Book Review: Sleights of Mind

I really enjoyed this book. A couple years back I read a couple books on performing magic as I thought there were some interesting bits that could apply to giving presentations, product demos and the like.

Sleights of Mind does reveal some secrets about how certain magic tricks are performed, but it's written by two neuroscientists who spent years researching magic (collaborating with magicians) to determine how and why we are susceptible to certain illusions and deceptions.

I not only learned a lot, but it's given me a lot to think about in terms of the future of graphics, user interfaces, gaming and more. Fascinating stuff!

Sleights of Mind: What the Neuroscience of Magic Reveals About Our Everyday Deceptions

MIGS (Post 3 of N): Jason Mitchell's talk

Jason Mitchell (former ATI, now at Valve a couple years) gave a talk on "Connecting Visuals to Gameplay)

Rough notes below. Lots of screens and some video from Team Fortress 2 and the recently released "left for dead". Worth googling both before you read the notes below.

We'll look at two games today

TF2

- Distinctive silhouettes

- Stylized rendering

Left for dead

- Dark, gritty horror

- “filmic effects"

- Lessons from TF2

TF - Orignally as a quake mod 10 years ago, then half life mod

Class is selected by players. 

Initial TF2 1999 screenshot.

- Screen showing more “realistic” FPS, nothing distinctive

- Evolved to be stylized

Why?

- Gameplay (different classes)

- Readability

- Branding

Read hierarchy: What does player attempt to ascertain?

- Friend or foe (color)

- Class – run or attack (distinctive silhouettes, Body proportions,  Weapons, shadows, hats and clothing folds)

- Selected weapon – what’s he packin?

- Highest contrast at chest level

- Greadient from dark feet to light chest

- Lots to draw your eye up to chest level.

Early 20th centry conmmercial illustration influences.

Cornell, Leyendecker, Norman Rockwell

JC Leyendecker was biggest influence

- Clothing folds

- Rim highlights (light source orthogonally) – helps silouetttes pop.

- “red terminator” – where normal crosses orthogonal – increases saturation at that point, makes it red – actually makes sense inf you think about subsurface scattering

Screenshot from early short videos (“meet the heavy”)

- Before and after 2D paintover to make the image pop – rim highlighting was awesome.

Character creation

- Character silhouette (showed silouette of elephant, everyone gets it, despite it being orange)

- Bulding block, Identifiable at first read

- Interior shapes, Keep it iconic

- Work out design in ¾ pose

- Model sheet

- 3D model

- Front rear views

- Base ambient occlusion map, that's then used as a guide for painting skin

- Final character

- Iterate on the above

Environment design

- Create a compelling immersive worlds

- Team districntion through material hue/saturation/etc desaturate relative to players.

- Uncluttered painterly look.

- Bases – blue featured concrete, steel; red featured wood, sand.

Miyazaki was an influence – brushwidth foreshortened example. 

--

Left for Dead

Co-op first person horror, Dynamic shared narrative – "experience an action movie with friends"

AI director

Photo  - The valve “Shipping machine” (When games go live, they have a BIG RED BUTTON on a control panel of sci-fi proportions that Gabe hits at midnight. This then sends an 'enter' keystroke to a person's PC. Awesome. Took a pic, will post soon).

Gameplay movie (awesome).

Lessons learned from TF2

Filmic effects

Shaders enhance dark setting.

Filmic

-Color correction

-Grain

-Vignette

-Local contrast enahancement

-Dynamically communicate game state

Showed step by step

Color correction made it a bit greyer, desaturated

Grain – detail in greyed out darker areas

Vignette – mainly along top, to focus attention down at center.

Local contrast – highlights area around the player

We’re not film, we’re an interactive medium, so we might have info and cues we are looing to give to player

“sideband communications channel” like music score to film director

Gave example of normal stress level vs high stress

(local contrast driven higher, more stark

"Third strike", totally washed out, stark contrast - almost black and white.

[one note is that all the filmic effects were weilded subtly, but in sum were dramatic. good lesson here]

Lighting for darkness

Support fiction

- Fires

- Headlights of abandoned vehicles “clearly something has gone wrong”

- aid naviation - players tend to follow light.

"Smoking the set"

- Separate foreground from background

- - fog, light colord fog in dark areas to contrast with silhouettes, of infected in mid-ground

- Particles – adds atmostphere and helps accentuate silhouettes.

- Subway example of grey fog.

- Particles coming up from manhole

Reload shove and muzzle flash

- Player is the light source

- - increase drama and immersion (when Flashlight is attached to the weapon, and you are using it to light up a dark hall, reloading has consequence).

