PNG to Wavetable for eXpanse

[FLVZ]

Is it at all possible to convert images into wavetables without having to buy Serum. I have eXpanse, no intention at all to buy any software synth in the near future. Is there some kind of png to wavetable VST or just standalone app maybe ?

[diminished]

In theory, all you need is to pick any set of vectors from an image. Like, one row of pixels == one row of the wavetable.

[dpcoffin]

That's a useful video, thanks! I'm even more of a graphics guy than a musician, by training anyway, and spent a couple of days full of excitement doing this with Falcon (it's the only way to directly create WTs for Falcon, though you can import Serum ones). Sorry to say, less than 5% of all the images I created and tested seemed even marginally useful or even interesting, despite looking like they'd be cool. All sorts of dramatic visual changes hardly made an interesting difference to the sounds, and I don't recall ever actually using any of the results, since there are already thousands of much more interesting sounding Serum-type WTs available for free, and exploring these eventually seemed like time much better spent.

Of course, YMMV; just saying I wouldn't loose much sleep over not finding a cheap way to try this out. Speaking of which, I've never seen a device for doing this that isn't inside something else that's very expensive, eg: Serum, Falcon or Live 10 Suite, but the M4L device that does this for Live is free (https://cdm.link/2019/02/free-image-wav ... eton-live/), so maybe rent Max for a month and give it a go if you really want to check it out without spending a bundle?

Of course, you could rent Serum itself for a month, too: $10

What I recall I learned from the experience, fwiw, is that wavy or diagonal verticals hardly sound different than the same shapes that aren't wavy or diagonal, nor does moving any shape side to side in the image. Changes in width and value, as well as gradual value changes compared to abrupt ones are where the main noticeable sound and sound-change events come from, because that's what produces changes in waveform shape. Just moving the same waveform from side to side isn't very dramatic if even noticeable. But I can certainly see how once you get what does what, this could be a very fast way to create WTs compared to all the algorithmic and wave-combining ways, given the tools for doing it that way.

[mjxl]

Unsure what exactly Serums output looks like, but try this one out at least !
https://www.google.com/amp/s/www.howtog ... files/amp/
he created sounds posted there sound usable :0

[Catblack]

https://github.com/PalmerHogen/ImagesAsWavetables

Haven't tried this, but looks interesting.

[FLVZ]

https://github.com/PalmerHogen/ImagesAsWavetables

Haven't tried this, but looks interesting.

Managed to get Audacity to convert the images, some very interesting movements in the audio. But the amplitude is over 700db and is frying every VST i have as well as expanse when i try to scrub the sample position knob after loading it. Managed to reduce the amplitube and threw a few compressors at it once quieter, still seems be corrupting RX Audio Editor

png wav glitch.png (226.75 KiB) Viewed 1381 times

png wav glitch close up.png (58.32 KiB) Viewed 1381 times

png wav glitch close up stats.png (41.1 KiB) Viewed 1381 times

[FLVZ]

Unsure what exactly Serums output looks like, but try this one out at least !
https://www.google.com/amp/s/www.howtog ... files/amp/
he created sounds posted there sound usable :0

The funny thing is i ended up with better results than the ones given here (converted to tiff) but just cant use the wavs for anything. i dont know if audacity is writing wavs in a weird way ? but theres no stats for the files exported as pictured above

[EpiGenetik]

After reading this thread I gave this a go. Smaller less complex images work best I find.

http://flexibeatz.weebly.com/paint2sound.html

[FLVZ]

After reading this thread I gave this a go. Smaller less complex images work best I find.

http://flexibeatz.weebly.com/paint2sound.html

Thanks for the input.

I have since succeeded. It turns out I need to export the file as a 16bit file for it to function normally (was exporting at 32 & 64bit floating point). My assumption is that there is a limited dynamic range with 16bit so the crazy clipping integers are mathematically struck off ? Therefore avoiding the file from corrupting. There is a lot of consequent DC offset but Izotope RX can take care of that like a piece of cake! eXpanse accepted all 16bit files I created problem free

I DC filtered the sample then used KClip in tape mode to tame down the clipping

success.png (157.84 KiB) Viewed 1198 times

[dpcoffin]