Side Project Ideas

[Ayydan]

I'm excited to see the output results! My input is actually pretty easy to work with, I just feed in a 1D array of uint8's that represent the image. The optimal modification for asciiplay would be if you could support yuv colour, so that way I could send grayscale, or append the chrominance signals to choose between colour and no colour

If you're just inputting a 1D array of uint8_t, that's basically when I'm doing as well.

I copy the entire image data over and can just index into the array to get the info I need. Since the image is scaled, I'll probably grab the luminance value of the scaled "pixel" around it and average that value.

As for your 1D array, it's just the greyscale values yeah?

Exactly, just grayscale values, which is why tuv would be so handy, because in theory I could just input 3 1d arrays with the same parameters each storing one of the components

[Jersey]

Ayyyyyyyyyyyyy lesgooooooooo

   

[Jersey]

Ayyyyyyyyyyyyy lesgooooooooo

Now all that's left is more image support, proper error handling, better encapsulation, YUV support, fractional scaling, optimization, better ascii rendering potentially and maybe optimizing the number of blocks/threads.

[Ayydan]

So you're basically done right? Hahaha

Ayyyyyyyyyyyyy lesgooooooooo

Damn Jersey that looks great!

Think we could port your code to x86-AVX? CUDA isn't supported on a vast many systems.

[Jersey]

Ayyyyyyyyyyyyy lesgooooooooo

Damn Jersey that looks great!

Think we could port your code to x86-AVX? CUDA isn't supported on a vast many systems.

I might look into that later. Maybe have a different backend for the actual rendering and use CUDA if it's supported and AVX if not (and it's support). That probably won't be till I'm basically done the CUDA version though.