Video texturing

hi!
i have a few questions concerning a video texture on a quad…

  1. i have the quad in a 3d environment and a very bright white block behind the videotexture -> what happens is the video fades out (is just white) because its transparent?

  2. when i hook up an FFT (spread count is 1) to the scale nodes of the quad, the video changes size but due to a “bug” (maybe no bug, just me being to stupid to understand why its like that LOL) there are still 8 spread counts forced in the FFTs output… is it possible to use an FFT on a videotextures with only 1 spread?

  3. i made a cube out of the videotextured quad (using spreads) and would like to use this “cube” as a real cube (BOX) - making it possible to transform the “cube” as a usual BOX

i hope someone san help out,
thx

mike

hi!

  1. depends on what you want: usage of Blend (EX9) or even the advanced one give you more possibilities than you perhaps want to have…

  2. don’t know exactly what you mean: try an RMS (DShow9) with mono on your audio in…

  3. i remember that i once posted a module called box (DX9) somewhere in the old vvvorums. perhaps it needs a bit rework because of version changes. think it was made with beta3 or 4. i’ll have a look into it soon.

regards
@kalle

when i hook up an FFT (spread count is 1) to the scale nodes of the quad,
the video changes size but due to a “bug” (maybe no bug, just me being to
stupid to understand why its like that LOL) there are still 8 spread counts
forced in the FFTs output… is it possible to use an FFT on a videotextures
with only 1 spread?

the fft changes audio samples from the time domain into the frequency domain by using a specific block size. the fft takes a specific block of audio samples and transforms them into frequency information. from the mathematics in the fft this is a 1:1 process. the fft is quite easy to calculate, but may be not be what you want. also the fft internally requires powers of two, so the blocksize gets rounded to the nearest power of 2 (like 16, 32, 64, 128 etc.)

having the block size to small doesnt really makes much sense: if you set the block size for example to 256 and you will have 44100 samples going in each seconds, you will process 44100/256 = 172 blocks per second. which means if you render 60 frames per second, vvvv will throw 2/3 of the data away. so you could use a 512 block size instead. or lower the incoming sample rate to save some cpu.

so a fft with a very small block size like 8 doesnt really make much sense, as most of the data would be thrown away anyway. indeed the fft has a buglike little feature allowing not less than 8 as a block size.

if you just want to draw less quads, just pick out some of the slices. Use GetSlice or delve into the many possibilites of Resample or Bounds (Spectral).

or as kalle suggested, just use RMS to get the volume of the incoming audio instead of the FFT. RMS is nice as it gives you the left and right channel as separate slices if you use a stereo input.

hi mike,

the module i wrote about is up now on kalle - Modules\DX9
and is named Cube (DX9 Quads).v4p
is ported to beta8.1 now…

i remember that i once posted a module called box (DX9) somewhere in the old vvvorums. perhaps it needs a bit rework because of version changes. think it was made with beta3 or 4. i’ll have a look into it soon.