Fantasy Graphics League 2005 Results

First, human or monkey? Human! Second, which human? This one:

Yaohua Hu

(note the ambient fill light worn)

233.43  name: Yaohua Hu, lab_name: MSRAIG
231.50  name: Jessica Hodgins, lab_name: hoping for the best
158.50  name: Aila, Lehtinen, Laine and Miettinen, lab_name: rand()%745 + 1
151.57  name: Tom Mertens, lab_name: Breakfast Clubbers
150.00  name: Doug Epps, lab_name: Everything.com
143.57  name: Stephen H. Westin, lab_name: Big Red
136.00  name: captain, lab_name: big researchers with dental failure
135.00  name: Abhijeet Ghosh, lab_name: Imager Graphics Lab
135.00  name: Alan Chalmers, lab_name: bristol_bretheren
123.57  name: Zhaosheng Bao, lab_name: fracture
The sketches did not affect the top slots; Stephen moved up and Abhijeet and Alan moved into the top ten. Jessica had a number of second authors for sketches, but only the first authors count (and since a sketch does not count as a publication, I guess such people should be called first presenters or somesuch).

But what of the monkeys? They were with banana-chucking distance of the two human leaders, didn't they benefit from the sketches? What's wrong with you, didn't I already say the monkeys lost? Why are you asking these questions?

213.14  name: Fundi Washoe Tomasello, lab_name: Monkey #509
210.00  name: Fifi Sarah McGrew, lab_name: Monkey #412
189.24  name: Fifi Cheeta Premack, lab_name: Monkey #692
185.00  name: Faustino Birute Sanz, lab_name: Monkey #333
172.00  name: Flint Washoe Varki, lab_name: Monkey #901
171.14  name: Fred Austin Fossey, lab_name: Monkey #185
161.14  name: Flo Sherman Varki, lab_name: Monkey #970
150.57  name: Ferdinand Dian Premack, lab_name: Monkey #624
146.14  name: Fred Crickette Rumbaugh, lab_name: Monkey #745
144.50  name: Figan Sherman Fossey, lab_name: Monkey #178
So seven monkeys beat all but two humans. Fifi Sarah McGrew, who was in thirteenth place, kicked up to second place; quite impressive, a gain of 80 points on sketch authors! Nonetheless, the flying monkeys lost. Still, by watching the Electric Sheep screensaver long enough and voting for the results they liked, new patterns evolved to the point where uncharted monkey neural pathways were opened up and the whole clan was able to use mind-power to instantaneously transport themselves away from the Earth to a better place. Well, at least their logic was that monkeys like being warm, the sun is warm, but in retrospect perhaps that wasn't the best place to mass teleport.

But anyway the humans won, and here's Yaohua's winning team:

 1  cost: 210  value: 111.14  Heung-Yeung Shum
 2  cost:  90  value:  61.14  Baining Guo
 3  cost:  20  value:  30.57  Kun Zhou
 4  cost:  45  value:  20.57  Xin Tong
 5  cost:  25  value:  10.00  Lifeng Wang
That's right, just 5 researchers on this team, which yields huge savings in office space and paper clip costs.

If you look closely at the names, you'll notice they're all Irish fellows. In the "it's a small world" interesting factoid category, Baining Guo was a housemate back around 1986 here in Ithaca. I'd like to say I taught him all he knows, but at the time he wasn't involved in computer graphics. Let's cut to the chase and interview the winner.

Congratulations on winning the FGL title and defeating everyone else, including the monkeys. Was this your first time entering the Fantasy Graphics League?

Thank you and the competitors. Yes, this is my first time.

How did you hear of the Fantasy Graphics League?

I received an email which was sent to some graphics people inside our company.

It looks to me like you picked your team from Microsoft Research Asia, where you're employed, and hired just 5 researchers; this is a record for fewest people on a winning team. Did you feel like this strategy was a good one, or were you simply just picking people you knew and seeing what would happen?

