FAQs

Please ask us in the forum, or contact us directly.

The AIM@SHAPE Shape Repository is a key component of the e-Science framework of tools and services for modeling, processing, and interpreting digital shapes, developed within the AIM@SHAPE project.

The repository is primarily meant to facilitate the research process.

For developing new algorithms, it is important to have a number of small and easily manageable shapes that cover all necessary test scenarios. This enables an efficient prototyping as well as a first proof of concepts. On the other hand, for practical evaluation, real-world or large-scale benchmarks have to be considered.

The Shape Repository provides digital shapes of varying complexity for any of these purposes.

Once a critical mass of models is reached, the AIM@SHAPE Shape Repository will become the European reference database of digital shapes, similar to other repositories available on the web, e.g. the Stanford 3D Scanning Repository, the NIST repository. Consequently, its content will be presented and referenced in numerous publications, serving as an important means of public relations for the entire AIM@SHAPE network.

Most of them are. At the time of upload, model owners are asked to specify whether they would like to open the models to public, or keep them restricted to AIM@SHAPE partners. A large portion of the models in the repository have been made open to public. However, each model is governed by some terms of use, which are prescirbed in the accompanying licence(s). Please go through them when downloading a model from the repository.

All such legal stuff is clarified in the licence(s) accompanying each model. We request you to take the time to go through these short licence(s) when donwloading a model.

Yes, we request you to be mindful of people's feelings. Some models represent artifacts of religious, cultural and/or historical significance, e.g. the Max Planck model. Please handle these models with the care and respect you would otherwise observe for the original artifact. Please refrain from conducting "amusing" experiments on them, e.g. morphing, animation, Boolean operations etc. For these purposes, feel free to use any of the other models in the repository!

Registered users, i.e. users with a valid username and password, can upload models to the repository.

Currently, the repository accepts 3D mesh models in off, vrml, ply (bin and ascii), and obj. Some third party resources for these file formats are listed below.
OFF: documentation Unix/Linux viewer Windows viewer
VRML: Unix/Linux viewer Windows viewer
PLY: documentation tools Unix/Linux viewer Windows viewer
OBJ: documentation Windows viewer

Version 1 of the repository accepts meshes in the indicated formats only. In coming versions, the repository will accept a broader range of data - 2D and 3D images, range data, point clouds, etc.

Uploaded models are manually inspected and given a rating based on their visual complexity, aesthetic value and rarity. The rating is on a scale from 0 (lowest) to 5 (highest).

Group models are envisaged to act as containers for logically related models. Some examples of models logically related to each other are: remeshings of the same model with modified parameter values, one shape in several file formats, multiple range images of a shape possibly accompanied by one or more aligned and registered models. Instead of adding these models to the repository as several individual models, they can be added as a single group.

When adding a group to the repository, you have to enter a group name and description. Typically, this is the description of Level 0. See the next question for more on group levels.

Levels within a group are meant to reflect a hierarchy among the models. Sometimes the hierarchy is obvious, e.g. range images (lower levels) accompanied by a final model (highest level). In other cases, there is no implicit hierarcy, e.g. the same shape represented in files of different format. Here, models may be assigned to levels arbitrarily.

The topmost and most special level is Level 0. When you choose to add a group to the model, you are immediately asked for information on this level. Apart from the group name, the Level 0 description serves as a description for the entire group. The model added at this level serves a `representative' of the group. This is the only model from the group whose thumbnail is displayed on the View page. For this reason, only 1 model is permitted in this level.

After level 0, each subsequent level is lower in the group hierarchy. Each of these levels has its own Level description, and imposes no limit on the number of models. While empty levels are allowed, they do not make sense and are not recommended unless you plan to add models to a level soon.

There has to be at least one level in a group (Level 0), and no restriction on the maximum number of levels. For most groups in the repository, 3 levels (up to Level 2) are sufficeint.

Please note that, at the moment, only registered users can add models to the repository.

Adding a group of models to the repository is much like adding the models in the usual manner, i.e. individually. The few additional steps are to input group-specific information, namely group name and description and level descriptions. Once you finish adding, a new group consisting of the added models in the specified hierarchy is added to the repository.

After choosing 'group' for upload type, you will have to enter the group name and Level 0 description. This serves as the group description. Level 0 accepts only 1 model which is added in the usual way. The description of the Level 0 model should be given in the subsequent model upload page.

At the bottom of the upload page for the Level 0 model, you can proceed by adding more models to the group which will automatically go to the 'Next group level' (Level 1), or you can 'Finish' uploading, in which case the newly added group will consist of the single Level 0 model. More on group levels can be found in the answer to the previous question.

Before adding models to a new level, you will have to input a Level description. Models are added in the usual way. At each upload page, you will have the choice to add more models in the 'Same group level', to complete the current level and continue adding models to the 'Next group level', or to 'Finish' uploading and add the group with models uploaded up to that point.

Most users encounter problems when adding a group model for the first time. Luckily, if you are logged in with the same username you used to upload the models, there is a lot you can do to remedy the situation, including:
- edit uploaded metadata
- edit group and level descriptions
- add more models to an existing group level
- create a new group level (created at the bottom of the hierarchy)
- convert an individual model into a Level 0 model of a group

However, at the moment, you cannot remove added models from the repository. If that is what you want, or if any combination of the above does not solve your problem, the only remaining option is to contact us with a brief account of the problem.

Using the MT library, we can internally convert some shape models to MT's native mtf format. An mtf file can then be queried with a quality term for a mesh at the desired quality. Higher qualities are closer to the original mesh, while lower qualities are drastically simplified. The output mesh can be returned in several file formats.

Supported models in the repository are automatically processed and made available for extraction in the desired quality and format.

Conversion to mtf format is done by the TriMesh2MT tool. According to its README file dated November 2005, TriMesh2MT accepts triangular surface meshes in OFF, IV and VRML (1.0 and 2.0) formats.

Meshes can be currently extracted from mtf files in OFF, PLY and VRML formats.

Try the homepage of the MT library!

We use the TriMeshInfo tool for metadata extraction. As TriMeshInfo accepts a restricted set of file formats, we first run a converter script for some additional formats.

When a supported model is added to the repository, metadata is automatically extracted from it and displayed along with the user's input metadata. The user can then choose to commit either the extracted metadata or the input metadata to the repository.