Rendering a big suite of cameras

[NNois]

Hello, I'm front of a problem.
I have a long scene with 200+ cameras following coming from a camera tracking software.
I need to render this... my first though was to constrain a master camera to the others (and FOV), I do it once on another project but it's a pain because all pose contraints have the same name, so changing a frame range on a camera is very difficult not to mention that the FOV can't have mutiple expressions.

So, it seems that I have multiples choices:
1. Use the motion mixer to record all this cameras (this seems to be a very long task, is it the good path to follow ?)
2. Copy/Past manualy all motions ? ( a scipt on ed haris is supposed to do that but doesn't work on SI2010 )
3. I'm not a programmer but if I have no choice, is it complicated to create a thing like a scripted operator who takes a frame set, sort it and contrain the master camera to the others ?

Humm, another thing, i'm used to apply expressions on animatable channels but is there a way to apply an expression on the top of the global transform for example ?

does someone had already managed a problem like that in the past ? an advice ?
Thanks

[owei]

Hi..!

Well, at first sight I would say, that a amount of 200+ cameras is not comfortable to handle, this or that way. The first thing I would like to know, how import is working. You import a scene with 200+ cameras, is that right? What is the naming convention for them? If the are named properly, it might be the easiest way to write a short script that builds a pass for every camera. That way, you have the biggest chance to manage changes "relatively" easy. But I never tried to build a scene with 200+ passes though ;) just one thought...

cheers,
oliver

[NNois]

well i haven't imported a scene with 200 camera, i've build it, so naming is not a problem (or not very long to remedy) for now it's indexed "CAM_P001, CAM_P002 etc..."
actually I've found that changing passes cameras/and frame range before submiting to rendering is the fastest way, but i've to do this 200 times ! (for now i'm testing with 10 cams but in a short time i'll need to render all this shots).
Passes I think is not a good way because passes takes too much memory i think... and it's not very manageable

[Rork]

You tracked a scene, imported it, added 200!!! camera's and are getting yourself into problems.........
Yes, read that again.... ;)

You really have to tell me why you need -in addition to the camera created for the track- 200 additional camera's. Isn't tracking suppose to give you a camera solution for that particular shot?

I am really curious how your scene is set up, and what's the need for the extra camera's. It doesn't make any sense for passes either....

Can you explain more how you started and what you want to do??

rob

[NNois]

Thank you for trying to help me.

well I think I expressed myself badly, sorry for that.
So, I have a filmed sequences with many cuts (green screen) aproximately 4 minutes, i'm tracking them separately, so the tracking software bring me all this tracked cameras.
In parallel, I've build an animated scene in XSI, then imported my cameras to match the footage.
I've started this way because the 3D part isn't changing dramaticaly over the 4min.
In parallel, I've composited each cuts with my rendered passes (beauty, reflection, etc....)

For now i'm working on the begining of the video (on the first 15 cameras) testing compositing etc... and hopefully I'll be able to automate the rest of the 3D rendering ! it's why i'm posting here...

But tell me if you think that I have not used the correct method !

[NNois]

Thank you for trying to help me.

well I think I expressed myself badly, sorry for that.
So, I have a filmed sequences with many cuts (green screen) aproximately 4 minutes, i'm tracking them separately, so the tracking software bring me all this tracked cameras.
In parallel, I've build an animated scene in XSI, then imported my cameras to match the footage.
I've started this way because the 3D part isn't changing dramaticaly over the 4min.
In parallel, I've composited each cuts with my rendered passes (beauty, reflection, etc....)

For now i'm working on the begining of the video (on the first 15 cameras) testing compositing etc... and hopefully I'll be able to automate the rest of the 3D rendering ! it's why i'm posting here...

But tell me if you think that I have not used the correct method !

Well, I think I forgot to clarify one important element. I replaced independently all my tracked cameras in the scene, because the tracking software does not place quite the right place. So I can not just copy paste keyframes to a master camera (sans parler de la target et du roll), I need to constraint them ...
what i'm looking is to dynamicaly contraint my master camera to the others (with the FOV), but like i've said before multi XSI pose constraint is a big confuses and i don't know the way to multi constraint one independant parameter like the fov

[NNois]

I'm posting it right here but, admins, did i need to post it on the pragramming part of the forum ?

Well, I think i have a syntax problem in this scripted operator:

I have connection to cameras FOV (fov0, fov1 etc...) and connection for a custom parameter (integer animated over time to switch camera)
it's a start ;-)

Code: Select all

logmessage ('--Debut--');
var activefov = "fov"+CamNumber;
logmessage (activefov);  // output "fov1" it's what i'm expecting
Out.Value = activefov.Value // doesn't read the fov1.Value , it's a syntax problem no ?

[Hirazi Blue]

I'm posting it right here but, admins, did i need to post it on the pragramming part of the forum ?

I don't think it NEEDS to be moved,
but if you want me to, I will move this thread to "programming"!
;)

[Rork]

I still don't get it.....

4 minutes of footage = 240 sec.
'many cuts'
200 camera's = a new camera every 1.2 sec.
result: epileptic seizures in your audience ;)

I still don't understand the need to copy the tracked camera data to something else. If you have measured the set, you can easily scale up the tracked data and 3D so they match.
It sitll feels to me you're looking at this the wrong way.

rob

[NNois]

hello ,
well, in fact i ask here because i don't want to do it the wrong way, so i've started this big work with the 3D (now it's well advanced) plus tracked a part of the cuts, worked on the compositing and when i've posted here i've estimated badly the number of cuts because i'm at 40 seconds...
so you're right and excuse me but it's more precisely (after a real count of cuts) 132 cuts, it's a music video with more or less epileptic cuts ;-)

I still don't understand the need to copy the tracked camera data to something else. If you have measured the set, you can easily scale up the tracked data and 3D so they match.
It sitll feels to me you're looking at this the wrong way.

rob: I need to render these 6000 frames but i don't want to prepare 132 scenes just to submit a different camera to render or having a scene with 132*5passes ! so i would simply like to assemble these cameras to render.
Olso, it's probably my fault if you don't get it (i'm a poor badly english speaker), like i've said before, it's a music video with live action footages, so i have 132 cuts along this video where i need to composite it with matched 3d footages. If you think i'm in the wrong way please guide me a little on how you will do that kind of work. It's for me a premiere on working on so many camera matched cuts.

Thanks all !