DRAFT MINUTES OF PAVR INTERIM PROGRESS MEETING

Ecole des Mines, Nantes, October 2nd, 1997

SUMMARY OF ACTIONS ARISING

PJW

send a letter to Siemens and pay them a visit as soon as possible

BA

arrange VRML tutorial at the forthcoming workshop in April.

JWP

chase own contacts on J-DF's behalf for speech recognition candidates

J-DF

help Josep Blat to obtain SoftImage developer package from Microsoft

WP Coordinators (PJW, DT, J-DF, Siemens, J-DG, FvR, JWP)

organize first meeting before the next management meeting in Palma

Anna

circulate minutes by email, as well as making them available on PAVR web page

update mailing list

check existence of hypertext links to the various sites from PAVR home page and check and update partners' URLs

chase missing contributions towards annual report from Bath, Glasgow, Limburg, Siemens and Nantes

All

refer any suitable applicants to partners with vacancies

provide early feedback to gauge likely attendance to forthcoming meetings

 

Present

University of Bath

Max Froumentin

University of Glasgow

John Patterson (chair)

Ecole des Mines, Nantes

Gérard Hegron, Jean-Daniel Fekete

EPFL, Lausanne

Tom Molet

University of Geneva

Pascal Volino

TU, Vienna

Francois Faure

UJF, Grenoble

Jean-Dominique Gascuel

IRISA, Rennes

Bruno Arnaldi, Stéphane Donikian (2/10 only), Jean-Luc Nougaret, Frédéric Beauchamp

UIB, Mallorca

Josep Blat

LUC, Diepenbeek

Peter Verhoye

Anna Beria was in attendance

 

1. Welcome, introductions, apologies and departure times.

John Patterson welcomed those present, and newcomers in particular. He presented Phil Willis' apologies and was asked to convey to him the meeting's best wishes for a prompt and full recovery. Apologies had also been received from Nadia Magnenat Thalmann, Frank Van Reeth and Michael Gervautz. No particular constraints on departure times existed for that day since most participants were staying on for the technical session on the following morning. Departure times for the next day were noted.

2. Minutes of previous meeting

Adoption of the minutes was postponed to enable those participants who had not seen the minutes before the meeting to read them. Minor corrections to the list of participants were made, viz:

UIB, Mallorca: Eric Chauvineau (not Chauvinau)

LUC, Diepenbeek: Frank Van Reeth (not van Reeth)

UJF, Grenoble: Jean-Dominique Gascuel, François Faure

TU, Vienna: Michael Gervautz

Action: AB to circulate minutes by email, as well as making them available on PAVR web page

The agenda of the meeting was adopted with the addition of Item 10. Date and place of next meeting.

3. Matters arising from minutes

The action list circulated with the minutes was checked.

Phil Willis had carried out all his tasks. This included repeated attempts at contacting Siemens to agree on their role and arrange a visit during the summer. Although an answer had been received, the date proposed was not convenient, and it had been hoped that Siemens would be present at the meeting as it was essential to clarify their position before the annual report was drawn up. Since this was not the case, and no apology had been received from Siemens, PJW was asked to send a letter and pay them a visit as soon as possible.

Action: PJW

Josep Blat had submitted UIB's proposed contribution to the project and also a report on work carried out so far.

John Patterson had not yet chased his own contacts on J-DF's behalf, but would do so in the near future.

Action: JWP

Jean-Daniel Fekete reported that no multi-site licence was available for SoftImage. A standard developer package could be obtained for 1000 CAN DLS for one year. Josep Blat had expressed an interest, but had received no answer from Microsoft.

Action: J-DF to help

On Nadia Thalmann's behalf, Pascal Volino said that no clear answer had been received on the subject of multi-site licencing for Alias Wavefront and 3D Studio, and that a more accurate description of the project was needed. Josep Blat reported that Wavefront offered three partnership levels, including a partnership programme for developers, below trade price, whereas 3-D Studio only offered standard commercial discounts. He proposed visiting Wavefront's website and reporting to the meeting during the technical session on Friday morning.

WP Technical Meetings: no technical meetings had been convened by the WP coordinators, as efforts at all sites had concentrated on finding good candidates for the posts offered and evaluating available software. WP8 expected to hold their first meeting in the new year, before the next management meeting.

