Internships
and
Jobs offered in 2012 in the REVES Research group
|
|
For any questions please contact Adrien.Bousseau
at inria.fr or George.Drettakis at inria.fr. If
you are interested in one of the positions
please email us directly.
Three
positions currently available:
Please
note that these positions are in the context of
an EU-funded project led by a games company, and
including several top-notch computer graphics
and vision labs in Europe. The starting date is
November 1st.
|
| Masters or Engineering
Internship |
|
No positions are currently open; please come
back in November 2013.
|
| Postdoctoral Fellowship
|
Computer Graphics/Computational Photography
Postdoctoral Fellow
Image Based Rendering, Relighting and
Perception for Games
(all bibliographic references below)
We are looking for a postdoctoral fellow (18
months renewable up to 36 months) to work on
Image-Based Rendering (IBR), Image-Based
Relighting and Perception for IBR, in the context
of Computer Games. This position is in the
context of an EU-funded project led by a games
company, and including several top-notch computer
graphics and vision labs in Europe. The starting
date is October 2013 or later.
The successful candidate will work with a team
consisting of a Ph.D. student, a software engineer
and two senior researchers at INRIA, as part of a
broader European consortium.
The project aims to provide working solutions for
image-based rendering (IBR) and relighting,
continuing the successful line of work in
image-based navigation in [Chaurasia 11, Chaurasia
13], intrinsic image with the goal of achieving
relighting [Laffont 12, Laffont 13] and the study
of perception for IBR [Vangorp 11, Vangorp 13]
(see below).
The successful candidate will develop novel
research solutions to outstanding problems in
image-based rendering, image-based relighting and
perception for image-based methods. Typical
research questions include: how to provide
real-time inpainting for regions which are not
present in input photos/videos during interactive
navigation; providing real-time relighting during
image-based navigation by developing robust
inverse global illumination and material models
based on sparse and possibly inaccurate 3D data;
developing predictive models for the perception of
artifacts in IBR (e.g., ghosting or popping), by
studying the human vision literature and
developing appropriate experimental protocols.
These are just examples, and the successful
candidate will have significant freedom in
proposing the exact research topic in the same
overall research area.
The ideal candidate must have a Ph.D. in Computer
Graphics or Computer Vision (with applications to
graphics). A top-level record of research
publications in these fields is required.
The candidate must be fluent in English.
The position will be at the REVES research group
in Sophia-Antipolis,
http://www-sop.inria.fr/reves, situated in the
beautiful French Riviera.
The position will
start Oct/Nov 2013.
|
|
Ph.D. Student |
Computer
Graphics/Computational Photography Ph.D Position
Available
Multi-view
Methods for Image-Completion and Image-Based
Rendering
(all
bibliographic references below)
Please note that
this position is in the context of an
EU-funded project led by a games company, and
including several top-notch computer graphics
and vision labs in Europe.
Image-based rendering (IBR) algorithms
have the potential to provide interactive
rendering of unprecedented realism. Our
recent IBR algorithms leverage sparse but
incomplete 3D information provided by multi-view
stereo algorithms to guide shape-preserving
warping algorithms
[Chaurasia2011,Chaurasia2013], greatly improving
the result.
However, several significant problems
remain. First, as soon as the user moves
away from the input views it is necessary to
synthesize content to complete missing regions.
Recent inpainting methods eg [Darabi12] are too
slow for interactive navigation. Our procedural
methods [Galerne12] could also be useful.
Second, various hard cases such as as
reflection, thing objects (railings) etc. cannot
be treated by current methods. Third, rendering
synthetic object coherently poses a number of
issues of coherence (lighting etc.) to allow
convincing combination of image-based and
synthetic assets.
The first research axis in this thesis will be
the development of a fastimage completion
approach which preserves structure, while
respecting the performance constraints of
interactive IBR. We will build on on
accelerating image-completion approaches using
the multi-view data. We will first apply this
solution to noisy depth input from Kinect-like
or RGBD cameras and then generalize to
multi-view and sparse depth scenarios. We
will then generalize the image completion
approach to combine the strengths of
patch/neighborhood-based solutions ([Darabi12])
and procedural methods ([Galerne12]) to develop
a novel hybrid solution.
The second axis in this thesis will involve the
use of object-level learning techniques (e.g.,
[Bleyer11]) to allow more powerful completion
strategies using similar texure information from
large image collections, by combining sparse 3D,
warps and image completion. This will also
potentially help with the hard cases such as
reflections and thing objects.
Finally, we will also investigate ways to
combine image-based and synthetic assets
convincingly. This will involve determining what
makes such a combination to be perceived as
realistic, and developing a powerful inverse
illumination model to perform such a
combination.
The successful candidate should have a Masters
in Computer Graphics (or Computer Vision but
with some experience in graphics). A solid
understanding of basic graphics and vision
principles is required, together with experience
in OpenGL/GLSL (or DirectX equivalent)
programming as well as knowledge of matlab. The
candidate must be fluent in English. Good
scientific writing abilities in English are
required.
