| |
Siggraph 2006
Bernard Lebel talks about the classes, presentations, parties and other things that took place at the SIGGRAPH 2006 Conference (Boston, MA).
September, 6th, 2006, by Bernard Lebel. Revised by John Syracopoulos
On July 31’st 2006,I had the opportunity to experience my first international SIGGRAPH conference ACM SIGGRAPH 2006 in Boston, Massachusetts (www.siggraph.org).
There were several goals I intended to accomplish at the 2006 SIGGRAPH conference. First, I wanted to attend as many educational classes and seminars that were available so that I may be able to learn and borrow new ideas, and techniques, to apply to my own practice. Secondly, I wanted to attend the Softimage user events to intercept old friends and acquaintances and build my personal and professional network. And third, I wanted to drink beer and discover what Boston’s nightlife had to offer. Before I move on, I am pleased to say that I did accomplish all my SIGGRAPH goals, and would recommend anyone interested in the academic or professional field of computer graphics to attend a conference at least once within their academic or professional career.
Saturday, July 29th BOSTON, MASSACHUSETTS
After a long flight I end up, as scheduled, at the hotel. The hotel was rather distant from the Convention Center, but despite the forty-five minute commute, it was worth it. At the hotel, I decided to wait in the lobby to intercept my dear friend Joe Saltzman (the distributor for my XSI Production Solutions and Tips training videos), but waited for three hours for Joe (who was running late due to his hotel shuttle service), to finally arrive. When he finally did arrive, we quickly registered with the hotel, unpacked our belongings, and headed towards another hotel where an acquaintance of ours, Kim Aldis, was staying. Ironically, after many years of email tag and interesting internet discussions, I finally had the chance to meet several 3D buddies of mine such as Kim Aldis, Ed Harris, Joey Ponthieux (pronounced: "Ponchey"), Christy Sprangler, and many others whom until SIGGRAPH, I never had the chance to thoroughly pick their brains over a pint. So in an effort to begin our SIGGRAPH properly, we headed to the nearest pub for a few drinks and a bit of relaxation from our travels.
Montreal's Pierre Elliott Trudeau International Airport, where it all started.
Downtown Boston, viewed from near the Boston Convention Center.
Left: Joey Ponthieux (pronounced "ponchey") getting a shot at me. Right: Ed Harriss.
Left: Adam Ferrall (green shirt at the left) looking at his cell phone or camera, Matt Lind (purple shirt) looking at the map, Byron Nash (dark blue shirt looking at Joey Ponthieux (flowers), Joey Ponthieux looking at Matt Lind, and François Painchaud (green shirt at the right) looking at the backside of his pulpills. Right: Kim Aldis (left) in a weird mood. Both Darren* (center) and Joe Saltzman (right) focus hard to figure what's going on.
Sunday, July 30th SIGGRAPH 06 – DAY ONE
Before moving on, I have critical advice for anyone who attends a future SIGGRAPH conference. BY ALL MEANS POSSIBLE, RESERVE THE HOTEL THROUGH THE SIGGRAPH WEBPAGE! Trust me, you may regret it if you do not. If you do this, you will save yourself a bit of frustration and unnecessary transportation expenses (via the SIGGRAPH shuttle bus). If you do not reserve the hotel through the SIGGRAPH website, you can still take the SIGGRAPH shuttle bus but at a heroic cost of seventy-five dollars U.S. per person. Unfortunately we did not do this, so we took the train, which cost us some money, but more time than anything.
Moving right along…
My first registered class was High-Dynamic-Range Imaging: Theory and Applications, conducted by Greg Ward from AnyHere Software, Karol Myszkowski from Max-Planck-Institut für Informatik, Helge Seetzen from RightSide Technologies Inc., Habib Zargarpour from Electronic Arts, Gary McTaggart from Valve Corporation and Drew Hess from Industrial Light & Magic.
Below is a summary of the contents of the course.
The first important thing to understand about HDRI (High Dynamic Range Images) is that pixels are stored as floating points (like XSI’s normalized color widgets,) instead of integers (un-sub-dividable whole numbers like in Photoshop). The main goal of HDRI is to have every pixel color, and intensity representative of the true intensity of the light source from the captured image (which is not the case in LDR or Low Dynamic Range). However there are big challenges involved in doing this.
