Vector rescanning using soundcard :)

i.m.klif · Post by **i.m.klif** » Thu May 31, 2012 3:58 pm

hi

i finally found time (due to broken arm) to turn on my new oscilloscope and try some vector rescanning techniques using soundcard. i used maxmsp with jitter for video input. so, equipment i used was: powerbook, max6, souncard (m-audio 410), tektronix 922 scope.

to my surprise i got it going in under 2-3 hours! it seems to work as expected.

i recorded quick and dirty demo using old panasonic camera. next week i'll reshoot using dslr.

[video][/video]

thank you dave and lars, if it wasnt for your advice in some other threads, i'd probably never try this

klif

i.m.klif · Post by **i.m.klif** » Thu May 31, 2012 3:59 pm

i forgot how to get the video to show inside the post. anyone?

lizlarsen · Post by **lizlarsen** » Thu May 31, 2012 4:01 pm

Awesome!!

I wouldn't have thought this was possible. Can you comment more about the resolution and frequency of the output from the sound card used to generate the effect?

i.m.klif · Post by **i.m.klif** » Thu May 31, 2012 4:10 pm

well it does have to do with output frequency. at the moment i have 96k card. also, scopes are both second hand and i use them for the first tme. they dont seem to be in perfect shape.

i get stable image using divisions of sample rate. frequencies in between are more unstable but look excellent. at 1,20 i started playing with x+y frequencies.

if you can suggest some sort of testing method i'd be glad to try it.

but i do have a feeling that the number of scanlines directly relates to output frequency

Matos · Post by **Matos** » Thu May 31, 2012 4:21 pm

Looks great. Midi control over the variables is like a dream come true. I might have to finally start learning max and jitter. Oh god, the thought scares me!

smrl · Post by **smrl** » Thu May 31, 2012 6:49 pm

sort of related:

these both came out of the demoscene... I know that there's a wav file around somewhere for 'youscope', oscillofun is particularly interesting because the x-y information is multiplexed with the music...

daverj · Post by **daverj** » Thu May 31, 2012 11:01 pm

Very nice!

At the low end it looks like you were down to about 10 scan lines. It was hard to tell what the high end was, but looked like well over 100 scan lines. That was cool.

mono-poly · Post by **mono-poly** » Thu May 31, 2012 11:21 pm

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 6:32 am

the resolution seems to be 100x100.... but i might be very wrong here.

it's kind of confusing. i seem to send 24 hz on y and 2400 on x to get stable image with max resolution. higher settings produce some very interesting fx, but the resolution doesn't get any higher. i don't understand how this low frequency can produce rather smooth looking video.

with 240hz on x i get 10 scan lines.

at 24 hz on y i get a flickering but optimal image, at lower hz i seem to get better resolution but at visibly slow scan rate

what is new for me here is that in vector land lo-res doesn't look lo-res at all!

another thing that confuses me is the image drift. when i change the scale of image i.e. when i change the amplification of the output, the image slowly drifts to the position. other parameter make instantenous changes. i guess it has to do with the scopes... like some capacitor discharging... but when i put the scope to AC setting, it reacts quickly, but the top of the image is squashed, like it gets exponentially more squashed from the bottom to the top.

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 8:10 am

now, how do i perform image rotation??

lizlarsen · Post by **lizlarsen** » Fri Jun 01, 2012 9:29 am

[video][/video]
http://thecreatorsproject.com/blog/expe ... filmmaking

Pretty cool!! But still not as pretty as an actual vector monitor in my opinion.

All the same, I bet there's lots of fun to be had with converting Kinect into a grayscale map and using it as a video source for an analogue video synthesizer.[/list]

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 9:53 am

i was on their workshop i belgrade couple of months ago. it was really nice.

the process is pretty involved. it is definitely not point and shoot.

the main thing about is the ability to change angle + other parameters of already recorded footage. vector look is just one of the options. and material really shines in full HD. so much detail... very nice

in my opinion, it is only vaguely connected to vector displays.... even visually

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 9:55 am

but, lars, to return to the original topic.
how do i rotate rescanned image? is it possible? i'd like to move the square in 3d space. i can move, scale, but i can't rotate.

lizlarsen · Post by **lizlarsen** » Fri Jun 01, 2012 10:00 am

Sorry, I had meant to post that video in a new thread but it ended up here! I suppose that works okay, though.

As for the image appearing smooth... if you are running Y at 24Hz that is still 24 frames per second, and also the vector monitor usually has more persistence than a normal television (the beam sticks to the display longer before fading.)

