Self-Organizing of Kanazawa Art Museum
1.1 Introduction:
The Kanazawa Art Museum’s open layout provides visitors with the freedom to explore and discover exhibits at their own pace. However, irregular exhibition area boundaries and random tour paths can present challenges for visitors. Our project aims to explore the qualities of such layouts and how we can enhance visitors’ experiences. Specifically, we will focus on the layout of exhibit areas and their qualities when exhibition area boundaries are irregular and the area of the rooms does not change. In this paper, we outline our approach to achieving this goal, which includes obtaining entrance locations and exhibition layouts, planning optimal navigation routes, generating possible routes, analyzing visual effects, identifying areas with poor visual effects, adjusting exhibit layouts to improve them, and proposing dynamic and variable space layouts that allow for periodic reorganization. Through this research, we hope to improve the museum experience for visitors and provide valuable insights for museum owners and designers.
2.1 Methodology — spatial reorganization and routes finding:
- Model:
In our model, we combined the spatial reorganization of exhibition rooms and route finding in a single continuous workflow. We first use the method of genetic algorithms to divide spaces that match the original room setting, but in different configurations that will enable different visiting experiences and fit into various types of exhibitions. Then, to maintain the main concept of the museum, which is maximizing visitor’s freedom, all corridors surrounding each exhibition space are designed to be accessible.
This will lead to infinite choices of routes, and the model will suggest four efficient routes for each configuration, while still keeping most possibilities. The model contains the following steps:
LAYOUT FINDING + OPTIMIZATION
Step01: 1st divider is controlled manually for generating different layouts
Step02: Subdividing spaces to reach 19 blocks within the boundary
Step03: Optimization to control the tolerance reaches 2%
DOOR FINDING + SORTING
Step 04: Dividing room spaces into 2 categories based on room size.
Step 05: Remove all door points on the boundary.
ROUTE FINDING
Step 06: Picking one of the entrance points and using Python script to find the closest door of each gallery.
Step 07: Find all possible paths based on room layout.
Step 08: Divide all paths by generating used door points.
Step 09: Find the Shortest Walk between all used door points.
Step 10: Find 4 shortest routes and calculate the total distance of them in each layout
- Input parameter:
Initial inputs include the boundary enclosure, the area of each exhibition room and courtyard space in the original layout, which is used in the genetic algorithm step as the main constraint of generating a new layout, and the first spatial division cut, which is a controller designed for the museum owner or exhibition curator to obtain more possible variations. Each cut is specified as an “h-horizontal cut” or “v-vertical cut” for cut direction, plus a number between 0–1 represents the cut position along the axis. All other inputs such as possible paths, possible entrances, and possible door openings are generated automatically from this process.
- Performance metrics and outputs:
The genetic algorithm will generate the new layout, in which each room size and each category of the program (including gallery, courtyard, utility, 19 rooms in total) is as close as possible to those input room sizes and types. Corridor areas are allocated to each room to create a layout without gaps, which makes it easier to carry out the model. Room shapes are also limited to certain ratios to avoid generating strange unusable spaces. To control the calculation time, we manually set the tolerance of room size’s difference to 2% for the comparing and matching process.
Rooms are categorized according to areas into two groups to have different door openings. After having all door point locations sorted and interrupted segments of paths, these become input for route finding: The strategy is to always pick the nearest exhibition door point considering a visitor’s location, which means finding a path among infinite possibilities to minimize the walking distance to the next gallery. This is a dynamic process that helps to guide visitors. We pick one entrance point as the current location, loop through all door points to find the nearest one, and use that door point to update the current location. And then loop this program again and again, until all galleries have been visited. We also consider all door points for one gallery space as a group, every time a gallery room has arrived, all doors belonging to the room will be removed from the remaining list, so the route will only enter each gallery once.
The method is repeated for all other possible entrances and generates one route for each. In order to quantify the accessibility for each layout, we select 4 shortest routes and calculate the total distance of them for this layout.
The output at this stage includes room layout, entrances, and recommended routes to be used in the view analysis, and they can be also converted into a map product.