The image must be captured in a larger or “dynamic” range of colors and intensities. To do so, we take multiple bracketed “exposures” of the same subject. This is the only way to remove unnecessary noise from the areas of the image that have low lighting conditions (see some of my low light pictures below for evidence)
The .RAW format is recommended to store the captured images due to its pure “loss-less” architecture.
Changes to the aperture of the camera should be avoided between exposures, and a lower ISO setting tends to capture the best detail.
Consideration of filters – where neutral filters that remove light and forces longer exposure times can be implemented to achieve a higher quality image.
And great care must be taken with mirror balls, as they are not 100% reflective, and a fisheye lens tends to create vignetting (which may result in a visible loss of color “ramp” data near contrasting areas and or banding) and possible cropping problems.
Secondly comes the data itself and how to manipulate it. This can be very complex, as the given display's response to light and its source must be taken into account. For instance, a digital monitors response to light is quite different from film's response to light. Therefore, any manipulation of the HDR data must be done with the output in mind. It is recommended to work in linear space (ie: perform all editing and rendering on the HDRI first) and then re-scale it to a logarithmic (or any other non-linear) space at the very end of the pipeline.
Third, there is the implementation of the HDR data to reproduce the lighting “source” in the 3D application (ie: “lighting with HDRI”). There are several problems involved, and multiple techniques that can be approached to achieve this. One of them is to sample the environment color (similar to how the Dirtmap Occlusion shader does). However, if the environment color is not properly sampled it will usually result in unwanted artifacts within the rendered image, and as we all know, if it is over sampled, it will punish your render time (especially with sample calculations near the areas of the images light source). Solutions to eliminate this problem include: creating 3D lights in place of the image light source and “cloning-out” out the visible light source from the HDR image (which is what ILM does). You can also replace the environment sampling approach altogether by creating an actual 3D light “rig” source (like the “Light Rig From Image” option in XSI if your 3D application supports it) and create and place more 3D lights if need be. Another technique includes "importance sampling" (take more samples in brighter areas of the map, less in the darker areas), use ambient occlusion with diffuse environment mapping, and/or use approximated geometry to calculate the shadows.
Nearing the end of the HDR implementation pipeline, there is the display or “how it will look” phase. Since most view displays are only LDR-capable, “tone-mapping” must be performed on the HDR image to make it visually acceptable. “Tone-mapping” is the process of achieving a greater diversity of detail and color in the medium ranges of the image, while maintaining the same contrast and light intensity of the original image. “Tone-mapping” is a science in itself and may require patience and time. There are many techniques and or approaches to achieving this, however none, by current industry standards, are considered to be the “best”.
There are operators that work on the entire image (global operators, which perform the same operation at every pixel), and there are operators that work in local areas of the image (called “I know you would have never guessed…” local operators). Basically, a local operator divides the image into regions, performs a Guassian Blur, and then interpolates between the divided regions. Local operators are preferred, however great care and attention to detail must be taken to detect the edges and to help minimize the overlaps of the regions with conflicting contrasts (which can cause halo’s).
Regarding the actual display medium, HDR displays were shown at the conference. This was very impressive! HDR displays allow the true intensity of each individual pixel of the image without having to perform any “tone-mapping”. What this means? Well, soon you will have HDRHDTV’s and HDRHDDVD’s! Let me catch my breath… The plan for the future is to have the image stream coming off the DVD or possibly a service provider as an HDR stream. The DVD or signal will detect if the output device (your monitor or TV or the translating receiving box) is HDR-capable. If not, it will perform an on-the-fly tone-mapping procedure to support the low dynamic display.
The Boston Convention Center. An impressive sight.
Monday, July 31th SIGGRAPH 06 – DAY TWO
My second class, was An Introduction to Sketch-Based Interfaces conducted by Joseph LaViola from Brown University, Randall Davis from Massachusetts Institute of Technology and Takeo Igarashi, from The University of Tokyo.
I was not quite certain as to what to expect from the class, other than fulfilling my curiosity in regards to what interactive interfaces actually were. I will have to admit though that the class was amazing, and I received more from it than I had anticipated.
