You are viewing an archive of the old fxhome.com forums. The community has since moved to hitfilm.com.

Looking to improve my CG Composites

Posted: Thu, 10th Sep 2009, 9:33pm

Post 1 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Hello All,
Hav'nt been around for a little while but im hoping someone can shed some light on a particular workflow im looking into.

I am heavy into compositing CG elements into live footage with the aid or matchmoving/camera tracking software.

Until now, i have rendered my CG elements as a single entity image sequence, .Mov or .AVI but obviously that's not very flexable.

A while ago, my CG software got an upgrade to include Multipass rendering (for anyone who does'nt know what that is - It means the ability to render seperate 'Passes' for lets say, ambient light, shadows, reflections, Ambient occlusion, specular light etc and to layer them seperatly on the timeline which allows you to adjust each layers intensity individually to get the look your after)

My problem is this:

I want to use this 'multipass' ability in Visionlab, so, for those who know, or maybe those who can point me in the right direction, which passes will i need ? which order do they need to be layered in ? what blend mode should each pass be set to ? do i need alpha channels ?? etc

Tutorials on the net seem to be so thin on the ground that i need an expert to guide me as i cant afford an expensive course.

sad

Can anyone help ????

I'd be so greatful

Paul
Posted: Fri, 11th Sep 2009, 7:56am

Post 2 of 23

Simon K Jones

Force: 27955 | Joined: 1st Jan 2002 | Posts: 11683

VisionLab User VideoWrap User PhotoKey 5 Pro User MuzzlePlug User PowerPlug User PhotoKey 3 Plug-in User FXhome Movie Maker FXpreset Maker Windows User

FXhome Team Member

Multipass 3D rendering isn't something I've explored. Would you be able to render out a sample set of passes and upload them somewhere so I can experiment a bit?

Doesn't have to be a video file if that's going to be too big in terms of filesize, just a single frame should be sufficient for testing.
Posted: Fri, 11th Sep 2009, 12:13pm

Post 3 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Tarn,
Thats a fine offer you have there, might take me a week to get that sorted as we are in the final stages of a conservatory build at the moment.

Certainly will be good for you to have a go at it yourself too as the experiment might help highlight area's that could be addressed in the 'Lab' lineup to help support this compositing method in the future. By adding a shadow pass (for example) as a seperate layer gives you the flexability of directly influencing it's intensity in your composite as well as allowing you to add grading to this layer like bluring and colour correction to get a better match to the background footage etc, this then wont affect any other attribute of the image except the shadows. If i rendered a complete sequence/video in 'one' with all passes combined, i could find at composition stage that the shadows were way too black for the background giving me no choice but to tweak in my 3D app and re-render over and over again until it looks right.

I have tried experiments myself in Vlab with this multi pass stuff but my success has been limited probibly due to a shortage of understanding and maybe features which Vlab might need.

Still hoping someone who knows this stuff might be able to offer me and Tarn some pointers here so dont be shy. eek

Thanks
Paul
Posted: Fri, 11th Sep 2009, 1:10pm

Post 4 of 23

Simon K Jones

Force: 27955 | Joined: 1st Jan 2002 | Posts: 11683

VisionLab User VideoWrap User PhotoKey 5 Pro User MuzzlePlug User PowerPlug User PhotoKey 3 Plug-in User FXhome Movie Maker FXpreset Maker Windows User

FXhome Team Member

Yeah, coincidentally I was reading about this stuff just yesterday. It'll be interesting to see how VisionLab can handle the technique and, as you say, to identify any holes where it has trouble so that we can fix them in the future.
Posted: Fri, 11th Sep 2009, 7:41pm

Post 5 of 23

spydurhank

Force: 1956 | Joined: 24th Jun 2008 | Posts: 1357

VisionLab User VideoWrap User FXpreset Maker Windows User FXhome Movie Maker

Gold Member

Damn!
That sounds like a great idea.
To be able to control all of those elements individually is pretty awesome and it should help even more with compositing cgi with live video.

