|
Info:
|
When the AI Layout Automation is used for automatically placing articles on a layout, it might be that certain aspects of what should be placed and how it should be placed need to be given more or less weight.
This can be controlled through so-called 'Placement Presets'.
Multiple presets can be defined, each one prioritizing a certain aspect such as copyfitting, the placement of images, a balanced placement, and so on.
How Placement Presets can be set up and managed is described in this article.
How it works
Placement Presets are configured in a dedicated worksheet of the spreadsheet named 'Layout Automation config' that is used for configuring a workflow for each Brand.
Each preset is defined as a row of parameters, including a dedicated name.
For each parameter, a percentage of importance is given, ranging from 0 to 100. Parameters with a higher percentage than other parameters are given more weight compared to those other parameters.
The sum of all percentages should equal 100. This means that when increasing the value of one parameter, the values of other parameters have to be decreased.
The spreadsheet is uploaded to the system, after which users see the names of the presets appear in a list in the Layout Automation dialog. Users can choose the preset they want to apply. The chosen preset is remembered for future sessions.
Note: The use of this feature is optional; when no presets are defined, the list in the Layout Automation dialog does not appear.
What can be configured
The following table shows the parameters that can be configured for each preset.
Note: Apart from the preset name, all parameters are defined as a percentage.
| Parameter | Description | |
|---|---|---|
| name |
A name that describes how the Layout Automation places the content. This should be descriptive: users will see this name in a list in the Layout Automation dialog; will not see or have a notion of the values of the parameters that are set. Example: 'Copyfit', 'Image priority', 'Balanced'. |
|
| placed_area | The total page surface that will be covered by the placed objects. A higher value reduces unused spaces. | |
| num_placed_objects | The number of objects that will be placed. A higher value increases the number of objects that will be placed. | |
| priority | The priority that should be given to the objects that will be placed, based on what is set in the Priority property for an article. A higher value leads to more high priority objects that will be placed. | |
| knn_fit |
How well an object fits in an Article Shape, based on attributes listed for the invoked objects in the Layout Automation dialog: words, characters, story type, images, and quotes.
|
|
| blueprint_non_reuse | Avoidance of picking the very same blueprint more than once within the same print section. A higher value strives for fewer ‘duplicate’ blueprints. | |
| page_aim | How much importance should be given to the page number specified in the 'Page nr (Aim)' field of the objects. A higher value increases respect for the user’s preferred page to place the object onto. |
Figure: An example of a preset configured in the spreadsheet.
Objective
The overall objective is for the algorithm to maximize each parameter.
The following table shows the specific numbers of the factory defaults that reflect the standard behavior. Use these as a starting point to tweak the parameters at your own insight.
| Parameter | Weight (%) |
|---|---|
| placed_area | 15 |
| num_placed_objects | 0 |
| priority | 15 |
| knn_fit | 15 |
| blueprint_non_reuse | 5 |
| page_aim | 50 |
The total placed area, knn fit, and priority each receive 15% attention, meaning that each contributes 15% to the overall objective.
Minimizing blueprint reuse receives 5% attention, while maximizing page aim hits receives 50% attention.
All attention weights must sum to 100%, which represents the full objective. By adjusting these percentages, you control how strongly the model prioritizes each goal relative to the others.
Attention correlations
The area objective part is defined as the sum of areas of all placed objects.
Notice the following relationship: maximizing the total placed area and maximizing the total number of placements are correlated quantities. Because the area is a positive number by definition, placing more objects necessarily leads to an increased total placed area.
In addition to the area, the same is true for knn fit and priority objective parts.
|
Example: Giving 20% attention to the number of placed objects, 20% to area, 20% to priority, and 20% to knn fit creates a bias towards placing more objects that is stronger than the 20% explicitly specified for the number of placed objects. Instead, take this into account and make sure that the correlated metrics do not overwhelm the rest. A good mental model is to think of the attentions as follows:
where the placement attention is broken up into 4 parts:
In this model, it should be defined how much attention to give to metrics that are driven by the number of placements, and then distribute that attention value among the 4 constituent metrics as desired. |
Attention examples
The following are some examples showing the attentions in action:
| # | Scenario | Configuration |
|---|---|---|
| 1 | Many articles should be placed, but it is not important when some remain unplaced. The chosen shapes should fit the chosen articles well to minimize manual work afterwards. | Set placed area, number of objects, and priority attentions to 0, and give all attention to the knn fit. |
| 2 | Place as many of the defined articles as possible, without focusing on how they fit in a shape. The result should be an initial layout that is refined later. | Give all the attention to the number of placed objects, rather than knn fit, placed area, and priority weight. |
| 3 | Cover as much of the area as possible, while still taking into account the priority of objects. | Split the attention between placed area and priority. |
Configuration steps
Step 1. In the 'Layout Automation config' spreadsheet for a Brand, add a worksheet named Placement Presets.
Step 2. Add the following columns in the following order:
- name
- placed_area
- num_placed_objects
- priority
- knn_fit
- blueprint_non_reuse
- page_aim
Step 3. For each preset that you want users to make use of, create a row and fill out each column. See section 'What can be configured' above for more details.
|
Notes:
|
Step 4. When done, save the spreadsheet.
Step 5. (Optional, only needed when making changes after the Layout Automation has been set up, else this step is part of the main setup process). In Studio, right-click the spreadsheet and choose Upload to Layout Automation from the context menu.
Step 6. In Studio, open the Layout Automation dialog for a layout and test the presets that have been created by choosing each one from the list.
Note: Users who have the Layout Automation dialog open when the spreadsheet is uploaded need to close and open the dialog to see the changes.
Revisions
- 6 July 2026: Marked the article as 'outdated' and added information about this to the introduction.
- 29 June 2026: Updated links to Setting up and configuring the AI Layout Automation feature in Studio.
- 28 May 2026: Updated the notes for Step 3 in section 'Configuration steps'.
Comments
0 comments
Please sign in to leave a comment.