pouët.net

So we now will have a category where we will be able to shine (I still remember the claps to Chaos Theory in the main theatre with nobody knowing it actually was realtime).

http://www.siggraph.org/s2010/for_submitters/live_real-time_demos

Since we are so proud (and often arrogant) about our coding and artistic skills, I hope we see lots of demos being entered and getting good mention. Will we able to impress them at all? Let's at least show them that realtime means A LOT MORE than dark corridors with weapon in the foreground, or molecular models and volumetric intestines in phong shading. Are we good enough to impress them?

Go go go!

Submitted as videos only! Great way to show the power of realtime! :)

Jury Criteria:
The jury will meet in May to review each submission on a five-point scale for the following criteria:
1. Innovation (pushes the boundaries of real-time technology)
2. Interactivity (how the submission is different from a linear, pre-rendered segment)
3. Creativity and originality
4. Interest and entertainment value for conference attendees
5. Production values (appropriate to context)

Makes sense to me, I don't think they want to spend their own time kkapturing. This makes me think how lucky we are to have ryg in the BP team.

About showing the power of realtime, I guess most people attending an event called "Live Realtime Demo" will get the point that it is Realtime, so the rest is left to the visuals to show them how powerful rt is.

The interactivity part indeed is a problem, but it's only 1 of the 5 points, so we can still get a 80% score (except the two FR and Plastic productions)

I am so interested in this! So, I think I will do whatever it takes to participate and I will make sure I personally come to LA to attend the event. Man, this is just plain awesome :D.

i'd be interested to know what's in their heads about what they want to see - a larger piece demonstrating a combination of lot of different (but possibly simpler) techniques (like a game), or one big effect.

yes, if one only could foresee what the judges would rank >)

iq, actually what I found interesting is that there's no way in their requirements for a jury to judge that the work was actually realtime.

Also, the "4. Interest and entertainment value for conference attendees" made me chuckle a little.

From what I remember every prod ran realtime this year.

Judging interactivity from a video doesn't sound like a good idea.

Can you rename this thread to "demos at breakpoint 2014"? It usually takes few years before demos use siggraph presentations.

btw, is it possible to display a video in 1080p at 60fps ?

jaw: if they do it at breakpoint I don't know why it wouldn't be possible at siggraph.

Why? By the time we get demos at BP, it's usually too late to find out whether they run nicely in kkapture. :)


All PC demo compos organized or co-organized by me were 100% real-time. (That's ms2002 and all BPs). Even if I didn't think that putting videos alongside real-time stuff would be unfair, it still would be grossly impractical to use videos. A lot of demos don't exit cleanly in kkapture, so you usually need at least two takes to get them right, shutting down recording at just the right moment. The resulting video files are huge. We need to encode videos since we don't have the hardware necessary to play back high-res high-framerate uncompressed video in realtime. Encoding takes time. To get good quality, you need really high bitrates, and playing back 1080p60 video with lots of motion and high bitrates requires significantly more CPU power than running the demo in realtime in the first place (it's lighter on the GPU though). Finding a good balance between optimal quality and the ability to actually play back the video at full framerate usually requires at least 3 full encodes for me. It's a goddamn mess.

I spent the better part of 3 weeks this year to prepare the videos for the scene.org awards, which was roughly as many demos as there are in a PC demo compo at BP. We don't have 3 weeks to prepare the democompo at Breakpoint; if things run smoothly, we have about 8 hours. Real-time is the only option; anything non-realtime takes too much real time! :)

Yeah, even high-motion video. But if you plan on doing it using a PC, you need a quadcore machine, and it very much depends on the codec. We use CoreAVC at Breakpoint to play back H.264 material. Even though H.264 is in principle more expensive to decode than MPEG-2 or MPEG-4 level formats, only new codecs are designed to really take advantage of multiple cores, and you need it at this resolution with high bitrates.

For the last two Breakpoints, we actually ended up transcoding some of the Animation compo entries to H264 because the original XVid codec couldn't play them back in realtime on any machine we had at our disposal. Good thing we have a small encoding farm at BP, else this would've sucked :)

That's for judging purpose only, if you read carefully, the maker of the 'demo' will have to go there and present the product, having 3 to 5 minutes to show it off, in real-time.
It's totally understandable that they don't want to spend time learning a statistics tool or a game just to see the potential of the application.

this is actually the second time there's a realtime category.
see last year here:
compo rules
entries

if there were really just these 4 entries, the bar wasn't all that high yet... but if the competition gets tougher over the time, demos may actually fit better to the computer animation category. i'm just saying that because a part of jury vote seems to be about interactivity.

knos: folks are even able to play the entries themselves. so if you have to go interactive, realtime or prerendered is not really an issue...

basically it seems to me that they just want to see games. but it could be interesting to see how they react to a demo or a generative demo...

Iq:

I suggest you to present an interactive version of Elevated, using the most powerful as possible GPU you can find for it, probably fully raycasted, and make the video explaining it is fully procedural. Probably you can make a little GUI for it in few kbs, so it continues being a little prod.

That's a marvelous idea! I second that.

Yay! That's a wonderfully ignorant procedure worth exploiting, right there!

1. Contribute non-realtime, wholly unrealistic video of amazing lies to a team of judges wholly incapable of judging the realtime aspect.
2. Arrive on stage naked, flinging poo at the audience during the presentation of the "actual" product.

So yes, it's essentially the process of how games evolve from tech demo to actual product, except nobody gets paid.