APP/APG v2.0 Control Points Editor
Ground Breaking Concept
- 1 Ground Breaking Concept
- 2 Interface Description
- 3 Local Help Window
- 4 Selection Windows
- 5 Viewer Toolbar
- 6 The Editor Grid
- 7 Optimization
- 8 Tools
Yet another Autopano Pro & Giga innovative concept; the Control Points Editor will allow you to manipulate your panoramas with great ease, in a very efficient way. The general idea behind the Control Points Editor is to be able to manually edit the Links and the Control Points to achieve an optimal alignment of the source images. In some cases the automatic stitching mode will have trouble aligning some images, and will even exclude some of them from the panorama. This problem can easily be corrected using the Control Points Editor.
But before going into more details, we invite you to take the tutorial.
Selecting the tool will modify the Editor Window display. The Editing Grid formed by a network of points connected by lines and labels will be now displayed on top of the panorama. The second window will allow you to edit the control points by selecting either the links (i.e. bounds between two or more images), or the individual source images currently included in the panorama. The control points are used by Autopano Pro to place the images within the panorama.
If you do not have a dual-monitor setup, I suggest you arrange your workspace as follow.
Local Help Window
This area will be displayed after entering the control points editing mode. It gives access to the most important functions of the editing mode and displays a summary of the panorama’s status related to the stitching quality.
The Image List is closely tied to the Link List and the Control Points List. By selecting an image you will effectively be able to visualize the links and the control points this particular image shares with the other images inside the panorama.
This feature will allow you to quickly see if an image is correctly bounded to another.
A duplicated image can easily be deleted using the keyboard Delete key.
Tip: Selecting all the images will allow you to display the global Grid of the panorama.
The Link List is closely tied to the Image List. Meaning that when selecting one or more images the list will display the links shared by the selected images.
It is a good practice to sort the list by RMS (i.e. link quality), and to delete the bad links using the keyboard Delete key.
Note: A smaller RMS value equals a higher quality link.
Control Points List
The Control Points List is closely tied to the Image List. This means that when selecting one or more images the list will display the control points shared by the selected images.
It is a good practice to sort the list by RMS (i.e. control points quality) and to delete the bad ones using the keyboard Delete key.
Note: A smaller RMS value equals a higher quality control point.
- Views Activation / Deactivation: /
- Portrait / Landscape Layout: /
- Synchronize the views:
- Launch Panorama Optimization:
- Add Control Points to the selected areas:
- Delete Control Points from the selected areas:
- Automatically add links/control points:
- Automatically delete bad Control Points:
- Automatically add and delete Links/Control Points:
- Find orphan images locations:
This icon initiates the panorama’s optimization calculations. This step will take in account all the changes performed on the panorama and adjust it to the new constraints.
This process can be configured in the Options Dialog.
This icon will allow you to automatically add control points. All you need to do is to define a common area on each image and hit the keyboard Enter key.
- A smaller area will lead to a higher accuracy in the control points placement. In some cases however, it is possible that adding control points will fail (i.e. no new control points are generated). In such case try a smaller area.
- Adding control points implies the creation of a link between the images.
This icon will allow you do delete control points. To do so, define an area on one of the displayed images. All controls points inside the created area will automatically be selected. Next, click on the icon or use the keyboard Delete key.
Note: If you delete all of the control points shared by two images you are also deleting the link between those two images.
The use of this button will perform a geometry analysis of the panorama in its current state. The geometrical analysis is used to find the existing overlaps from the source images current locations. For each overlapping area found, the software simulates the adding of control points, just like you would have done yourself by defining the two zones. You can configure the way the software will be adding the links in the Optimization Settings Dialog.
The screen capture on the right illustrates this concept. The overlaps analysis forces Autopano Pro to automatically add links in the center of the panorama.
The main appeal of this function is that its use generally raises the panorama's quality. The number of links between the images is increasing; the individual links are then becoming stronger. But it’s not always the case. In the case of a 360x180 panorama, this analysis is less efficient at the poles.
This icon will allow you to delete control points on all images following two methods:
- Error threshold method: all points over a maximum error threshold will be deleted.
- “N” best control points method: Only the "N" best control points per link are kept.
Each method has its own upside and downside. We personally have a tendency to give a plus to the second method as it will never break a link (i.e. even if all control points are really bad, some will always remain). This is not the case with the Error Threshold method as it can very well delete all control points from a link. The choice of method is performed in Setting dialog. It can be accessed using the icon.
Selecting this tool will perform in one click the sum of actions performed by the previous two icons.
Selecting this icon will enable you to find the location of those images that would otherwise have been left out during the initial stitching pass. This can arise if you chose, for example, to force using all images in a group. This method can be useful when finding yourself with images left out because of their high degree of resemblance (examples: highly homogenous blue sky, rows of identical windows on a building, etc.).