2.2 Methodology — view analysis:
- Model:
In the previous model, the study of spatial reorganization and recommended routes generated multiple plan layouts for us. In this step, our model focuses on further visual optimization of these plans. Since all exhibition spaces are surrounded by opaque materials and most of the corridors are narrow, we take the courtyards inside the museum as the object of visual analysis. For the visual analysis section, there has been a large amount of research over the last 40 years about extracting quantitative measures from architectural space and in turn evaluating the space with these measures. In our visual analysis model, we mainly use “spatial openness” and “visual complexity” as quantifiable metrics, which can be extracted by the Isovisit component in Grasshopper. This model contains the following steps:
Choose analysis objects and define obstacle
Step01: Trans all the courtyards into mesh
Step02: Generate even distributed grid inside each mesh area
Step03: Select vertices inside mesh area as analysis objects
Step04: Designate the boundaries of all space other than courtyards as obstacles
Run Isovisit in Grasshopper and quantify output
Step05: Input analysis objects and obstacles and run Isovisit
Step06: The first output of this component shows the first intersection of the multi-rays from the selected objects with the obstacles, connect these intersections to form closed figure
Step07: Display the area and perimeter of these closed figure as output, take these measures as quantified spatial openness and visual complexity
Visualize the output in Grasshopper
Step08: Remap the area and perimeter values calculated earlier to a range of 0.0 to 1.0 and use them to create a range of colors for visualization
Step09: Use the Colour RGB (f) component in Grasshopper to compose colors based on normalized 0–1 values for each color channel.
Step10: Use the Construct Mesh component to build the visualization mesh, use the ‘area’ measure to drive the ‘Red’ channel of the color and the ‘perimeter’ measure to drive the ‘Green’ channel of the color
- Input parameter:
The initial input parameter includes the courtyard area to be analyzed as well as all other spatial areas. Take other spatial areas as obstacles, the rays emitted from the vertices will create numerous intersections with obstacles. These intersections will be used as intermediate metrics for the final analysis.
Other input parameters are mainly used to regulate the intensity of the analysis. For example, the reconstruction of the mesh can increase or decrease the number of objects to be analyzed, while the second and third inputs of the Isovisit component regulate the number of rays emitted from the analyzed objects and the distance of rays. To some extent, these input values influence the objectivity and accuracy of the output results
- Performance metrics and outputs:
As mentioned above, there are two kinds of outputs in this model — One quantified one and one visualized one.
1. Based on the work of M. L. Benedikt (1979), who provided a set of measures which could be extracted from Isovist to quantify spatial experience. Two of the measures Benedikt focused on in particular were the area and perimeter of the isovist, which are also the quantified output of our visual model. They not only depict the line that connects the intersection of numerous rays from a particular point in space with the obstacles surrounding it, but also reflect the spatial openness and visual complexity. In our model, we try to keep the output values within three digits and test the model with several plans generated in the former model. For the area output, which indicates spatial openness, the number ranges from 97 to 118. For the perimeter output, which indicates visual complexity, the number ranges from 62 to 69. These values effectively help us to understand the characteristics of the space and help users to select the right plan layout.
2. Besides the quantified value, the visualized output provides more visually readable characteristics of the space. We provide three different diagrams as visual output. The first diagram illustrates spatial openness with a gradient color from red to black, in which lighter regions of the floor plan are more open while darker regions are more enclosed. The second one use a gradient color from green to black to indicate the visual complexity. In this diagram, Lighter regions have higher visual complexity while darker regions have lower visual complexity. The last map is formed by overlaying the first two diagrams. With the RGB color coding, it displays the space in multi different colors like orange-ish, yellow and mustard, showing the spatial openness and visual complexity at the same time.
3.1 Results :
After several attempts to optimize the design by genetic algorithm, we generate 6 new planes and measure the suitability of the planes in different scenarios by view analysis to provide different choices for owners in different situations.
4.1 Conclusion :
Our product meets the requirements of both museum proprietors and visitors by optimizing the exhibition experience while preserving visitors’ freedom of exploration. Additionally, we provide a streamlined solution for gallery reorganization, allowing owners to administer their galleries efficiently and effectively. With our product, museum proprietors and visitors can have a pleasant and seamless experience.
