Adding Images to Serum Wavetables - Combining Art and Sound

One of Xfer Serum's many offerings is the ability to load images into the wavetables to create brand new sounds. This, for many is an experimental game, but I'm going to show you how to design and pick images for what you need. Also, if you haven't already, check out our free wavetable pack, much of which used techniques outlined below.

Firstly, I'll show you the absolute basics, then we'll go onto more complex images, I'll show you what is good for both hard and soft sounding wavetables, and then it's on to you to make some absolute sonic gold!

The above image has a black side, and a white side. The way Serum converts an image into wavetables, is it looks at the colour, and depending on how dark it is, assigns the waveshape. In this instance, loading this image into serum will give us the following:

Its results in a nice Squared wave. This is pretty good as a single waveform, but serum offers a whole wavetable where you can travel along the y-direction. If we move the wavetable position in this, nothing changes. Let's resolve that by adding a slope in the image:

Now, when you change the wavetable position, you get a kind of pulse-width modulation. This means we can get different harmonic content from just moving the wavetable position, which is very useful. Let's make our image a bit more complex:

As you can see, the more complex the image, the more complex the sound is. These waves are very square and the more changes from black to white across the photo, the more parts (or harmonics) there are in the resulting wave. So far, we have just looked at black and white, resulting in a very square wave. Lets look at a gradient.

As seen here, when we blur black, through grey, into white, we get a slope instead of a square wave. This is because the height of the section of the wave is chosen, by it's shade. This means that black is the bottom, white is the top, and grey falls in the middle. By using greys, we can get smoother slopes, and a rounder, less harsh sound.

The above two images were loaded into Serum (below) and you can see the effect of blurring the hard lines. The blurry image results in a much rounder, smoother waveform, despite it's complexity. 

Using these techniques we can create complex waveforms. I enjoy using Microsoft Paint to draw in waves as an experiment to see what is possible, I often then open the images in Pixlr which is a free online image editor to add filters and effects.

Effects like Kaleidoscopes make interesting results with wavetables which often end up symmetrical, this means that the waveform will be the same at 0%, and 100% but changes in the middle. With the Kaleidoscope image above, it is bordering on "too-complex" where there is so much harmonic content and it starts to sound quite harsh. So this does mean that there is an upper limit if you just want to use the raw waveform, however it does lead to a much more interesting sound design if you decide to incorporate filters.

The complexity issue can be quite difficult to predict, the image above seems quite simple but with the amount of horizontal lines, the waveform will jump from black to white a lot which means that sweeping through a wavetable (unless you morph the frames) can be quite stuttery. Images like the one below have lots of fine detail in high contrast, which leads to lots of very high harmonics too - take any "slice" and the number of times it changes from black to white is the number of sharp peaks in the waveform.

This means that we want to pick photos which don't have much sudden changes in any horizontal slice. The image below is an example:

Likewise, there's no point choosing a picture where the majority of it is one solid colour as most of the waveform will remain the same:

In the photo above, the whole top portion will just be maxed out and no resulting waveform is likely to occur.

This brings us onto how to choose the perfect photos for the job of making nice waveforms. From what we've discovered above, for harsher sounds, you can get away with harder contrast, but you still don't want it too busy.

I've always found warning signs to be quite good for heavy sounds, they have a stark contrast, but are never too busy. For Softer sounds, you have lots of options. Anything which is lower contrast (with lots of greys between black and white) and not too busy horizontally can lead to nice smooth waveforms.

Height maps are one good option as the gradients are normally quite smoothed out and there isn't too much horizontal busy-ness. One of my favourites is pictures of wood, which have a fair amount of detail but the contrast is fairly low between any bit of detail.

Other good examples are portrait photos, which depending on the hair especially, can make excellent softer wavetables.

All these photos have been made in MS Paint or found online, and there are plenty of pictures everywhere across the web to try and input into Serum. This guide went into quite a bit of detail, but until you play about with it yourself, there is not much more I can say. 

Remember, any photos need to be a .png file but that is easy to do when downloading from google images as you can just change the .jpg in the file save name to .png. 

I guarantee you that 90% of the sounds you'll end up making will be rubbish, but remember to save any of the wavetables you like!