Self shadowened normal mapping

- Normal mapping locally alters surface orientation, causing detailed lighting effects

- HL2 “radiosity normal mapping”

- Turned out to be free by refactoring shader code

[note effect here was noticable but not huge, felt like he delved a little to deep into this one vs others. Hmm...]

Wet environments

- Film technique

o Wash down the set to geth that moview dark look

o Film noir

- Adds details to dark settings while still feeling dark

Then showed gameplay elements to show the above.

Count the monitors

French-canadian news clip on day traders playing the current market volatility. The guy in the red sweatshirt is my buddy Dave.

For those of us in tech, it's fun to watch just to see the crazy multi-mon goodness. I think the biggest setup I spotted was 14 monitors plus a TV. 

Sightseeing in Liberty City

Awesome flickr set of side-by-sides of GTA4 screens and NYC pics.

Cult of Personali-tea

Several folks linked last week to this story about a woman in the middle east arrested for joining a cult that worships a giant teapot.

While the lack of religious freedom is sad, I have to feel that the majority of folks linking to it were doing so more because they found the idea of a cult of people paying worship to a teapot to be shockingly newsworthy.

Evidently, none of these people have been to Siggraph. :-)

Unintended Effects

Like many, I've been using the holidays to catch up on a backlog of games. This being such a killer year for content, this is a problem, and I haven't even touched a bunch of the "must plays" of the season (CoD4, Assassins Creed, and *bowing head in shame* I haven't picked up Portal yet). Still, I have a year's worth of games with the shrink wrap on them, so I figured I'd get through them before buying more.

One of them is Blue Dragon. I'm not a fan of Japanese turn-based RPG's in general, but I'd heard it was good and I got a free copy, so figured I'd give it a go.

I'm still not a fan of turn based combat in these games, and I have no patience for the character power/attack customization pedantry that Blue Dragon presents me with. Nothing wrong with it, just not my thing.

One thing I did note, though was excessive use of depth-of-field shader effects in both the cutscenes and realtime animation. I think it's a good study in excessive use of a visual effect that is detrimental to the end result.

Depth of field effects can be used for a couple things. It can be used to draw attention to a particular element of a scene (in particular when dynamically drawing the focal point from one distance to another, for example). It can also be used to suggest scale such as in macro photography.

In Blue Dragon, there's a lot of grandiose scenery and scenes in which they are trying to suggest an immense scale. However, in overdoing the depth of field effects, I found it reminisent of tilt-shift miniature faking photos like this:

(from flickr tilt-shift miniature fake pool)

Some examples from Blue Dragon:

I find the bottom shot in particular has a look like an animated scene of plastic toys, not a vast, grandiose world. Perhaps that was the look they were going for, but I don't think so. I think it was just a case of the team getting carried away with the latest shader effects.

There's a lot to be said for using such effects with nuance, but it's difficult to do in an interactive medium where you aren't sure how much it's going to affect the player - esp since some will be playing on different screen sizes/resolutions, etc, and where you are trying to compete for screenshot/trailer drama with a bunch of other titles.

Carmack's Back

Having not heard anything from the guys at Id in a while (then again, I don't follow the hardcore space so closely anymore), a long interview with him popped up on Game Informer.

I've always been a fan of Id's games, and of Carmack as a developer. While they seem to have fallen behind guys like Valve and Epic in terms of their development model (old skool small team vs large team w modular development, etc), I secretly (not so secretly, I guess) root for Carmack to come out and ship something that kicks everybody's ass.

Of note, this quote about Gfx HW, and when and what users should upgrade to:

I don’t think that there’s any huge need for people to jump right now. All the high-end video cards right now—video cards across the board—are great nowadays. This is not like it was years ago, where they’d say, “This one’s poison, stay away from this. You really need to go for this.” Both ATI and Nvidia are going a great job on the high end.

Wow. It's not just me then. If HE doesn't care about the latest and greatest graphics, then who does?

Also, a game-biz-101 bit from Todd Hollenshead for those that think that digital distribution will make their publisher-dependence woes go away (Steam is the topic of discussion, but to be fair, they are just a proxy here for all digital distribution services):

there were serious flaws in the economic analysis that [Valve] laid out for developers. The problem for most developers is not one of not getting paid enough once the game is out, it’s that they don’t have the seed funding necessary to internally fund development of their titles. That’s why they work for publishers on milestone schedules and advances against future royalties, and Steam offers no solution for that. It also doesn’t offer any solution for the marketing spend question, where if developers don’t even have enough money to fund themselves internally to develop their product, then they’re not going to be able to pay for a multimillion dollar marketing campaign, which is a huge amount of risk that as an industry standpoint is offloaded from developers to publishers.