|
|
Post by nathanmyersc on Mar 25, 2024 13:35:50 GMT
|
|
|
|
Post by reden on Mar 25, 2024 13:47:58 GMT
This thread has been moved to the Corkboard, as it is here where members can advertise their services.
|
|
|
|
Post by AnthroHeart on Mar 25, 2024 19:05:42 GMT
Thanks reden. I didn't think of doing that, but great idea.
|
|
|
|
Post by AnthroHeart on Mar 26, 2024 18:23:15 GMT
How did you create the WAVs on your Patreon?
Is it something like my wav converter that converts any image into a WAV?
Did you find a good algorithm to map the bytes of the image into a WAV?
How do you map a pixel byte value from an image to a sample in a WAV file?
|
|
|
|
Post by nathanmyersc on Mar 26, 2024 18:43:09 GMT
How did you create the WAVs on your Patreon?
Is it something like my wav converter that converts any image into a WAV?
Did you find a good algorithm to map the bytes of the image into a WAV?
How do you map a pixel byte value from an image to a sample in a WAV file?
I find that real photos of actions and objects hold power. also for pixel in pixel_data: # Extract RGB(A) values if has_alpha: r, g, b, a = pixel adjusted_data.extend([ 32767 - r if r % 2 == 0 else -32767 + r, 32767 - g if g % 2 == 0 else -32767 + g, 32767 - b if b % 2 == 0 else -32767 + b, 32767 - a if a % 2 == 0 else -32767 + a ]) else: r, g, b = pixel adjusted_data.extend([ 32767 - r if r % 2 == 0 else -32767 + r, 32767 - g if g % 2 == 0 else -32767 + g, 32767 - b if b % 2 == 0 else -32767 + b |
|
|
|
Post by AnthroHeart on Mar 26, 2024 18:53:54 GMT
Does that map the full range of pixel values to the full range of WAV sample values?
What is the range of r for instance? Is it 0-255?
|
|
|
|
Post by AnthroHeart on Mar 26, 2024 18:56:42 GMT
And if the pixel value changes every sample, is that too fast for the speakers to be able to respond when playing back the wav?
Should it glide gracefully over 10-100 samples between each pixel to give the speakers time to adjust?
|
|
|
|
Post by nathanmyersc on Mar 26, 2024 19:17:09 GMT
And if the pixel value changes every sample, is that too fast for the speakers to be able to respond when playing back the wav? Should it glide gracefully over 10-100 samples between each pixel to give the speakers time to adjust? Good questions, its experimental. i find the loudest i can be is the best. so i map it so 32767-value remains near the loudest. I suppose we could hold onto the current note for 100 samples or so and then queue the next pixel. its a thing to try. Like i could duplicate each R G B A 100 times or 10 times or whatever. and see how it goes. |
|
|
|
Post by nathanmyersc on Mar 26, 2024 19:50:22 GMT
i did some hilarious things they all sound vastly different its a very interesting idea. here test these. |
|
|
|
Post by AnthroHeart on Mar 26, 2024 20:00:09 GMT
Try this one. In between each pixel, I converted it to the decimal value of the RGB. I created 100 steps to gracefully change the WAV sample value from one pixel to the next. And alternate between negative and positive WAV values each pixel.
I used your Marijuana image from your Patreon as the input.
It created a WAV output 48000 sampling rate, mono.
And the Python code:
It requires: pip install numpy scipy Pillow
Try it if you like, and let me know what you think. I think it's better to gradually go from one pixel to the next so the speaker has time to adjust.
|
|
|
|
Post by reden on Mar 28, 2024 1:48:38 GMT
And if the pixel value changes every sample, is that too fast for the speakers to be able to respond when playing back the wav? Should it glide gracefully over 10-100 samples between each pixel to give the speakers time to adjust? You should do 10 samples instead of 100, if you are going that route, as speakers aren't weaklings and can handle it. |
|
|
|
Post by nathanmyersc on Mar 28, 2024 3:33:47 GMT
Try this one. In between each pixel, I converted it to the decimal value of the RGB. I created 100 steps to gracefully change the WAV sample value from one pixel to the next. And alternate between negative and positive WAV values each pixel.
I used your Marijuana image from your Patreon as the input.
It created a WAV output 48000 sampling rate, mono.
And the Python code:
It requires: pip install numpy scipy Pillow
Try it if you like, and let me know what you think. I think it's better to gradually go from one pixel to the next so the speaker has time to adjust.
Im going to try and make my noise less jarring. going for wind like sounds or fire like sounds or water like sounds. messing around with gpt to try and emulate these sorts of results. |
|