Action: All WP coordinators (PJW, DT, J-DF, Siemens, J-DG, FvR, JWP) to organize first meeting before the next management meeting in Palma

4. Review of appointments

J-D. Fekete (EMN) said that interviews had been held, but no appointment had yet been made, although a candidate from Lintz seemed promising. He asked whether travel expenses for candidates to attend interviews could be charged to the project. After consulting Phil Willis by 'phone, it was confirmed that this could not be done, although it was possible to reduce costs by interviewing candidates at any of the partners' sites when visiting on PAVR business.

J-D. Gascuel (Grenoble) said that although no appointment had yet been made, a German post-doc candidate, Bernd Eberhart seemed likely to be appointed.

J. Patterson (Glasgow) said that Jean-Christoph Nebel had now been appointed and would start in November.

P. Volino (Geneva) informed the meeting that Laurent Moccozet had been replaced by Marc Eicher.

Action: AB to update mailing list

F. Faure (Vienna) said that he had been appointed as a post-doc and Chris Faistenauer (I) had filled the PhD post.

M. Froumentin (Bath) stated that he had recently been appointed as a post-doc and Frédéric Labrosse, another post-doc, would start in November, thus completing Bath's contingent.

Bruno Arnaldi (Rennes) had interviewed a PhD applicant from Greece. The decision would be made in two weeks' time.

J. Blat (Mallorca) announced that Eric Chauvineau had left the project at the end of September and might join the Diepenbeek team. No other candidate was in sight and everyone was urged to advise of any good candidate.

P. Verhoye (Diepenbeek), who represented his site at the meeting but was not working on PAVR, said that no suitable candidate had yet been found.

Action: ALL to introduce suitable applicants

Action: AB to check existence of hypertext links to the various sites from PAVR home page and check and update partners' URLs

 

SUMMARY OF RESEARCH APPOINTMENTS

(Vacant posts in brackets)

LOCATION

POST 1 (Post-doctoral)

POST 2 (PhD)

University of Bath, UK

Max Froumentin [FR]

Frèdèric Labrosse [FR]

-

University of Glasgow, UK

Jean-Christoph Nebel [FR]

(PhD)

Siemens, Munich, DE

?

?

EMN , Nantes, FR

-

(PhD)

EPFL, Lausanne, CH

-

Tom Molet [FR]

Amaury Aubel [FR]

University of Geneva, CH

-

Pascal Volino [FR]

Marc Eicher [FR]

TU Vienna, AT

François Faure [FR]

Chris Faistenauer [IT]

UJF, Grenoble, FR

(Post-doc)

-

INRIA/IRISA, Rennes, FR

(Post-doc)

(PhD)

UIB, Mallorca, ES

(Post-doc)

(PhD)

LUC, Limburg, BE

(Post-doc)

-

 

5. Review of progress

The project annual report to the Commission was being compiled. Contributions had been received from most sites, but Bath, Glasgow, Limburg, Siemens and Nantes were still missing.

Action: AB to chase

The most imminent deliverables deadlines were D5, D16 and D17. D5 had already been achieved as a standalone demonstrator had been produced. D17 depended on Siemens' stance. As for the proposed workshop (D16) J. Blatt said that if Siemens were not willing to host or organize it, then UIB would be happy to do so, to coincide with the next management meeting.

6. Technical meetings

See under point 3.

7. Mobility of researchers

J. Blat said that the project should strive to produce jointly supervised PhDs and joint papers as these were considered important in TMR projects and would look good in the final report to the Commission. However, pending the choice of a common harness and the appointment of more young researchers, it was difficult to arrange visits and exchanges. Discussion would nevertheless continue after the formal meeting and any planned visits would be announced the next day during the technical session.

 [The following was subsequently agreed:

J-D Gascuel to visit TU Vienna in the new year

J Blat to visit LUC in the near future and Glasgow at the end of November, possibly to be joined there by P. Willis]

8. Mailing list updating

The PAVR mailing list was circulated for updating. It was felt that at this stage it was unnecessary to create separate mailing lists for administrative vs. technical communications as the volume was not so great. It was suggested that adding an indication of the nature of the message in the header (eg. [ADMIN] or [TECH] would help the recipients to sort out which were relevant to them.

J. Blat asked all colleagues to avoid replying to individual messages using the PAVR list, as this wasted a lot of people's time.

9. Any other business

The minutes of the Geneva meeting were adopted and it was agreed that a decision on the common harness would be arrived at during the technical session on the next day.

