How to add more photos to an existing project?

[maddin]

Let's say I have a set of 80 photos that were nicely processed with PhotoScan (alignment, adjusting bounding box, computing dense point cloud), just to discover that I forgot to cover a certain part of the object to be reconstructed.

So I start the camera with the same settings as before and shot 20 more photos to cover that part - now what is the best way to add these photos to the project without having to redo everything?

Thanks!

[stihl]

1. In your project, right click in your workspace pane to create a new chunk

2. In this new chunk, import the photos by going to workflow in the toolbar and choosing add photos.

3. Align the photos in this new chunk with the same parameters you used in the initial chunk.

4. Go to workflow > align chunks  and let it align by the point based method with high accuracy and generic preselection turned on if applicable (can't remember exactly what the chunk alignment parameters all were).
((If you happened to use markers for your initial chunk, add the markers for this new chunk as well and use merge chunks instead of align chunks))

6. If you've used align chunks, you'll end up with a new combined chunk which should have the tie points of both chunks. Just run a dense cloud of this chunk.

[James]

You will have to create a new chunk and align and merge it with the existing chunk.

Add your new photos to a new chunk, and at least 3 (but preferably more) images from the existing chunk, which overlap with your new photos.

Align the photos in the new chunk, and then align chunks using 'camera alignment'. This will just look at the positions of the photos common to each chunk and is very fast.

If that worked ok, then you can build dense cloud in your new chunk, and then merge the chunks to get a 3rd chunk containing the dense cloud from both chunks.

If you have Pscan Pro you can align chunks by markers which is a bit more flexible.

The other option is to align chunks by 'points' but that can be much more time consuming, although possibly more robust as it looks for common features in the photos of each chunk to calculate the alignment.

[maddin]

Great, thanks! Will try that!