A “sketch-based interface” is basically intelligent software and hardware which enable you, the end-user, to perform the vast majority of graphic based operations through hand-drawn strokes with a drawing instrument. For example, mechanical engineering software that uses a sketch-based interface that’s simply a white 2D canvas instead of a cluttered interface full of tools and operators. The user draws engineering symbols on that canvas, and the program correctly interprets the overall pictures as an engineering system.
So far, sketch-based interface research has been done for these domains of application: music notebooks, interface design, engineering simulation, mechanical systems, computer animation, mathematics, graphics, rapid prototyping, medical visualization, and teaching. The need for sketch-based interface is driven by the fact that current computer graphic tools have cumbersome interfaces, which limits the accessibility and learning curve of the software for all potential users. Sketch-based interfaces are at the forefront of simplifying the technical curve so that artists can concentrate more on creating instead of learning the technical components of the software and then creative ones.
The challenge for sketch-based interfaces is extremely high and the “under-the-hood” mechanics of it are quite intricate. To exemplify, first, the stroke made by the artist must be analyzed by the software system. How does the software know that the artist is intending to draw a straight line (which, when drawn by hand, is never perfectly straight) and not a curved or parabolic line? The software attempts to determine what the user is trying to do by analyzing the velocity and angles of the stroke. Graphics can be built and coupled to perform this analysis. Other problems need to be solved. At a very small scale, any stroke is a mess. Hooks (little naturally occurring decorative artifacts we create when we write or draw) must be identified and removed. How does it interpret our subtle inconsistencies of our personal writing and drawing styles? To solve this, segmentation is performed, where strokes are decomposed into smaller strokes, and classified by groups of similarity. In some cases, strokes may be "beautified" by the software (intended straight lines are actually redrawn as true straight lines), and the software may need to add information for the missing ones.
Past the strokes, the entire sketch must be evaluated. It's good that the software knows how to interpret linear and cubic shapes, but it must be able to interpret what the end-user is trying to express overall. Again, advanced intelligent analysis is necessary: the image is parsed into pieces in order for the software to be able to recognize the overall construction of the object and how it is assembled together. Also, the software must understand the specific domain of the representative objects application in order to minimize occurrences of misinterpretation. There is also the notion of noise, which can be caused from irregularities within the strokes, over sampling, or the sensitivity of the hardware. It's a big challenge to identify which components of the graphical information is valuable and what is not.
There were several prototypes shown at the class. One of them enabled the participants to create 3D volumes from 2D strokes. Then you could use tools to bend, scale, cut, and extrude these volume objects, as well as plot them from their volume form to a tessellated mesh.
Also demonstrated was a fabulous cloth simulator. On the left of the interface there was the three-dimensional model, and on the right resided, the drawing canvas. The application allowed the sketching of two-dimensional cloth pieces. Once the pieces were constructed, point markers were created which would allow the user to draw perfect dynamic cloth objects on the three-dimensional mesh by evaluating the seams from the markers, and their response to the environmental forces within the simulation were visually perfect. This class was amazing and it is truly inspiring to see scientists developing tools that will ultimately make the artist’s job much easier. I strongly encourage everyone to visit Takeo Igarashi's homepage at:
http://www-ui.is.s.u-tokyo.ac.jp/~takeo/index.html.
The next class that day was Taxonomy of Digital Creatures: Defining Character Development Techniques Based Upon Scope of Use, by Tim McLaughlin from Industrial Light & Magic. I somehow expected something more technical. It was more of a discussion about the various steps involved in creating a digital creature, and how to determine things like level of detail and features. Many subjects were covered, like distance from camera, screen time, reference materials, action, and performance. It was still a very educational and interesting seminar to see how an expert from ILM approaches the creation of a digital character asset.
After that, I met with many eminent folks from the Softimage community for something special, unfortunately however I cannot write about it.
