Nuke Plugin for reading VRAY Cameradata from EXR released!

[ultrasonic]

Hey Guys,

i have just compiled a nuke plugin which u can use to extract all camera information from a vray exr file. I think it'll only work with vray4maya exrs...i think max doesn't write cameradata.

Find it here.
http://www.fullblownimages.com/scripts.html



Just plug the node named "PostlightCamera" into your read node and watch what happens

It's only compiled for Nuke 5.2 win_x64 yet...let me know if u need it for other versions.
EDIT: I just got the first crashreport. If u experience crashes too,please send me your exr if possible!

cheers
Oli

[ultrasonic]

I updated the plugin...there was a checkup missing which caused crashes when the attributes were not present!

please download the archive again!
http://fullblownimages.servebeer.com/scripts.html

[Inlakechh]

Hey Oliver wie gehts ? ,

this seems to be a fantastic plugin.
Is there an option to make Mental Ray render out Camera Data in Exr`s aswell and get it into Nuke ?

cheers,
Marco

[ultrasonic]

@marco: i'm fine....hope to see u guys again soon! greetz 2 everyone!

It is possible to write all this information from mentalray as well,but this would require coding an extra output shader. i have done this once some years ago,but since mental images changed the whole framebuffer api it is going to be hell to port this over to the latest mr-versions.

Also i think using MR these days is a waste of time. If you want fast good looking visualization or commercials use VRAY. If u wanna do feature film use any renderman render. I think MRs time is over. *lettheflamesbegin*

cheerio
Oli

[beaker]

Any chance you could just include the source code so people can compile it for osx and linux? I am mainly asking for it because people post precompiled plugins and then the author will get busy on projects and won't recompile it for new versions of the software. So people and companies don't want to put a tool into their pipeline that there is no guarantee that they could use it in future versions of the software. I have run into this countless times over the years with Maya, Shake and before that Power Animator.

[ultrasonic]

Hey beaker,

that's a very good point...haven't thought about this yet, but i will see what i can do.

I am also currently working on a reimplementation of the technique using nukes built in metadata streams. I think one could also realize this tool as a script instead of a plugin which would not create that much hassle as a plugin would.

stay tuned for updates
Oli

[beaker]

yes, I started exploring something like that too. Till I saw your plugin I didn't realize the data was right there in the metadata in vray renders. My only worry is if you retain that metadata when you render .vrimg files and then use vrimgtoexr.

Your plugin kind of flips out when you have a live connection between it and a camera. The camera jumps all over the place in the viewer. Also Nuke is weird about it because it doesn't see animation so when you try to bake out an expression between the plugin and a camera it only bakes the frame your on for the entire sequence. So you have to first bake the keyframes in the plugin and then copy them to a camera.

[DrJones]

Great Plug, I have been waiting for some thing like this to be released for a while.

but since most of our work stations are Mac an OSX version would be great.

-DrJ0nes

[ultrasonic]

yes, I started exploring something like that too. Till I saw your plugin I didn't realize the data was right there in the metadata in vray renders. My only worry is if you retain that metadata when you render .vrimg files and then use vrimgtoexr.

Why do you render .vrimg?

Your plugin kind of flips out when you have a live connection between it and a camera. The camera jumps all over the place in the viewer. Also Nuke is weird about it because it doesn't see animation so when you try to bake out an expression between the plugin and a camera it only bakes the frame your on for the entire sequence. So you have to first bake the keyframes in the plugin and then copy them to a camera.

can u please send me a .nk file that shows the problem....

an OSX version would be great

Stay tuned ... i am working on a python implementation at the moment which will make this platform independent

[alkali]

Why do you render .vrimg?

Because when you convert .vrimg files to EXR using the command line tool, you get a ZIPS-compressed scanline-based file with a correct BBox (i.e. Nuke-friendly), as opposed to the tile-based EXR that you get from a bucket-based renderer (not sure if VRay's framebuffer saves BBox info).

Think faster image IO in your comp.

[beaker]

vrimg renders faster and the vrimgtoexr converter will give you datawindow(DOD) and convert the file to a scanline in one swoop. I already have to rerender the exr out of the box from vray because they are tile exrs.

[ultrasonic]

Because when you convert .vrimg files to EXR using the command line tool, you get a ZIPS-compressed scanline-based file with a correct BBox (i.e. Nuke-friendly), as opposed to the tile-based EXR that you get from a bucket-based renderer (not sure if VRay's framebuffer saves BBox info).

Think faster image IO in your comp.

We have stopped using this workflow because it's slow as hell to convert .vrimgs! We convert to scanline-exrs with dod over here too, but by using a postrender script that launches a nuke script for the entire sequence.

@beaker: have u found out if vrimg2exr retains the metadata?!? And i highly doubt that vrimgs render faster...are u sure?!?!?

cheers
Oli

[alkali]

I can't speak to the speed of the vrimg2exr workflow (you could always spawn multiple instances...), but 9 times out of 10 it's a heck of a lot better than tying up a Nuke render license every time you render a 3D sequence.

I would imagine metadata stays intact (they can put whatever they want in the .vrimg files), but I haven't tested this (last VRay build I worked with was 1.5RC4). And speaking to the speed of the renders, I would imagine it's probably a tad faster to render .vrimg files, since they're basically a buffered brain-dump of data, as opposed to any kind of formatted, compressed file. Not sure though.

I've used the post-render Nuke method too (in non-VRay pipelines), but I now prefer using RVIO, since it's faster, more reliable, and is actually built for that sort of thing.

[ultrasonic]

I can't speak to the speed of the vrimg2exr workflow (you could always spawn multiple instances...), but 9 times out of 10 it's a heck of a lot better than tying up a Nuke render license every time you render a 3D sequence.

Yeah maybe, but also the filesize of the vrimg tends to get huge...in my cases more than 600MB per frame which made it no option for me.

I've used the post-render Nuke method too (in non-VRay pipelines), but I now prefer using RVIO, since it's faster, more reliable, and is actually built for that sort of thing.

I don't know RVIO...can u tell a little bit more about it? Does it handle or even convert .vrimg?

[pingking]

i just want to throw one thing into the discussion about exrs. the latest 3dsmax with connection extension can write out scanline based exrs with bounding boxes, so i believe you will see it in the next maya release too.

i know this isn't very helpfull for you but i just wanted to say it