Project management

CHAPTER 02

Hundreds of photos,
neatly sorted by face.

In this chapter you will learn to

• Create a project with one set per wall face
• Organise projects with search, copy and archive
• Filter images by judgment and processing status
• Cull duplicate photos using location data

9 min read

In CRITIR, one survey (building) is a project, and each batch of photos inside it (north face, south face, …) is a set. To create a project, you just point each set at its photo folder.

12345

Creating a project

Select + New Project on the project list screen.

Basic settings

• Project name (required) — Enter a name that identifies the project, such as the building name
• Description (optional) — Free-form notes or remarks

Configuring sets

A set is a unit for grouping images. Orthomosaics are generated per set, so put each facade of a building (north face, south face, and so on) in its own set.

• Assign one or more image folders to each set
• Click "+ Add Set" to add another set
• Folders can be added to or removed from each set individually

Adding image folders

Choose a folder with + Add Image Folder, or drag a folder from Explorer onto the Drop folders here area.

To align infrared and visible images, both must be located in the same folder.

Finishing creation

Click "Create". The images in the specified folders are scanned automatically, and the main screen opens.

Automatic fine alignment during import

When images are imported, fine alignment runs automatically for newly completed visible and thermal image pairs. When adding images to an existing project, it targets the newly imported pairs. Existing manual adjustments are preserved.

Check the alignment in overlay mode after import. For further adjustments, use fine alignment in the viewer.

Managing the project list

The list screen — shown right after startup, or opened with the Projects button at the far left of the toolbar — is where you organise your projects.

1234

For projects with several sets, hovering over a card cycles through each set's photo (the dots under the card show how many).

Per-project menu

Use the button on each card, or right-click the card.

Copying a project

You can choose between two copy scopes.

Filtering images

Use the controls at the top of the image list panel to narrow down which images are shown. This is handy when you are looking for specific images among hundreds.

Filter panel

Press the Filter button to open a panel with one row of choices per axis.

• Selecting several choices within one axis keeps images that match any of them (OR); combining axes keeps images that match all of them (AND). An axis with nothing selected is ignored
• The count next to each choice is the number of images after the image-type tab and filename search, before the axes are applied. It does not change when you toggle other axes
• While a filter is active the button reads Filter (N) (N = number of axes with a condition) and its tooltip lists the conditions. The panel header shows Showing N / M, and the image-type tab counts reflect the filtered result
• When no image matches, the list shows "No image matches … at the same time." together with a button such as Remove the "Judgment" condition (→ 12 images) for an axis that would bring results back
• Clear all removes every condition at once

Badges on images outside the image link

Images that are not part of the image-linking (SfM) result show a badge in the top-left corner of the thumbnail image. Images that are fine (included in every product) show nothing.

• Hover over a badge to see the registration state for the ortho and the 3D mesh (registered / not registered / not generated) in its tooltip
• The Image Linking: Not linked filter shows every image carrying any of the No ortho, No 3D or Unlinked badges. Linked shows the images without a badge
• Ortho: Adopted / Excluded is separate from image linking: it filters by the images actually used in ortho generation (Adopted) and the images you set to be excluded from ortho generation (Excluded)

Marking images as inspected

Click the badge in the top-left corner of a thumbnail to mark the image as inspected (a green check badge). Click it again to clear the mark. Besides tracking which images you have reviewed, it also works well as a flag for images of interest you want to revisit later.

You can also select multiple images, right-click, and choose Mark as Inspected / Clear Inspected to set them in bulk.

Image right-click menu

Right-click thumbnails to act on the selected images (one or more).

Culling by location

From Process ▾ → Cull by Location on the toolbar, you can automatically exclude duplicate images captured while hovering and the like. Running this before image linking, 3D reconstruction, or orthomosaic generation shortens processing time.

Move the Target spacing slider (0.5–30.0 m, default 3.0 m) and the result updates instantly, e.g. 361 total → keep 154 (207 excluded). Click Run Culling to apply.

How culling works

Following the capture path, an image becomes an exclusion candidate when its distance from the previously kept image is below the target spacing. Because this is not a simple "keep 1 out of N" rule, bursts from slow or hovering segments are consolidated without leaving gaps in fast-flight segments.

In addition, the following images are always kept, even where spacing is dense:

• Images where the camera heading (yaw) changed by 25° or more, or the tilt (pitch) changed by 20° or more
• The first and last images of the capture
• Images without GPS information

COLUMN

Why not just keep “1 out of N”?

A drone does not always fly at the same speed. While it stops to frame a shot (hovering) or creeps closer, the camera keeps taking photos. When it moves on to the next wall it flies faster, and the photos spread out.

Thin a flight like that mechanically — "every third photo in shot order" — and the dense parts keep plenty of near-duplicates while the already sparse parts develop gaps. Where a gap opens, neighbouring photos may no longer overlap enough.

CRITIR instead follows the capture positions along the path and decides by the distance from the last photo it kept, so the photos it keeps end up evenly spaced. On top of that, a photo where the camera turned sharply is kept even when it is close to its neighbours, because it is looking at a different part of the scene.