Textures and rendering

Hi,

I’m starting this topic in order to get some advices that will help me to earn some time.

Actually i’m a one week VVVV user and i’ve to achieve a project for the 05/04/07. This project consist of a non-realtime 50’ duration video that’ll be displayed during a concert. This video deals with textures based on photographies exclusively because it’s a showcase of a photograph work. Not easy at all in such short period of time.

Actually the output of the photograph are images scaled at 1600x1600 (i asked her power of two textures after reading the faq).

I’ve made some patches, with my little knowledge, based on texture transformation or quad coordinates transformations reacting with audio analysis. And smoe shader stuff too still over textures.

So i’ve two questions:

  • If you have some advices, ideas, examples dealing with textures as a starting block (glitch, transition, morphing?, shaders or whatever) it will help me to earn some time (i’ve to test the kalle module textures transitions)

  • How to render 1280*1024 video output smooth and with best quality possible? Actually i’m using ampop-nrt writer to write frame by frame .jpg files making a video about that with quicktime but the result look jerked. You can see an example there http://perso.orange.fr/louloup/ESSAIS_VIDEOS/02.mov

Maybe i’ve to reduced the size of the texture or compressed them with the file texture node. Or should i save each frame in another format with the writer node ? I need the better quality output as the video will be projected by a beamer.

I can post patch or some more of my gear if it can help all of you understand that better. My english isn’t that good too. But i’m in emergency don’t sleep at all theses days and a little bit ot time earning would be really great.

Thx a lot by advance.
And thx to all the vvvv community it’s really a great stuff.
VVVVery addictive.

How to render 1280*1024 video output smooth and with best quality possible? Actually i’m using ampop-nrt writer to write frame by frame .jpg files making a video about that with quicktime but the result look jerked.

@jannis had an issue with missing antialiasing and patched a pretty good workaround. see that thread

i don’t know what exactly you mean with jerky?

btw: what software do you use for quicktime-conversion?
i often had problems with that.

Actually the output of the photograph are images scaled at 1600x1600 (i asked her power of two textures after reading the faq).

for best image quality i wouldn’t chage the original image size, unless the performance is to worse. if you still like to scale the images, it should be a square power of two size to get a noticable performance improvement.

  • If you have some advices, ideas, examples dealing with textures as a starting block (glitch, transition, morphing?, shaders or whatever) it will help me to earn some time (i’ve to test the kalle module textures transitions)

yes, try kalles module and the Blend.fx shader from the shader gallery.

  • How to render 1280*1024 video output smooth and with best quality possible? Actually i’m using ampop-nrt writer to write frame by frame .jpg files making a video about that with quicktime but the result look jerked. You can see an example there Orange

vvvv ships with the module Writer (EX9.Texture NRT), which is ampops nrt writer.
for non realtime rendering its important to use the Increment time mode of the mainloop. the MainLoop (VVVV) node is already in the nrt writer module, you just have to set your desired output framerate.
second important paramter is the size of the rendered images. its set either at the renderers backbuffer size OR at the DXTexture node inside the writer module.
png is recommended as fileformat, because of its lossless compresseion. for best results you should render the images in double resolution (2560x2048).
to render the frames into a video i can recommend VirtualDub. it loads the image sequence by drag&drop and has a good downscale filter to get the final resolution. dont forget to set the framerate to the value you used during the rendering in vvvv.
if you don’t need the depthbuffer for 3D objects, you can further improve the image quality by enabling the antialiasing of the renderer…

thx for the reply i’ll try that tonight thx a lot :)

yes actually my problem is that with these textures my pc is slow so it’s hard to set up parameters to have the correct value during the rendering process.

Maybe i could increase that hiding the rendering window during exporting but if i do that i won’t have any vision over the final rendering. Anyway i’ll try your suggestions of both and see what i come with.

btw: what software do you use for quicktime-conversion?
i often had problems with that.{POSTIT(title=>Title, date=>date)}textPOSTIT}

i use quicktime pro player for that actually but as i saw on my exemple it doesn’t look natural in the animation it look quite jerky (saccaded ???)

Kalle i’m trying to use ur texture transition modules but it seems nothing is displayed on the renderer window even after wiring the quad to it and right click on the modules nothing is draw on the renderer window. Am i doing smth wrong ? What a pity these modules seems very usefull for me.