The position will be at the REVES research group
in Sophia-Antipolis,
http://www-sop.inria.fr/reves, situated in the
beautiful French Riviera. Please note that
Ph.D. programs in France are 3 years. The
position will start Oct/Nov 2013.
|
Software Engineer
|
Computer
Graphics/Computational Photography Software
Engineer Position Available
Image Based Rendering and Relighting for
Games
(all
bibliographic references below)
We are searching for an
engineer for a 18 month position (renewable up to
36 months) to work on Image-Based Rendering and
Relighting for Computer Games. This position
is in the context of an EU-funded project led by a
games company, and including several top-notch
computer graphics and vision labs in Europe. The
starting date is October 2013 or later.
The successful candidate will work with a team
consisting of a Ph.D. student a postdoctoral
fellow and two senior researchers at INRIA, as
part of a broader European consortium.
The project aims to provide working solutions for
image-based rendering and relighting, by
continuing the successful line of work in
image-based navigation in [Chaurasia 11, Chaurasia
13] and [Laffont 12, Laffont 13] (see below).
The main task of the successful candidate will be
to develop software components within the common
platform (e.g., the Unity3D engine) which will be
used throughout the project. This will typically
involve porting and cleanly re-writing initial
research code developed in the research projects,
thus providing more efficient and robust version
for use by the game company testers in the
project. The engineer will also participate
-- when appropriate -- in the development of novel
algorithm, and/or provide software support for the
more preliminary research results developed.
A smaller percentage of time (less than 1/5 of the
total working time) will involve manageiral tasks
for the EU project: handling EU project reporting
(writing deliverables, participating in
coordination meetings etc.).
This EU project is a very ambitious endeavour
which we believe has significant potential to
change the way game developers create extremely
realistic lightweight backdrops (urban scenes,
including detailed buildings, vegetation
etc.). There is thus a significant
possibility of technology transfer or the creation
of a startup at the end of the 3 year project.
Candidates interested in such a development will
thus be considered favorably.
The ideal candidate should have a Masters in
Computer Graphics or Computer Vision, good
knowledge of C++ and related development tools in
both Windows and Linux, preferably with good
knowledge of matlab. Knowledge of image
processing, computer vision and computer graphics
algorithms is a must. The candidate must be fluent
in English. Candidates with Ph.D.'s who do not
wish to pursue an academic career, but would like
to remain in a research environment, are also
encouraged to apply.
The position will be at the REVES research group
in Sophia-Antipolis,
http://www-sop.inria.fr/reves, situated in the
beautiful French Riviera.
The position will
start Oct/Nov 2013.
|
References
|
[Laffont 13] Pierre-Yves Laffont, Adrien Bousseau,
George Drettakis, Rich Intrinsic Image
Decomposition of Outdoor Scenes from Multiple
Views, IEEE Transactions on Visualization and
Computer Graphics - 2013:
http://www-sop.inria.fr/reves/Basilic/2013/LBD13/
[Laffont 12] Pierre-Yves Laffont, Adrien Bousseau,
Sylvain Paris, Frédo Durand, George Drettakis,
Coherent Intrinsic Images from Photo Collections,
ACM Transactions on Graphics (SIGGRAPH Asia
Conference Proceedings), Volume 31 - 2012:
http://www-sop.inria.fr/reves/Basilic/2012/LBPDD12/
[Chaurasia 13] Gaurav Chaurasia, Sylvain Duchêne,
Olga Sorkine-Hornung, George Drettakis, Depth
Synthesis and Local Warps for Plausible
Image-based Navigation, ACM Transactions on
Graphics, Volume 32, 2013
http://www-sop.inria.fr/reves/Basilic/2013/CDSD13/
[Chaurasia 11] Gaurav Chaurasia, Olga
Sorkine-Hornung, George Drettakis,
Silhouette-Aware Warping for Image-Based
Rendering, Computer Graphics Forum (Proceedings of
the Eurographics Symposium on Rendering), Volume
30, Number 4, 2011
http://www-sop.inria.fr/reves/Basilic/2011/CSD11/
[Vangorp 11] Peter Vangorp, Gaurav Chaurasia,
Pierre-Yves Laffont, Roland Fleming, George
Drettakis, Perception of Visual Artifacts in
Image-Based Rendering of Façades, Computer
Graphics Forum (Proceedings of the Eurographics
Symposium on Rendering), Volume 30, Number 4,
pages 1241--1250, 07 2011
[Vangorp 13] Peter Vangorp, Christian Richardt,
Emily A. Cooper, Gaurav Chaurasia, Martin S.
Banks, George Drettakis, Perception of Perspective
Distortions in Image-Based Rendering, ACM
Transactions on Graphics (SIGGRAPH Conference
Proceedings), Volume 32, Number 4, 2013
|
|