Later, many of us went to the Seaport Hotel beers and more beers! Upon doing so, I met many more long time XSI users, like Michael Donovan, Bill Blakesley, as well as a bunch of Softimage people (my good friend Will, Halfdan, Chinny, Gareth, Jennifer, Alexandre Jean-Claude, Dominique, Jeffrey, and others). Also present were friends like Matt Lind, Graham D. Clark (hey buddy!), Kim Aldis, Bradley Gabe, Olivier Ozoux and Maggie Kathwaroon. It was a blast!
Tuesday, August 1st - DAY THREE
I have to say upfront that at this point, I was already exhausted. The plane trip as well as 2 days of intense technical “head-stuffing” and 3 nights of binge drinking was really beginning to take a toll on me. There was no way I would do anything without the king-size Dunkin Donuts coffee cup at hand. And yet I knew today was going to be the biggest day of them all.
With unmatched vitality and unparalleled receptiveness I attended my fourth class, RenderMan for Everyone, given by Hal Bertram from Walt Disney Feature Animation, Dan Maas from Maas Digital, Moriz Moeller from Rising Sun Pictures, Saty Raghavachary from DreamWorks Animation, Tal Lancaster from Walt Disney Feature Animation, and Heather Pritchett also from Walt Disney Feature Animation.
Although I'm not much into RenderMan at the moment, and had only recently begun writing shaders, I was still interested in the class, and was glad that I participated because of the inspiration I had achieved from it.
The class started with an overview of the RenderMan specification history and main features. For readers not versed in RenderMan topics, "RenderMan" is not a piece of software, but a technological specification for a programming interface. By following the specification rules when writing a render engine, scene files, and shaders, the source code becomes compatible with existing RenderMan-compliant renderer’s. However, not all implementations of the RenderMan interface strictly follow the specification. The goal of the specification is to allow different artists who use multiple software packages (such as: XSI, Maya, Houdini, 3D Studio Max, etc…) to render scenes using a RenderMan renderer.
Dan Maas gave many tips and tricks to solve common problems with RenderMan, like how to solve topology, transformations, memory optimization, bump mapping, texture seams, depth-of-field, as well as many other material based operations.
Moritz Moeller than spoke about a concept called "Magic Lights". Before I go on, Moritz Moeller, a guru at Rising Sun Pictures, is the individual who wrote the “XSI-3Delight” pipeline. 3Delight is a RenderMan-compliant renderer. Yes, there are people at RSP who use XSI and the RenderMan renderer for production.
Magic Lights, simply put, are lights used not to illuminate, but function as parameters and surface shaders. While Magic Lights can be used to illuminate an object or scene, its primary use is to be called by the surface shaders and pass information to that surface shader that would be difficult or inconvenient to transmit to otherwise. There are several advantages in this. Surface shaders are not necessarily called by other surface shaders, whereas lights can be called by “all” surface shaders. So using lights to pass information to a surface shader is a means of storing and passing data that is global to the scene. Also, it can be impractical to have too many parameters on surface shaders. Magic lights can be used to pass extra parameters to surface shaders without the impracticalities.
The power of Magic Lights includes the ability to texture multiple objects without any UV data on the objects, control procedural shaders without UV’s, explicit shading effects, and/or matte paintings. For matte painting, imagine that you have multiple lights in a scene, and each light projects the matte painting into the scene. Each light has adjustment parameters, and you can flip between the lights to view all versions of the matte painting.
Walt Disney folks talked about their RenderMan pipeline and some of the work they did for Chicken Little. They have developed their own preprocessor language, which is then parsed by a Perl program, which then outputs the final source code. Portions of the class included very “hardcore” demonstrations, some of it very technical and not intended for the unseasoned programmer. In the end, I left the class with a plethora of knowledge, a brain filled to its capacity, and an empty king-sized Dunkin Donuts coffee cup.
So I started lurking around the exhibition, waiting for the Softimage VIP cocktail party to take place. I have to say I haven't seen very cool stuff at the Emerging Technologies area. Some things were interesting, but nothing jaw-dropping. The Teapot Exhibition was rather boring too. One thing worthy of mentioning is that there was this big mosaic of a cityscape, made entirely of mini-teapots. Fortunately I do not have the time to do this kind of work.
Some obscure experiments from the Emerging Technologies area.
Emerging Technologies area.
Left: A piece of art. Right: another Emerging Technologies mystery.
