Moho Forum

I'd like to share some preliminary results of the generative music-to-visuals code I've been working on.

There's no moho involved in this at this point, but I could see myself using something like this as an animated background image layer. The way the blobs are randomized and rendered is mine, the way the image changes over time is mine, and the way it's filtered to "match" the audio spectrum is also mine. The music ain't mine, but as I continue to improve on this I intend to use it on some of my own music.

The filtering effect can be used separately, on any other images; it doesn't have to be the random blobs.

Some things I want to improve:

*The aliasing on borders between areas.

*Add the ability to manually zoom in on different parts of the image, depending on what part looks best ... as it is now, it's kind of backgroundy and not focused on anything (although if used as a background, that's not a problem). This is something that could be done after the fact in Moho, now that I think of it, but the detail will be sharper if I zoom during the initial rendering.

I have lots of other ideas branching off of this, but it all takes a lot of time to realize. Hope you like it so far.

Haven't gotten any feedback on this yet ... maybe it got lost in the shuffle? Anyone interested in this kind of thing? Even in just the background or whatever?

It's a very neat effect. What did you use to do this? Was it pure programming and math, are you using some of the shelf software, or a combination? I'd love to see a higher resolution version of it. It has a nice hand drawn quality to it.

Is it linked to the music somehow? If so, what does it look like with other music or voice?

Pure programming and math. C++ code compiled in G++, generates a series of bitmaps, and then I use another app just to convert those to a video file.

It's linked to the audio in two ways: one, when there is more change in the audio (from analysis window to analysis window), the shapes move faster; two, each frame has a visual equivalent of a "graphic equalizer" applied to it (i.e. high frequencies emphasize contrast and noise) modeled after the audio as well. These are separable; the filtering effect can be applied to anything. I think I'll run it on a regular photograph next so you see what I mean.

There are some clearer, still examples of images generated with the same code here.

The still pictures are really nice. I especially like the last two - with the more subdued colors and the blur effect. It's just a personal opinion, but I like how those two make me almost think I'm seeing something, even though they're just as abstract as the others. I'd like to see those ones animated.

Yeah, the effect of increasing blur as it gets further from the center of the picture almost has a simulated depth-of-field effect, even though there's no actual depth or depth information.