Hey guys... Lets make a Competition 2.0 !

[TanookiMonk]

Hey everyone !! More than a year has passed since the last one ! Here is a link to it viewtopic.php?f=2&t=563

If you remember well, the first one was about no main use of kaleido, starfield, plugin and sound source. Well this time lets make the complete opposite.

So if you are interested, here are the rules that we could start with :

Any plugin and external material can be used but the final product must come from magic.
Make the most beautiful Kaleidoscope the entire universe crave for !
MUST absolutely contain a sound source and some kind of music visualisation with it waveform, spectrum or anything.
ONE SCENE ONLY and plesae something longer than 30 seconds XD
Once done, upload a video of it and post it on this thread.
Please add a little description on how you did it since we will not upload the source file in this competition.
Deadline on Febuary 1 ! If you want to enter. just write into the comments "Im in!"

Eric will be the judge, ok Eric !?!

Again please try to show what the engine is capable of !

[artnik]

I'm in!

nik

[TanookiMonk]

Ok i maybe a little bit early to post mine but i am in the mood for it !

So basicly here is what ive done :

I began to create this flame image using jwildfire :
https://dl.dropboxusercontent.com/u/56060582/Flame.png

This image create the background and all the black clouds that are coming by the starfield. I made a hypnotizing effect using the background to slowly go back and forth while varying the hue.

Here is what the magic project look like :

Magic project.png (739.47 KiB) Viewed 11460 times

And then i created my song in renoise :

https://dl.dropboxusercontent.com/u/560 ... ng%202.mp3

Here is the final product :

(For a better experience listen ind 1080p, fullscreen with a good volume)
(And feel free to use my Flame image and my song for your own use too !)

And its kind of sad the youtube codec dissolve the quality :S !

[Sadler]

This might not stay up very long because of the music but that's the inspiration.

As to how, it is probably pretty obvious - some bitmaps squirted through a kaleidoscope (and a 3d banana).

[D1gits]

Im in!

Just hope to find enough time to play around : )

[D1gits]

Had some fun playing around with 2.0 and some Dance Hall vibes : )

Based on a project I did with my group Sirqus Alfon with some new elements and testing some new ways to synchronies movements...

[blackdot]

Had some fun playing around with 2.0 and some Dance Hall vibes : )

Based on a project I did with my group Sirqus Alfon with some new elements and testing some new ways to synchronies movements...

very interesting . how do you get the cogs mapped on the sphere in realtime?

[D1gits]

Hey there Blackdot!

Thanks for the comment
I have used the BBC 70's Globerizer as shown on this picture:

A screenshot of the composition

Magic.jpg (437.02 KiB) Viewed 11318 times

[blackdot]

Hey there Blackdot!

Thanks for the comment
I have used the BBC 70's Globerizer as shown on this picture:

ah nice, thanks. guess I gotta take a second look at all those ISFs ^^.

[artnik]

Here's my entry – "LIDAR".

It's a single scene, utilizing a single ISF module, the generic extrusion grid "LIDAR" that I hacked out of TexF's amazing "Interstellar" space warp shader last year. I wanted to do something simple, but dramatic, that harkens back to the Rutt-Etra visual synth. The scene contains several instances of LIDAR with differing settings that are cycled through using a MULTIMIX module. It's an excellent way to make it seem like there's more going on than there is, akin to several playlist entries, but in one scene.

I've obviously interpreted the idea of a "kaleidoscope" very loosely, in terms of a roughly symmetrical tunnel-like view where the image is composed of small visual elements manipulated through rotation. The primary driver of the extrusion is a spinning waveform.

ORIGINAL: https://www.shadertoy.com/view/Xdl3D2
ISF HACK: https://www.interactiveshaderformat.com/sketches/1024

Music track is "The Place Beyond the Clouds" by MindsEye (CC-BY-SA), available here: http://freemusicarchive.org/music/Minds ... the_Clouds

[artnik]

Hey Sadler, are you a Mighty Boosh fan by chance?

[Sadler]

Ha! I haven't seen much of it but thanks for this. It does appeal to my sense of discombobulation.

I like your broken rectangle tunnel in 'lidar' - clean but twitchy.

[dj0le]

One scene, kaleidoscopes all over the place, and one spectrum in the center. Original footage was processed through Google deep dream before insertion into Magic,

[Sadler]

Like very mucho the vocals on that track.

If I may, what's your workflow for Deep Dream? Doesn't it take days for video?

[dj0le]

If I may, what's your workflow for Deep Dream? Doesn't it take days for video?

Basic Workflow - I dual boot Ubuntu on my laptop and use it to run the Fast-Neural-Style network with my own pre-trained models. As long as you have a good GPU and use the feed-forward process it isn't too bad, but it can still take hours depending on the total number of frames and the resolution you are doing. But, the network is fast enough to do low FPS webcam in real time which is pretty cool in some situations.

The model and instructions are here if you want to try it: https://github.com/jcjohnson/fast-neural-style

The end result is usually worth it though. It never looks good uploaded to youtube, but in person the detail is too stunning. I mostly use it to do fake low poly stuff in real time:

Flyer1.jpg (372.63 KiB) Viewed 11212 times

[Sadler]

Seems unix boot is the only way right now. Thanks for the link. And that image is intriguing!

If Windows users don't mind lengthy process then Style App from MacDaddy.io is fairly easy to use. It could use some polish and acceleration but it's pretty cheap to try things out with.

Here's one I did earlier...

[Magic]

One of my very good friends, Jon Shlens, was one of the lead researchers on the style transfer project at Google: https://research.googleblog.com/2016/10 ... nsfer.html

He told me that with a very fast GPU it can almost be done at 60fps. But for most users, it would probably be way too slow, otherwise I might look into implementing it for Magic .

[dj0le]

One of my very good friends, Jon Shlens, was one of the lead researchers on the style transfer project at Google: https://research.googleblog.com/2016/10 ... nsfer.html

He told me that with a very fast GPU it can almost be done at 60fps. But for most users, it would probably be way too slow, otherwise I might look into implementing it for Magic .

My experience with the fast neural style webcam I linked above is that I can get 60fps as long as the resolution is at most 640x480. That will use ~1 gig of video ram. Anything above that though, and it grinds down to like 5-10 fps. Of course, if you train your own model instead of using the defaults, you have to budget 2-4 days of letting your computer sit there running the training code first.