The Teapot Exhibition area.
A mosaic of tiny teapots form this cityscape.
Left: The Teapot Exhibition area. Right: Believe it or not, all the people you see in this picture are actually in line for the Pixar booth, where "walking teapots" were given out.
Afterwards, I ended up at the Softimage booth for the presentation by Bradley Gabe, from Stan Winston Studios who is known as whitanar on XSI Base, about Eight Below. Brad explained how he rigged the seal that attacks the dog. His character rig demonstration was very versatile and allowed many different ranges of movement and performance for the character. For example, he presented the modularity of the rig to accommodate the cross performance of the seal from a crawling action to a swimming one. Brad also demonstrated his competitive spirit by mentioning that two of the dog names were Max and Maya. Ironically the production was made with XSI.
Bradley Gabe's presentation on Eight Below.
Full scale creatures and animatronics were created at Stan Winston for Eight Below.
After Bradley’s demo, I ended up at the Softimage VIP party, where everyone’s favorite “free” beer (and we even had Maudite, an ass-kicking Quebec Ale) and food complimentary of the hotel. At the party I ran into Sue, who is a fellow XSI Base member, (who did not tell me her XSI Base name! Unless I have forgotten....) the legendary Jeremy Birn, and to my great surprise, Moritz Moeller, all of whom I had lengthy conversations with and was given a great opportunity to really pick Moritz’s brain about RenderMan and XSI. Well, it was more about him trying to convince me to turn my studios pipeline to 3Delight!
Left: Éric Ballandras, obviously enjoying La Maudite beer. Right: Bradly Gabe and Ronald Beirouti. For many people, Ronald was the star of SIGGRAPH 2006.
Left: Greg Punchatz shaking hands with Jason Dexter. Right: Jason Dexter (standing up) and Steven Caron, the youngest XSI TD on Earth.
Left: Guy Rabiller (what's wrong buddy?) Right: François Painchaud, who seem to really dislike being photographed. Don't sue me please.
Left: Jennifer Goldfinch. Right: Pimpin' Will Mendez.
Left: Jeremy Birn. Right: Joe "Bring Me Beer" Saltzman.
Left: Moritz Moeller, obviously enjoying too much Quebec's beer. Right: Matt Lind.
Left: Sue (she didn't tell me her family name!). Right: Steven Caron (left) and Ed Harriss (right).
After the cocktail, it was time for the Softimage user group to start, at the hotel conference room.
Left: Michael Isner and Jordi Riera. Right: An overview of the user group room.
First, Chinny came on stage and spoke about the past 20 years of Softimage. On the video screen there were pictures of a few known XSI users being displayed. One of the pictures featured a recent portrait of that user, while the next one was a picture of the same user 20 years ago. It was nice to see Steven Caron as a baby!
Chinny starting the user with few memorable feature films from 20 years ago.
Left: Steven Caron now. Right: Steven Caron 20 years ago.
Following Chinny's introduction, Mark Schoennagel did a great demonstration using the legendary SOFTIMAGE|3D platform running on an dusty old SGI Octane work horse. Can you believe it? Mark created for us a hot sexy spaceship, and then he showed how cool it was back then to be able to create IK rigs. And then he showed how it could be used to animate a Godzilla-like character. Then he imported the character in XSI, the heavens parted, and we could truly see the sheer power and glory of XSI’s performance and versatility in contrast to the good ol’ SI3D days.
Left: Mark Schoennagel creating a high-res polygon spaceship on good ol' SGI Octane running SOFTIMAGE|3D. Right: The tyrannosaurus is brought into XSI so we appcriate the sheer difference in power between SOFTIMAGE|3D and SOFTIMAGE|XSI.
Then Mark Stevens took over the stage for a speech. I felt it was a bit long, but necessary. Mark talked about Softimage’s history and where its future is headed, in general terms (like "character animation"). After Mark gave his speech Mattias Worch of Factor 5 Games enlightened us with a brief overview of how Factor 5 achieved the high resolution (both geometric and texture based) in their latest game.