10. Date and place of next meeting

J. Blatt confirmed UIB Mallorca as the venue for the next management meeting, to be held in conjuction with technical sessions and, possibly, the D16 workshop from 15 to 17 April 1998. He encouraged participants to include the Saturday in their travel plans both to qualify for cheaper air fares and to enjoy a day off in pleasant surroundings. J. Patterson suggested that the following mid-term review meeting take place in Glasgow at the end of August, dates to be confirmed. Early feedback is required to gauge likely attendance to both meetings.

 The formal meeting closed at 5 p.m., but most of the participants gathered the next morning to continue the discussion on the packages which had been tested and evaluated, so as to reach a decision on a suitable common harness for the project.

 

TECHNICAL SESSION SUMMARY

 

 The following is not a verbatim report of that discussion, but rather an attempt to convey the participants' positions and the way in which a degree of consensus finally emerged. Feedback from participants is urged to clarify positions, fill in gaps, correct misunderstandings!

 Jean-Daniel Fekete did not mind which platform would be used, since at the end it was necessary to produce something that would be directly usable by others, without further integration work. SoftImage was possibly not the right solution, but VRML had no API and no set of functions; it could describe movement, but not create it and no testing would be possible.

Jean-Dominique Gascuel was not in favour of starting work with VRML now. It was not useful for PAVR purposes for two reasons: (1) VRML is a "file format", not a framework in which plugins with our knowhow can just be added; with such an approach, most of the time would be spent re-inventing the wheel, and (2) the project needed a "sexy" demonstrator, hence the need to use a commercial package with special effects (SoftImage, Wavefront). He favoured SoftImage.

François Faure reported that Micheal Gervautz would prefer Open Worlds, but the non-Beta version was not yet available. Personally, he preferred VRML 2.0: it was small, clean and open.

John Patterson's main concern was asset generation. VRML was best suited as exchange format. Was it possible to generate VRML objects? Possibly so. He tended to prefer Soft Image.

Josep Blat said that VRML was ok as exchange format. He wished to try out SoftImage and develop it further, but had more experience with Alias and good contacts with Silicon Graphics. On the whole he reckoned that not sufficient experience had been gained by the various sites to arrive at a satisfactory decision.

Peter Verhoye said LUC had no experience with any commercial package, but he felt that VRML would provide a good basis.

Frédéric Beauchamp felt that a VRML-based platform could be more versatile and make it easier to integrate other parts. Investigating tools required a large amount of work. His site had started work with Alias Wavefront in 1995 and it would be a problem to move away from it now. He felt that VRML 2.0 was probably a good choice for PAVR.

Max Froumentin felt that an exchange format was more important than the actual package employed, given Bath's tasks. He himself only had experience of VRML1.0 and wondered how open the commercial packages were. He would not venture to decide or even give an opinion pending further investigation.

Pascal Volino said that Miralab considered Alias to be a solid framework, except for its lack of real-time capability. Personally, he felt that a start could be made using VRML, to which a package could be added later.

Tom Molet said that none of the packages supported real-time interactions and wondered whether VRML 2.0 would. It certainly could provide key-frame support and enable the exchange of animation between partners.

Gérard Hegron suggested working with two platforms. Since VRML was no good for interaction, a PAVR platform could be developed for one demonstrator and a SoftImage-based platform for another. He suggested using the forthcoming workshop to examine actual work done by the labs. End-of-year-2 deliverables were very independent and a final choice could be postponed until April next year.

John Patterson stressed the need for an exchange format and noted that a strong consensus was emerging for using VRML as an exchange format.

Jean-Daniel Fekete did not believe that VRML could be used for exchanging animation and wondered whether sufficient time would be left at the end for integration.

Bruno Arnaldi suggested holding a VRML tutorial at the forthcoming workshop in April. He wished to test a real plug-in with a VRML platform. Jean-Dominique Gascuel would supply one.

Action: BA to organize tutorial.

Jean-Daniel Fekete could see the need for an interface during development to reduce integration time at the end.

Bruno Arnaldi felt it necessary to convene a technical meeting for WP3 and WP5 to decide on a common basis for collision detection. This would be held in Rennes in January. WP4 might also wish to participate, subject to Siemens' reaction to the initiative. It was not necessary to create a new mailing list, even though collision detection was only relevant to WP3, 4 and 5. Prefixing relevant email messages with [COLLISION] would ensure that only interested parties would read them.

John Patterson summed up the technical discussions held over the two days. Consensus was clearly emerging towards using VRML as a common exchange format, although experimentation should continue with other packages. He saw recruitment as the main problem affecting the network. He felt that the two days had been a fruitful, mind-concentrating experience, even though no overall winner could be declared.