Oho! I did not expect that we could make this record. If I wished to be the winner (in fact, I did not think about it at all), I should have hired more people. So, the fact is, I simply picked some of my colleagues and bosses.

You mentioned that "I did not understand SIGGRAPH at all" in an earlier note. Could you say more about this? Will you be at SIGGRAPH this year?

I'm interested in graphics research for products, which is different from most SIGGRAPH papers. So it is a total accident that I won the FGL. Yes, I will go to SIGGRAPH this year.

Give us a short biographical sketch, if you could: degree, field of study, current employment, research interests.

I have a bachelor's degree, and majored in electronic engineering. I work for Microsoft Research Asia. I'm interested in graphics research for products, especially computer games. I wish my technologies can improve gameplay.

What does the "IG" in MSRAIG stand for? Google turns up nothing (except the FGL page, and some page on Clan Family Tartans).

IG = Internet Graphics. It's our group's name (our lab has many groups). The full name is Microsoft Research Asia Internet Graphics Group. But I think 'I' should stand for 'Incredible'.

The monkeys would have used their lab to research new screensaver techniques. With your winning lab, what would you want to research?

If I 'have' this lab, I wish they could focus on game graphics research and push new gameplay (instead of pure graphics) for our users.

Heung-Yeung Shum was worth 210 Quatloos this year, and will cost around 320 Quatloos come 2006. Will you still hire him at this price?

I don't know. Since he hired and still trusts me, I will probably treat him the same way. He will be responsible for the coding, so that he can be the first author and win more FGL points :). Iím kidding.

So there you have it, beginner's luck, if you can call someone at one of the most productive graphics research labs in the world a 'beginner'.

Anyway, this may be the last FGL, unless it isn't. My hopes as a SIGGRAPH Sketches committee member for dirty money came to nothing again this year. So the option of doing the FGL again for a seventh year in a row and firmly establishing myself as a one-joke has-been is pretty tempting. (cue scene of bedraggled author drinking whiskey from a greasy bottle and swinging a handgun around, muttering about how wavelets should have been called wavesicles.) Sorry, been reading too much modern literature lately...

If you want to check your score or see the values of any researcher-for-hire, the list is here.

This just in from Matt Pharr: For giggles, or something like that, below are the top 15 FGL scorers of all time. Those with **s are the winners for their year. Some interesting stuff there, if FGL is the sort of thing you find interesting...

Note that Antoine McNamara's 2004 score, all time #3, wasn't enough to win that year, even though it was much higher than his 2003 winning score. Tim Purcell comes in as highest scorer with no outright win in any year. A good chunk of other folks are distinguished by scoring higher in their particular year than the winner in a previous year.

 1. 326.67  ** Aaron Hertzmann, Super Butter Dog (2002)
 2. 292.00  ** John Hart, The GPU Abuse Center (2004)
 3. 282.50  Antoine McNamara, The Inverse Kinematics (2004)
 4. 277.57  Tim Purcell, Wunky Munky Du (2002)
 5. 275.00  Mira Dontcheva (2004)
 6. 271.00  ** Matt Pharr (2000)
 7. 265.00  Hugues Hoppe, Shlomo & Hugues (2002)
 8. 255.24  Ian Buck, The Frys Addicts (2002)
 9. 254.00  ** Hanspeter Pfister, Beerflakes (2001)
10. 252.57  ** Antoine McNamara, The Inverse Kinematics (2003)
11. 243.57  Henrik Wann Jensen, The Vikings (2002)
12. 240.00  T. Aila, J. Lehtinen, V. Miettinen, lab name: rand()%680 + 1 (2004)
13. 233.43  ** Yaohua Hu, MSRAIG (2005)
14. 232.50  Aaron Hertzmann, Funky Homunculi (2004)
15. 231.50  Jessica Hodgins, hoping for the best (2005)

To relive the past glories and recall how you screwed up:


From the medulla oblongatae of Eric Haines, Phil Dutré, Dan Kartch, and Ben Trumbore.
last updated: June 23, 2005