I am having some difficulty registering a series of images; I have a photo database of total of 837 images. After some experimentation with alignment I have realized a component of 810 registered images. My guess is that the 27 non-registered images have low features. I do want to include the non-registered images in my project due to viable information.
how can I register the non-registered images, in any case get them to form into components?
I have tried various ways with adding control points, in various places and in different ways. Adding cp’s manually and also via the control point application way in the 3D component/2D view combined (probably the new tool). Control points added in the large component, went back to the 2D non-registered images and placed the same control point in same places and run alignment. Is this the way to register images and make them available for alignment?
Also when applying a control point in 3 or 4 images will automatically cause other images to follow and make the green plus appear in all images(1D). This says ”validate suggested control point image measurements”.
Is it necessary to accepts and click on the green plus, even when automatically pointed cp’s are visibly out of range?
When it is necessary to click on the green plus, is there a faster way to accept all plusses in a control point? It seems a bit time consuming to accept all individuals by mouse-clicking…
Is it an idea to change the “feature source” of non-registered images to make them register? It is now set to use all image features, my guess is when an image is low in features turning it into merge using overlaps/ use component features can be useful? If you already have a large component will “use component features” be more of value?
regards,
Youri
Ho youri_v
You can see how many features each image has in the 1Ds view.
You cannot force RC to use those images with CPs alone - RC needs to find the matches by itself. So if you can’t convince it, it won’t do it. Using CPs is a good way to achieve that, but no guaranty for success. The (in most cases) quickest and easiest way is to take more images in the problematic area.
Your questions about the automatic CP placement are answered in the help. They are only suggestions and you need to adjust them if neccessary. You do not need to use them all and you can also use other images that are not being suggested. That one has to click every suggestion individually has been criticised resently here:
https://support.capturingreality.com/hc/en-us/community/posts/360009259991-Navigation-and-Suggested-Control-Points
You need to try and see for yourself if the suggested changing of the feature source will help your model along - it is different in every case. Keep in mind that already existing components will influence the newly calculated ones. So if you don’t make any headway anymore, you can try and delete all (make sure you can undo that though) end re-align. Sometimes this will result in a larger component, especially if you have added control points before.
Related…in having “fun” trying to get detail out of one particular area of my model. I’ve added control points in the hope of at least giving it some structure.
at first, the suggested CPs were a long way from their “real” position…but later ones graduall migrated to their correct places. Background features is on, so I guess that’s why?
ipdo I need to realign to take advantage of these, or can I just go straight to reconstruct again?
I’ve been wondering what ‘Background Features’ does - can you give a hint?
Not 100% sure but I believe it starts some limited processing in the background rather than waiting for you to click “align”…
I thought prob that i.e. feature detection at least as soon as you load the photos (which takes only seconds tho). But you said
“at first, the suggested CPs were a long way from their “real” position…but later ones graduall migrated to their correct places. Background features is on, so I guess that’s why?”
so do you think it does early, even premature work on processes further down the line?