Japanese | English

Kobe Studio Seminar for Design

Workshop: Toward Problems for Production Pipeline in CG

In the previous seminar, we have agreed with an opinion that a sort of ideas and theoretical approaches in computer science are used in other fields (including CG and not limited to). For example, recent complicated use case and specification, a robustness guarantee for long-term proper operation, and for those requirements, meeting the global standards, giving academical proofs and developing verification methods from the academical viewpoint, and so on.

In CG, easiness of management/maintenance tasks in several stages of production, and artistic flexibility/freedom (i.e. freedom of non-systematic operations) are trade-off. This trade-off will be an important factor for designing production pipeline and system. Moreover, we have realized that this issue is heavily depending on understandings and awareness between designers (of pipeline, system and so on) and users.

In this seminar, we would like to focus on their theoretical backgrounds, and will have a discussion from the academic viewpoint of related subjects only. However, to make the discussion better, we will give some fundamental sessions in such related subjects (e.g. system verification and so on).

This seminar expects mathematicians, computer scientists, CG-practitioners (involved in research from the academic point of view), R&D personnel of companies and students studying these subjects. The participants may read the information on this page (which will be available) in advance. If you are not any specialist as noted above and you'd like to attend our seminar, please contact to the corresponding organizer: Kosaku Nagasaka in advance. We will give you more detailed information on our seminar.

Date:
2014/09/06
Place:
Place: Graduate School of Human Development and Environment, Kobe University. (No. 10 at the Campus Map)
Room:
A739 with A721 (eduroam is available)
Workshop Organizers:
Makoto Takahashi (Kobe University),Kosaku Nagasaka (Kobe University)
Note:

Please note that this workshop is co-organized with Kobe Studio Seminar for Mathematics and Kobe Studio Seminar for Studies with Renderman.

Please see also the references below.

Program

Time:
10:00-10:50
Title:
Production pipeline accompanied with education pipeline
Speaker:
Yusuke Kiriu (Studio Phones)
Abstract:

In our current film making, rule-based html-generator gave us several ways of education pipeline for self-leaning, brain-storming, in many stages of production pipeline. In this talk, we discuss an experimental approach for programmable education pipeline to reduce our internal communication costs from the view point of our current failure and success knowledge.

Time:
11:00-11:50
Title:
Introduction to Pixar's USD from a viewpoint of Symbolic and Algebraic Computations
Speaker:
Kosaku Nagasaka (Kobe University)
Abstract:

TBA

Time:
13:20-13:50
Discussion:
USD as a pipeline for research
Moderator:
Yusuke Kiriu (Studio Phones) , Kazuhisa Takemoto (Honda R&D Co.,Ltd.)
Abstract:

There are interesting discussions in the world, whether we have a pipeline that is practical in the production and also theoretically extensible for research. Unfortunately, USD has not yet been released, however, we would like to discuss this possibility and the talks in the morning.

Time:
14:00-14:50, 15:00-15:50
Title:
Program Verification and Hoare Logic
Speaker:
Makoto Takahashi (Kobe University)
Abstract:

The aim of program verification is to prove correctness of programs relative to expected requirements (specifications). Hoare logic is a deductive system for proving program correctness and consists of a set of axioms and inference rules about asserted programs. In this talk, we shall give a brief outline of Hoare Logic and talk about some verification tools.

Time:
16:10-17:00
Title:
VC-dimension in model theory and statistical learning theory
Speaker:
Kota Takeuchi (University of Tsukuba)
Abstract:

One aim of learning theory is to give a good estimate of the whole set from a given small fragment. VC-dimension, a measure of a complexity of a class of sets, was first introduced by Vapnik and Chervonenkis in their study of learning theory around 1970, and it was found that the finiteness of VC-dimension is equivalent to PAC-learnability. Several years later, Shelah found the same concept with a lemma known as Sauer-Shelah lemma now. It gives a characterization of an important class, called NIP, of structures in model theory. Recently I found a generalization of VC-dimension in the study of the class n-NIP introduced by Shelah in 2009. In this talk I'll give a brief introduction to the theory of VC-dimension and discuss the meaning of the generalization.

Time:
17:20-17:50
Discussion:
Is there any advanced challenge on Production Pipeline in CG?
Moderator:
Makoto Takahashi (Kobe University), Kosaku Nagasaka (Kobe University)
Abstract:

We would like to have a discussion on any existence of advanced challenge on Production Pipeline in CG from the practical and theoretical points of view. For example, we may be interested in any common problem in several use cases with production pipeline softwares and newly developed functionalities, any research topic to solve such a common problem, any approach to solve or resolve the costs brought by some production pipeline constructed inappropriately, any future direction to get permanent or universal production pipeline in CG, and so on.

References

For the participants, we would like to cite the following documents for your preparations in advance. These documents may be helpful for considering the pipeline standard, constructing technical services for artists, or thinking of pipeline development.

In addition to the above documents, the following sites may be useful.

The following references are useful for studying fundamentals for this workshop.

References for Program Verification and Hoare Logic. Please note that we omit refenreces in Japanese, which are noted only in Japanese page.