After that, Steve Perlman from Mova gave a presentation on Contour, a unique facial motion capture system. Basically Contour consists of painting spots on the subject's face with phosphorescent ink, which appears transparent under normal lighting conditions. The camera then takes several “tens” of pictures a second. One half are with the subject under normal light conditions, while the other half is under lights which illuminate the phosphorescent track points. The captured ink information can then be used to drive the motion of a 3D face. Those who are interested in the Contour system might want to read this article:
http://www.vfxtalk.com/forum/showthread.php?t=6319
Steve Perlman doing an overview of Contour, a facial motion capture system that relies on a special phosphorescent paint to track points. I apologize for the poor quality of this picture.
Afterwards, Michael Isner came on stage to discuss some various ideas regarding the much-anticipated Face Robot animation pipeline, as Face Robot 1.5 was announced. Michael also showed some tests the Softimage Special Project team did with the data provided by the Mova guys. FaceRobot 1.5 includes tools to work with the Mova technology.
Following Michael Isner's presentation, Chinny previewed some of the technologies expected to find their way into future versions of XSI. To summarize:
- Performance enhancements and improvements with high resolution meshes and data sets demonstrated via the translation of a character arm control enveloped to a mesh with sixteen million triangles in near real-time.
- Viewport playback performance has been enhanced and was demonstrated by Chinny playing back six characters whose mesh density consisted of one hundred thousand triangles each in near real-time (both in wireframe and shaded viewport modes).
- There are now shader balls. You can view a shader ball in every node of the Render Tree, and generate a larger rendered shader ball to assist in material creation.
- Multiple render regions can now be created.
- Also, there has been the addition of animation layering which allows the user to add a layer of keyframe animation on top of an existing animation source in the mixer.
- Animation re-purposing. There are now new tools to drive the animation of one object to another object, which is completely operator-based and requires no animation property transfer, or animation clip transfer either.
Unfortunately there was no word in regards to any updates in the particle components of XSI, or in mental mill/MetaSL either, as well as other highly hoped for features. Maybe next time.
The points previously mentioned are only technology previews. It means that there is not absolute certainty it's going to make it in the actual released software. With that in mind, they did not revealed much of what Softimage may or may not have in store for its XSI users in the immediate future. Despite the fun from the engaging presentations and some interesting tid-bit’s I found the user group to be a bit frugal for the insatiated user that I am. Something worth mentioning however is that it was verified at the user group meeting that Softimage is in the process of acquiring CAT technologies, which could re-affirm Softimage’s future aim to dominate the character animation realm with its XSI platform.
The performance has been improved dramatically. There are 6 characters of more than 100,000 polygons each, and both interaction with the rigs and playback are fluid.
Shader balls. Let's pray they actually make it to the released software!
Left: Animation layering. Right: Animation re-purposing. The left character's animation is inherited from the right one through operators, and extra animation can be put on top of it.
With the user group being over, it was then time to move on to the Boston World Trade Center, where the final Softimage party was taking place. The weather was humid, but other than that conditions were favorable for drinking beer. At this party, I met several XSI users from all around the globe, including artists and technicians from Omation studios (and Graham, who was armed with a workman belt where pockets held papers, and business cards were being stored for distribution). Somehow I managed to acquire drink tickets throughout the evening, which rendered the event fairly inexpensive for me. Thank you Softimage for the great party, I had a memorable time!
The entrace of the Boston World Trade Center, where the Softimage party was held after the user group. An impressive sight.
Wednesday, August 2nd - DAY FOUR
For the first time in weeks I could actually wake up late! I showed up at the convention center at about 12:30. My first objective was to meet up with Todd Akita, from PSYOP, who was giving a presentation about the MTV HD opening. First off, I am a bit jealous of PSYOP, I think they do some of the best work in the industry. And the opening of the MTV HD just lives up to the standards they set. It really is a fantastic piece of work and if you would like to see it, you can do so here: http://www.psyop.tv/main.php (click the MTV HD button at the bottom left).
Todd showed us how the raven from the piece was rigged. Basically, there is a single character and character rig, which performs everything that’s needed. There are controls to handle the orientation of the feathers on every axis, the wings as well as the rig were carefully crafted to fold properly, and there is also a layer of hair on top of the mesh to add that extra important level of detail.