For rotation, you're going to need a way to mix the ramp sources together. I'd have to think about it a bit, to come up with an equation for the math part. If you crossfade between the X and Y ramps (swapping them, essentially) you will get a 90 degree rotation. If you do this sequentially, and mix in the inverted versions of the X & Y ramps, you can get all 360 degrees. There's probably a simple equation that will make sense for your code.

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 10:19 am

uuuuu! it started working

skew is there as well

i find this approach very good for learning and understanding how these things work. very fast.

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 2:08 pm

ok, rotation + skew works great. i can get some nice 3d style rotations here.

what i don't understand is the difference between AC and DC setting on the scope.

with dc i get slow response when resizing the whole image (changing the volume of the audio out), but in rotation i can specify anchor point.

with ac i get fast response when resizing, but there's distortion in scale (like log or exponential function), and the image is always central when rotating.

both methods are usable and have their strenghs, but i'd like to know why this happens. does it have to do with ac/dc coupled audio interface?? at the moment i use ac coupled interface.

lizlarsen · Post by **lizlarsen** » Fri Jun 01, 2012 2:11 pm

For DC accuracy you'll definitely want an audio interface with DC-coupled outputs like a MOTU. I'd definitely expect some weirdness there, otherwise.

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 2:33 pm

can you elaborate a bit on this ac/dc stuff??

lizlarsen · Post by **lizlarsen** » Fri Jun 01, 2012 2:35 pm

Well, a DC signal is going to maintain a consistent bias -- which makes things like positioning the signal on the display possible and accurate. With an AC signal, there is no offset/reference, so the image will tend to drift to a "center position" assuming a bipolar input signal.

daverj · Post by **daverj** » Fri Jun 01, 2012 2:38 pm

For rotation you want to use a pair of "4-quadrant" multipliers. (aka bipolar VCAs) for each ramp. You mix each pair and control them with a sine and cosine quadrature oscillator.

The idea is that you are basically crossfading between positive and negative versions of each the X and Y signals (the ramps or counters that create the raster).

I believe the following is correct:

Matos · Post by **Matos** » Fri Jun 01, 2012 2:44 pm

Brilliant guys! I'm learning a lot from this thread. Cant wait to get home and explore. Max seems ideal for these kind of transformations. Curious. To see how far I can get with the modules I have.

daverj · Post by **daverj** » Fri Jun 01, 2012 2:44 pm

Audio typically has no DC signal. DC is the ultimate low frequency (0Hz). Most audio interfaces are AC coupled. If yours can do DC coupled output, then you want to use that. Otherwise the AC coupling in the interface will slowly drift the signal towards the center, though depending on it's cutoff frequency it might be slower or faster than the AC coupling of the scope. It sounds like the scope's AC coupling is faster.

With AC coupling you lose the static DC voltage and also roll off very low frequency signals, which can cause non-linear distortion.

daverj · Post by **daverj** » Fri Jun 01, 2012 2:52 pm

Everything MAX can do can also be done with analog circuitry. It's just a matter of having the right modules. MAX is nice because it has an unlimited number of each module, and has all sorts of "modules" that you might not otherwise have bought or built.

For example, a mixer gives addition. Mixing a negative signal on one side is subtraction. A bipolar mixer can add or subtract. A VCA is an unsigned multiply. A "4 quadrant" multiplier (bipolar VCA) is a signed multiplication.

lizlarsen · Post by **lizlarsen** » Fri Jun 01, 2012 2:55 pm

Great diagram, Dave!

Some video-rate 4-quadrant multipliers have been on my prototyping list for a while. need to play around with that soon.

i.m.klif · Post by **i.m.klif** » Fri Jun 01, 2012 5:41 pm

wow, thanxs for all the explanations.

dave, i solved rotation in the exact same manner, but it was hit and miss affair until i got it right. now i have skew as well so i do get almost convincing 3D rotation. it is little bit rough and bouncy right now but it looks fine.

i made a short movie to show how it goes. i'm uploading now

another nice convinience is that the movies are read in jitter and directly scanned to audio. the ramps for the raster are used to scan the movie so the sync is perfect.

as for AC DC, thank you again. now it is a lot clearer. i don't have DC coupled card at hand right now, so i'll have to stick to AC.

i'd like to experiment a little bit more and then clear up the patch and start programming the user interface, so that i can perform with it.

p.s. how do i get the vimeo video to show in post??

MOD WIGGLER

MOD WIGGLER

Vector rescanning using soundcard :)

Vector rescanning using soundcard :)

RGB+D Toolkit