"Hi, im kind of at an alright level with pure data, and im working on a project, a spectral synthesizer in pd vanilla and Ofelia,
The way im planning on handling this is to have a lowpass and a highpass squeezing in one one chunk of audio. ive realized that the lop~ object and the hip~ object have a frequency rolloff that smoothens out the cutoff over time. its like a gradual slope that gets quieter before completely choking out the audio. i am aware that this is how most filters work,

but is there a way i can have a square shape filter that just completely straight up does not let any frequencies above the limit at all? im fine with using externals if i need to, or if there's some kind of vanilla work around. it needs to be done in real time."

SOLUTION: My friend Mike Moreno found a way to do this with the "I03 resynthesis" patch, shaping the table itself with sliders and an until loop. it creates a square shape on the frequency table itself. heres the patch: resynth-1024.pd
Credits to @mianmogra

@Kitty-Dyson This might help....... https://forum.pdpatchrepo.info/topic/9564/is-this-msp-patch-amplitude-gate-possible-in-pd-extended/4 although you should expect some artefacts I think.
David.

@Kitty-Dyson because the inverse fourier transform of a square shape is the sinc function, the perfect lowpass filter kernel is the sinc function. This function is impossible to be applied because it has components that go off to infinite positive time and infinite negative time. However, there are such things as "windowed sinc" functions which will introduce a phase delay in the output that you may be able to use in real-time. The term for these "absolute" filters is "brick wall" filters.
https://en.wikipedia.org/wiki/Sinc_filter

@Kitty-Dyson as far as practical filters in Pd, iemlib has some steep filters named lp8_bess~ or lp10_cheb~ etc, all of these types of filters have different tradeoffs in phase and frequency response, and "ripple".
https://en.wikipedia.org/wiki/Chebyshev_filter
https://en.wikipedia.org/wiki/Butterworth_filter
https://en.wikipedia.org/wiki/Bessel_filter

@Kitty-Dyson also it is possible to do with an fft as gsagostinho does in the link whale-av shared. This is equivalent to convolution with an FIR windowed sinc filter I think (but I'm unsure if the timing would be the same with overlapping windows and all)

edit: here's more info on how to use the overlap-add fft to implement windowed sinc convolution if you're interested:
https://ccrma.stanford.edu/~jos/sasp/Overlap_Add_OLA_STFT_Processing.html
https://www.dspguide.com/ch18/2.htm

@Kitty-Dyson I forgot that there's actually an example in the Pd help that does this - in example 3.audio.examples/I03.resynthesis.pd you can set all of the points in the array $0-gain below the frequency you want to cutoff to 1

another edit: if you're already working with spectral data as your original post suggests (that uses the overlap-add fft) you can simply set the high frequency bins to 0 if you're using a large enough fft (and using appropriate windowing of course)

@seb-harmonik.ar sorry for the very late response, ive been trying all these methods, and the fft example is probably best for the task. is there a way i can do this controlled with a single number cutoff point instead of a whole table? maybe one for the lower and one for the upper limit, making a square shape?

PS, I also am not a master with fft, and looking at this is slightly confusing, ive been trying to figure it out for days.

@Kitty-Dyson here's a simple example with 2 numbers to create a bandpass filter (lets through frequencies between 2 frequencies in hz).

You'll need zexy for the signal-rate relationals (>~, <~, &&~) and also an object from my own library shadylib (pinb~) because I couldn't figure out another way to get the sample number in the block without making an external (but if you can figure that out, that's all that pinb~ does)

fft_filter_example.pd