Cheers,

i’m trying to use ur texture transition modules but it seems …

i’ll have a look at them today. promised.

thx a lot this community is very welcoming :)

which ones are you referring to?
did you check the helpfiles? are they working?
check backbuffersize pins inside the modules?

all of them i’m using the helpfile actually and have the problem with these files. i didn’t checked the backbuffersize i’ll try that this evening at home. and let you know :)

ok seems to work with the group node but the quality of the texture seems to be very low maybe i’m missunderstand smth sorry to be so newbish !!!

yes, yes, i wonder why thats missing.

i am going to download my modules and check, why…

OK I’ve export the images at the 2560x2048 px as png files and activing the antialias via the inspektor (switching the antialias to 4). Is that correct ?

I’ve tried to change the fps in mainloop of writer nrt but i don’t really understand what it change. The bigger the fps is the more images/second will be saved so that the animation sequence will go faster ???

I’m using virtual dub with the 2.1 high quality filter but i don’t really know which framerate would be good (in vvvv too).

The save process is very slow so that i can modify parameters in realtime (actually i’m hiding the render window) so should i try to scale down the image resolution or borrow a more powerfull computer.

Thx by advance !!!

i did not get why you want to use the nonrealtime renderer with all the virtualdub issues at all?

a realtime solution straight from vvvv would be much easier and a better quality?

hi oschatz.

That’s mainly because i won’t be in the event where the video will be projected. So i need to let a movie file that will be displayed during the concert. That’s the mainly reason if i could be at the event i’ll do real time vjing. So i need to export all the stuff i’m doignt as video files and then end up with a final .avi or .mov that will be projected during the show. Won’t be audio reactive or in synch with the music. But the aim of the project is to show the work of the photographer (that’s why i’m dealing with textures) and to do better transitions that could be achieved with power point.

hmm. we routinely just leave our pcs running while away - vvvv is very stable and not more likely to crash than a video player. especially when dealing with photography vvvv is a great option to display uncompressed high definition content in very high framerates - its a shame to render that down to a compressed avi…

I agree but i currently live in bordeaux(FR) and the event will occur in brussels(BE) so far away… Actually i don’t even know which gear (computer, beamer, whatever…) will be available during the event and have no way to know that (i’m not directly involved into the project). And alsomore, they won’t be anyone dealing with vvvv during the show. So it seems to be quite complex…i’m still wondering how to find a better way to do that. And my pc can’t handle manipulations in realtime btw that’s why i’m into nrt rendering.

My goal is to achieve the better result in a short period of time (not fully due to my work) so any suggestions that would help me in that direction would be great. Just have to make good transitions with textures and make it looks good.

If you see a better solution i’d please to know that. It’s true that vvvv is very easy to install and very stable but i don’t even know if it’d be possible to have an access to the computer that’ll handle the projection.

When i’ll be at home i’ll post my patchs maybe my computer can’t handle the processing due to bad patching (still learning vvvv since two weeks not full time).

again ppl,

I’m asking you some recommandations because i’m becoming crazy.
I’ve use the sven particles patch and modify it a bit to play with some text. Everyting is fine when i play with the patch in vvvv (my comp is at 100 % cpu i’ll try this afternoon on another comp) but i can’t have a correct rendering in video about it. Yesterday i past 2 hours exporting 1000 png files modify all the parameters in the scene with my midi CC so slowly i hurt my fingers and the result is a disaster.

I don’t know how to achieve that successfully and as i spent 12 hours working on that yesterday i think u can help me a lot (i’m still a 2 weeks user of vvvv) so i post my patch i’ve try to clean it a bit and to let some comment. Any modifications recommandations is VVVVERY welcome as i’m so tired and so sad to not result in good rendering. The deadline of the project is monday…and still have 0 min of 20 finalized.

particles_mask.rar (28.7 kB)

i’ve just made a test with fraps and everything works fine.

so happy !!!

hi still working on that project.

Sorry if that question has no deal directly with vvvv but it’s the best place i know there for asking this question and having an usefull answer to it.

I’ve end my work with vvvv exported at 1024x768@25fps directly grap from the backbuffer using fraps. The size of the files was very big so i used auto gk in order to compress the files. Now i’ve video files with resonable size (the avi was compressed using divx 5 codecs at quality 100%). I’m quite happy with the quality and the results.

Now come my problem. I’ve put some black space (~10 sec at the begining and the start of my movies) and after showed that to the ppl that involved into the project they told me it’d be necessary to cut a little bit that black space.

Here i come with virtualdub and directstreamcopy video mode. But as far as i’ve understood i can only cut the video to keyframes that wouldn’t me allow to cut precisely the video as they are ~10 keyframes per video. I’ tried then with virtualdubmod which allow me to cut the video whatever was the frame but then i’ve some gaps in the video.

So if someone have a tool or any recommandations in order to cut my avi files with a good precision (approximately frames precision) without any loose in the quality.

It’d be vvvvery great !!!