Collecting_data

This file contains the results obtained in the user study made in our research work on best-view-selection.

The results are currently not visible, as we are awaiting acceptance of our research paper.

models

Our 3D model database consists of 44 regular models. We have acquired 26 models from Visual Attention for Rendered 3D Shapes, and the other 18 are freely available on the web.

Pre-processing has been applied to these models. All models are :

• non-textured,
• positioned with the Y axis running from bottom to top: gravitational upwards (when it is meaningful),
• resized between -1 and 1,
• centered at the origin.

json_files

Context:

There are 203 json files corresponding to the results of each of the 203 users recruited using the platform Prolific.

Each user studies 10 different 3D models, and must choose 3 viewpoints which he feels satisfy the main question of the study. The 10 models are randomly selected from the dataset and displayed according to a random viewpoint.

Once this study is complete, each user must explain, for a subset of 5 studied 3D models, why he made them. Keywords are suggested to help them justify their choices : Side view, Front view, Global view, Eyes contact, Pleasant, Recognizable, 3/4 view. A free text field is also available for users to enter their reasons.

Camera positions:

The user can switch between 26 different camera positions distributed over a sphere centered on the mesh, as illustrated in the figure.

Camera positions C_{i,j} are defined by two angles (\alpha_j, \beta_i):

\forall i \in [0,7], \ \forall j \in [0,4]

\begin{equation*} C_{i,j} = \left\{ \begin{array}{ll} x_{ij} = R \cos(\alpha_j)\cos(\beta_i)\\ y_{ij} = R \sin(\alpha_j)\\ z_{ij} = R \cos(\alpha_j)\sin(\beta_i) \end{array} \right. \end{equation*}

with R=2.2, \beta_i = i.\pi/4 and \alpha_j = (2-j).\pi/4.

When |\alpha_j|=\pi/2 \Rightarrow \cos(\alpha_j) = 0, we obtain the same camera position eight times. Only one of them is kept because we always have exactly the same view.

Organizing json files:

"Participant"
    |__ "Age"
    |__ "Gender"

"Tasks"
    |__ "Task_1"
            |__ "obj_name": name of the 3D model that was viewed during task n°1
            |__ "random_initial_position": random intial position of viewed 3D model
            |__ "choices"
                |__ "choice_1": first selected point of view defined by its two angles (α,β)
                |__ "choice_2": second selected point of view defined by its two angles (α,β)
                |__ "choice_3": third selected point of view defined by its two angles (α,β)
    |__ "Task_2"
            |__ ...
    .
    .
    |__ "Task_10"
            |__ ...

"Analyses"
    |__ "Analyse_1"
            |__ "obj_name": name of the 3D model that was viewed during analyse n°1
            |__ "keywords": ordered list of selected keywords
    |__ "Analyse_2"
            |__ ...
    .
    .
    |__ "Analyse_5"
            |__ ...