Access violation on texture queueing beta23

hi joreg.
I m on another project, and i m having this warning arriving in TTY when i insert in my project the resample patch attached here, and connect it :

02:11:43 ERR : Error caught in the act: TAddFlow : Access violation at address 0067E253 in module 'vvvv.exe'. Read of address 00000360


0:08:22  *  : [validatevalue: error occured in ](validatevalue: error occured in ): Access violation at address 0043B0E6 in module 'vvvv.exe'. Read of address 00000001

This project is very different and in a way far less complicated than last time.

I have ONE Ip camera goig to html renderer, that i m aquiring with VHSCrap to feed the videoin ( ipcamera driver doesnt show up any image when called directly from enum pins).

From this camera i m sending its texture with S(node) and filtering it in differents shaders before mixing those differents filter in Blend’s svvvitcher’s shader.

When i insert and connect in my patch RESAMPLE_BANG.v4p, i began to have troubles in the TTY.

This patch enables me to capture in live what is coming from the camera, and to reinsert it in place of the original signal. a kind of video sampler.

At the beginning i was doing this on frame base methode, taking the mainloop. Here is my post :

I was thinking that maybe frame delay was a problem, as I had blank frames on this technique so i changed all the patch and i went on lfo technique, wich doesn t show up troubles of capture, but is now showing this access violation.

I presume there is somewhere a conflict with image queueing ? is there any way to prevent from it ?

I have from 5 to 8 shaders on 16 possible working continuously in direct in 1024x728 with a rendering in 1024x728.
Graphic card is a ATI 5770 with 1Go DDRAM and seems ok.
Gui is very simple and in GDI to let some memory left. Patch is remoted in artnet.
I have some latency with beta 23 when opening subpatches and calling vvvv menu

If i make a test with only the camera and this subpatch, i dont have the access violation.

sampling texture in live (26.1 kB)

ai karistouf,

nothing obviously bugging in your patch.
probably unrelated:
do you know MonoFlop (Animation)? it should help you get rid of both your lfo->framedelay->backtolfo constructs in this patch and make everything more clean…

hi joreg, thanks for your reply

aie aie aie…

we are actually on creation ( new work, new company, new license to come) and last experience of access violation makes me definitively says: “i do n t want that in a patch” ;-)

my feeling is that i m reaching the limit of my graphic card and in this way strange things may happen ? and that some kinds of tetris blocks are moving in memory and sometimes colliding when inserting a new subpatch in the actual set. ( very personnal impression with vvvv)

if nothing bugging in this subpatch, can you tell me what type of error the tty is talking about ?

it will help me to find out why it happens on all the patch…

by the way, i never understood the interest of monoflop, and until now this is quite a big question … if anybody can shows me the practical interresy, to open my eyes…

about the monoflop:

in your patch you want to:

  • OnBang: start recording
  • AfterSomeGivenTime: stop recording
    i see two similar such parts in your patch. right?

you are using an lfo, a framedelay a > and a < node where you could use a single monoflop. the monoflop switches its output to 1 OnBang and resets it to 0 AfterSomeGivenTime via the Time. also see its helppatch.

about the errors: when it says TAddFlow…it is a gui fukc. shouldn’t occur during runtime if you don’t touch the gui. the other one is not very descriptive and as long as you don’t find a simple way to reproduce it we won’t be able to tackle it.

hi joreg, many thanks for your time ( you are also working sunday …) .

wow , never noticed the time pin!
but i m not understanding how i can stop lfo precisely to the end value of 1.0000 ( thats why all this mess about > 0.96 )

bugs: i put my petzl on my head and i swim deeply to find out what s happening.

about gui, same trouble of access violation founded with “as string”, but in default format.

i m using also +(string) and framedelay for incoming data. i began to suspect seriously +(string)

but my gui is very simple: 3 lines of states in string, and 3 buttons on/off, all of it in GDI. in this project whitecat is remoting vvvv in artnet from a second computer, so all ressources are only on video work.

lfo_once_monoflop.v4p (3.3 kB)

hey karistouf!


you only have to change the value at the switch from 0.0000 to 1.0000

ok got it. but really it would be nice to have a lfoonce node, than doing all that stuff of logic ( poor poor poor karistouf !) ;-)

hi joreg.
I have more news, after a nightmarous rehearsal yesterday
I think the error is reproductible with VHSCRAP on a HTML renderer connected to an IP cam.

the access violation is ALWAYS when i m aquiring texture, and some time there is a blanck image coming out with the vhscrap logo. its at this time ( on recording) that the access violation happens.
nothing else in the patch seems to create this error. I have putted out the GUI and i have now only shaders work, trautner and contour working.
i don t see any buggy subpatch.

my configuration is in main loop 30fps front and back. 30fps on vhscrap driver. i have screens and video projectors in 60hz.