. Is there a way to tell what zoom level is or is it trial and error thing (sorry, I'm not very photo savvy)?

Marzipano: So when you added CPs manually you mention it was "1 pair" do you mean ONLY one CP in each photo? Is it not better to add multiple CPS manually, preferably spread out to give APG a better idea of how to optimize? My second question is do you use the quick optimize, full optimize or right-click and local optimize (to N neighbour)?

I meant just one pair (as that was all I could find !) but I agree more would probably be better but I don't think you would need that many. I found another pair so this example has 2 pairs. I did a full optimise in answer to your other question

PostPosted: Thu Jan 24, 2013 10:02 am
by ThomasV
Hello,

You can add manually control points between 2 images. Just select the 2 images in the control point editor, and select the manual CP adding tool.
To add a pair of point, click once in image 1 (a blue N appears), then click the image 2 at the matching point. The CP now appears in the 2 images.

Once you add all your pairs, click 'Quick optimize'.

Hope it helps.

Regards,
Thomas