The Editor Grid
The goal of the Grid is to visualize the links existing between the images and to give information on their quality. This Grid is displayed on top of the panorama preview and allow for a rapid visualization of the existing links, their quality, missing links, etc. It also allow for a new selection of the images and the links.
- Selecting an Image:
represents an image and its index in the Image List.
When selecting an image, the Editor Window will highlight the selected image along with some red lines showing you witch images are linked to the current selection.
In the example given on the right, the selected image (#13) is linked to the images 3, 4, 5, 6, 12, 14, 16 and 17. And if we look closer in that area we can see that there is no other image present; Autopano Pro found all links!
- Selecting Multiple Images:
In order to edit the control points shared by two images, all you need to do is select a pair of images. Click on the first image to select it, you can then select the second image by holding the keyboard Ctrl key down and by clicking on the second image. The Preview Windows will refresh and display the selected images. You can now add or delete control points.
- Selecting a Link:
represents a link with a RMS value of 1.79 and a number of control points of 47. When selecting a link, the Editor Window will highlight the link itself, along with the two images located at each extremity of that link. A red line connecting the two images is also displayed. A color code is used to quickly identify good and bad links:
- Green (RMS <5): Good quality link.
- Orange (RMS between 5 and 10): Medium quality link.
- Red (RMS > 10): Poor quality link.
If you find yourself with red links (i.e. poor quality links) it means you have bad control points. The control points must be edited. It is possible that the images do not relate to each other. In such a case, delete the link.
- Selecting Multiple Links:
It is possible to select multiple links at once. The Grid display will then change to reflect the selection.
This set of parameters determines how the panorama will be optimized. It is during the optimization step that each image position is computed based on the control points it holds. By default only the optimization profiles are displayed in the dialog. The Advanced Options are only displayed if you click on the "Show Advanced Options" button.
Four (4) optimization profiles are available:
- Quick: this will perform a very fast optimization that can possibly fail if some bad image links exist in the panorama.
- Strong: (default) suited for most stitching projects. This profiles leaves room for intervention in case some bad links exist but, as a trade-off, is slower.
- Gigapixel: select this profile for all projects over 200 images.
- Fisheye: select this profile for any project using fisheye images.
These profiles are presets for the several optimization steps.
- Use grid position: this option will only be available after an import from a panoramic shooting head that will provide the images coordinates (typically Papywizard or Clauss). This option will then be active and will let the optimization engine know if it needs (or not) to take the images coordinates in account during the optimization process.
- Assume row/column shooting: was the shooting performed in a row/column pattern or not? This option is activated for panoramas issued from a Gigapan shooting head. We suggest using this option only in that particular case because this mode needs enhanced accuracy during the shooting process (n row, p column). It is best not to use this option if you find yourself with an extra source image in one row or in one column, or if the spacing between two consecutive rows varies too much.
- Optimizer stages
- 1) Local approach: this is a very fast algorithm that allow for a local pre-location of the images.
- 2) First optimization: this is the main optimizer, taking all links into account in a global way.
- 3) Clean up control points or links:
- Keep only control points below this error (RMS): useful when only quality matters.
- Keep only the N best control points: useful when only the number of control points per image matters.
- Remove links over this error (RMS): useful when only links quality matters.
- 4) Final optimization: the same main optimizer used in stage 2 is called upon (if it hasn't been yet) or recalled upon (to take into account de deletions performed in stage 3).
- 5) Adjust lens distortion: this option allows for lens distortion correction (very useful on short and medium focal length). It is better to uncheck this option if you are stitching a gigapixel panorama as it will slow down the detection process.
- Advanced distortion: allow for better distortion correction on fisheye lenses
This tab regroups the Control Points Editor Settings.
CP detector quality
The default value is inherited from the value used for the initial detection of the panorama.
- Detection Quality:
- Low is a good compromise between control points quality and detection speed.
- Standard (default) increases control points density but slows down detection speed. In the case of a set of images presenting a small overlapping ratio this option can enhance the detection quality on the overall panorama. It is also used with panoramas presenting a lot a bracketing.
- High very slow, mostly reserved to desperate cases (… worth a try before giving up).
- Control Points / Image Pair : This is the number of control points used per image pair. Influences the number of control points exported to PanoTools as well as the overall detection quality.
This tool performs a geometrical analysis and tries to find similarities between the source images overlapping inside the panorama. This analysis can be performed in three different ways:
- We only want to find new control points for overlapping images that do not share any control points.
- We want to find new control points for all overlapping image pairs and add them the already existing control points.
- We want to find new control points for all overlapping image pairs and replace the existing control point by the new ones.
Bad control point cleaner
- Keep only control points below this error (RMS): Useful when only quality matters.
- Keep only the N best control points: Useful when only the number of control points per image matters.
- Remove links over this error (RMS): Useful when only links quality matters.