Todd Akita's presentation for MTV's HD channel opening.
Details of the raven's rig.
More details of the raven's rig.
Last notes about the raven rig, few words about the particle flocking.
Right after Todd's presentation I headed to the Electronic Theater. The Electronic Theater features multiple shorts shown on “the” big screen (I believe it was a 4k screen). Before the shorts, there were interactive games where all the spectators would hold up a double-sided stick, with a different color on each side. There were captors on top of the screen, and it would pick the color of the side you were presenting to the screen. By showing up these colors to the screen, the crowd would control a puck running on the screen. The crowd was split in two teams. The games consisted of making artwork during the game play. Half of the crowd controlled the up and down movements, while the other half controlled the lateral movements. Ultimately, you get a crowd yelling "RED", "GREEN", "GREEN"! That was funny.
Many of the shorts were amazing. I will not discuss all of them, but will mention a few.
- Discord: metal and meat was really impressive. I liked the style, and even though there were no typical characters, there was still an interesting story.
- 458nm (http://www.cgportal.de/458nm/index.htm) had an amazing aesthetic.
- My Date From Hell (http://www.mydatefromhell.com/) was superbly executed on the artistic and technical levels, but found the story to be boring.
- One Rat Short (http://www.oneratshort.com/) was just awesome! Even though the story had no dialogue, the realism of the character rats was quite impressive, and their action carried the dialogue and emotion of the performance. A definite must see!
The only downside to the Electronic Theatre was the subzero temperatures from the air conditioners within the theatre. Between the shorts, many people would scream to (what I believe was no one) to turn down the air-conditioning. I had a long sleeve shirt on top of my t-shirt, and it was still cold. Even the tip of my nose was cold, I think I could see my breath!
The rest of the day was spent lurking around the various booths at the conference. I stumbled upon Bradley Gabe, so we started chatting. Upon doing so, Gordon Cameron, the father of the XSI Animation Mixer, now employed at Pixar Animation Studios, passed us by and engaged in the conversation with us. Afterwards, Brad, Graham (who also joined after some time), and I headed to meat up with Matt Lind and a few others to eat dinner at a restaurant. The evening ended pretty soon for me (at about 11 PM), and it was the first time in many, many, days that I had actually remained sober!
Thursday, August 3rd - DAY FIVE
I showed up at the Convention Center at 9:30 AM, to watch the presentation of Todd Lawrimore and Bill Blakesly, from Shaw Science. They gave a tour of how they create and use normal maps as well as rigid bodies. It was more of a brief overview than an in-depth presentation, but I was happy to bombard them with questions afterwards (in regards to normal mapping procedures since at the studio I work at, we are in the process of implementing more normal map procedures to replace the computational constraints of traditional bump mapping in our pipeline).
Todd Lawrimore and Bill Blakesly's presentation about normal maps.
After that, I went to attend Mark Schoennagel's presentation about GATOR. The thing is that I have never used GATOR for anything advanced, but since we are moving our pipeline to XSI 5.11, I figured it was better to know more about it and not need to use it, then to need to use it and not know how. The presentation was amazing. Mark transferred UV’s, envelope weights, and even shape setups between meshes effortlessly. It is tools like these that really put Softimage ahead of the competition. Between these tools and construction modes, it is impossible for an artist to back themselves into a corner. It truly is a non-linear and non-destructive system.
Mark Schoennagel's presentation on GATOR.
And finally I went to watch Jason Barlow’s (Omation) presentation about The Barnyard. Unfortunately though, due to legalities and the timing of the release and distribution of the film, Jason was unable to really go in-depth in the making of the Barnyard feature film.
Jason Barlow's notes on the Barnyard. On the right picture, a picture of the guys actually acting shots for the movie. Those were used as reference material by the animators.
To conclude, there were the sweet sounds of bagpipes. The official end of the SIGGRAPH 2006 conference had finally arrived. Everyone said their goodbye’s, packed up their stuff, and that was it for me! And that’s it folks!
* I could not remember or get the name of a few people. I sincerely apologize about this, I hope you don't hold hard feelings toward me.
Discuss Article at the forum
| |
|
