Muff's Modules & More

[lizlarsen]

In a raster order display when you are synthesizing shapes and patterns directly, creating the illusion of "depth" in a patch is a challenge. One technique would be the use of a low pass filter to blur out certain parts of the image. Another technique, used in this clip, would be to use certain output waveforms in the patch to control the brightness of the image. A combination of both techniques would probably be effective. If you have any ideas, please discuss.

[MrDys]

Nice. I really like the zoetrope effect going on around 1:23.

How exactly are you patching up the brightness modulation? Is it some kind of luma key?

[lizlarsen]

One of the inputs to the Video Blending Matrix in this patch is the triangle waveform of one of the audio-rate oscillators (which is modulating the oscillator we're actually seeing), and it is subtracting from the entire mix (this mix is mostly monochromatic, but i am tinting some shapes more than others.)

Another way to think of it is that the oscillator in question is both frequency modulating the oscillator that we are viewing, AND voltage controlling the brightness of the final output mix at the same time. Oscillator inputs/outputs can be more versatile (seemingly, to me) in a video synth, since you can control many different things with them with very different results -- where with audio synthesis, most of the time you're not interested in feeding every single output on a VCO into your output mix or to VC something in your patch.

To be purely technical, you can think of the red, green & blue inputs on the Color Video Decoder as "brightness CV inputs" for each color. So if you subtract from the mix going to the encoder, you are "darkening" the image.

Here's a pic of one of the LZX owners' systems I'm about to ship out, showing a patch that inserts a bunch of different sync'ed waveform outputs into the color encoder (the CRT image is a little washed out in this photo):


I'll make up a patch diagram illustration for this "folding cloth" patch in a bit.

[SynthPunk]

cool vid! I think ive achieved a similar effect in a similar way on my diy vsynth but ive not used a low pass filter with much success. do you mean like a normal audio range VCF? how would it be patched? when i was first prototyping mine i seem to remember terminally upsetting most of the ICs in a LM13600 based vcf that i patched some video signals into...

[lizlarsen]

Hey Rich! I mean a VCF similar to an audio range one, but with cutoff range up to 1 - 2MHz or so would be nice. If you CV it with part of your shape page (or an actual video image) it should create kind of a "blurring mask" for the processed image. By mixing horizontal and vertical modulation sources you could possibly create some kind of 3D filter of a kind...
So I guess the idea would be that control over brightness (gain) and have a direct relationship to low-pass (fuzziness) and brightness... so that things get blurrier as they get darker "further away", etc.

Kind of like the Buchla Low Pass Gate, how it simulates acoustic velocity by directly integrating low pass filter with a VCA.

[SynthPunk]

hmmm - d'you reckon it would be easy to convert a VCF to video speeds, maybe by simply changing the relevant caps for smaller ones? im guessing that maybe a simple sallen key design with as few ICs as possible, like the original ms20 type maybe? ive noticed when using one as a fader that incresing the resonance makes solid lines appear on the screen as the filter self oscillates at low (audio) freqs. or maybe the Bi-n-tic? hasnt it even got a sync input? ive not built one..

[lizlarsen]

Yep, but you'll want to use video rate OTAs. I have a video filter design I used in this clip (but high pass mode):


There are high speed VCF designs in the LT1228 and LT1251/1256 datasheets.

[giorgio]

the very end of the folding cloth video is great. i feel like these kinda effects really come alive when things are moving a bit slower such as 1:28-end

[lizlarsen]

[adamf]

I think the melting pot approach of using various techniques including direct video synthesis, digital and analogue processing, source mixing, keying and also feedback techniques can be a good way of simulating or giving the impression of depth. Alternatively, sometimes minimalist approaches or leaving space on screen for a pattern to breath and unfold can be a way to achieve depth. eg: when not flooded with content and colours that consume a whole screen, the mind can focus more on a pattern that emerges from a background and imagine it as a 3 dimensional entity that has its own depth and texture... possibley??? Sort of ties in with the animation speed discussed. Slowly evolving patterns may be easier for the mind to visualise as having depth and texture than rapidly changing ones?
Another thought: the juxtoposition of elements in a patch that are generating and modulated at different rates may help to give the impression of texture?? Perhaps getting too esoteric...

[lizlarsen]

Thats very true, about more minimal compositions having an impact -- something I need to work on. I find working in monochrome before colorizing the patch helps a lot, too, for finding the right dynamics for the image.

"Layering" is definitely a possible technique too! In the modular environment, with enough keyers and logic modules, it's easy to infinitely create separate layered elements, and then the illusion of making them appear in front of or behind one another more realistically could be done with filtering and brightness controls on each separate "layer".

I guess another big thing that's lacking in my own demos and patches is the lack of longer format evolving sequences (I need more low frequency modulators and sequencers in my system -- a new module in the works, that will be specifically good for that.) Being able to fade in and out different sets of modulation parameters, or signals into the image mix, etc. over longer periods of time can create more depth I think.

Another point is alternative colorspaces like YUV and HSB (speaking in the LZX context of course, with triple signals) is that in those colorspaces you can much more intuitively control the overall brightness of the image since it's on a single channel. Especially in HSB, since you can make things less saturated as they are "further away" on a single channel too. I need to get crackin' on YUV/HSB converter module prototypes.