I'm not trying to oversimplify this because I know how complex it can be when working with cgi but you could kind of think or take it like this... it makes sense to me anyway but... Working with layers in photoshop or gimp and actually Vision lab as well.

You have your live video, bring in your individual render passes shadows, lights, reflections and everything else and either grade them individually or all together with a single grade object.

What matchmoving software do you use by the way? I've tried a few of them... some work great and others.... not so much.
Posted: Fri, 11th Sep 2009, 8:16pm

Post 6 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Tarn,
There are some issues from what i can tell, hopefully you may notice these issues also. As spydurhank said, photoshop, gimp and also after effects are prime examples of getting the exact look your after by manipulating layers individually, it can save a huge amount of time in post production especially when CG elements are in the works.

This could be an interesting insight into fxhome software workflows, having the ability of adding lightning, muzzle flashes and light sabers etc is bloody great to say the least but CG intergration into live footage is also important and is now

WOOOOooo HOOoooooo

spydurhank,
I use Syntheyes which is not free like Voodoo is but is way more powerfull, and, for 'pay for' matchmoving software it's at almost the cheapest end of the price scale....It's not an easy thing to learn but very rewarding to get a good result.

Regards
Paul

P.S We're still hoping for an expert in the multi-pass field to put in some feedback or i'll have to cut & paste info from other forums for reference....Please please please, throw us a frickin bone *LOL*
Posted: Fri, 11th Sep 2009, 8:42pm

Post 7 of 23

spydurhank

Force: 1956 | Joined: 24th Jun 2008 | Posts: 1357

VisionLab User VideoWrap User FXpreset Maker Windows User FXhome Movie Maker

Gold Member

Right on.
I downloaded the trial version of syntheyes a few days ago and it is one of the better ones that I've tried.

Voodoo actually works pretty good as well but I keep having the same issues with all or most of the matchmoving software that I've tested. I can't seem to open an image in the software' preview window for a reference, I know that they only open certain formats or image sequences but I dunno, they just don't... I'm thinking that maybe it's a bug in Vista because those programs work great on my friends XP. Oh and the weird thing is that even though there isn't any video shown on the preview window... the programs' actually track the footage and export the track data into Blender even though it's not on the viewer... really strange. Syntheyes works just fine though, just have to save the cash to buy it.

I'm really glad that you posted this thread. I always thought that it'd be easier and faster to just render one pass with all the elements in one file but I thought about it a little more when I saw this thread and I realized how much control you would actually have if you had all of those elements split up into individual clips so that you could manipulate them one at a time. So yeah, my 1st thought was... this could be like working with layers like in Photoshop or Gimp. Anyway... I can't wait to get the insight of a pro who actually knows about this stuff.

Oh and thanks again for this thread Paul.
It's gonna be very informative and thanks to anyone who can help.

Frank.
Posted: Fri, 11th Sep 2009, 9:02pm

Post 8 of 23

Axeman

Force: 17995 | Joined: 20th Jan 2002 | Posts: 6124

VisionLab User VideoWrap User PhotoKey 5 Pro User MuzzlePlug User PowerPlug User PhotoKey 3 Plug-in User FXhome Movie Maker FXpreset Maker MacOS User

SuperUser

I'm not an expert, but the way I see it, there isn't a clear-cut answer to those issues. If there was, then you could just render them all at once at the best settings and be done with it. The whole point of multi-pass is that you can manipulate each of those elements separately, isn't it? And experiment with different composite techniques on each one?

I could be totally wrong there, but that seems logical to me. Then again, there is likely a good starting point for each element that will get you close (shadows would want 'multiply' or similar, speculars would want 'screen' or 'add,' etc.) So if someone who is experts chimes in, they may still have some good advice for you.
Posted: Sat, 12th Sep 2009, 11:30am

Post 9 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Ok,
Found a video tutorial on the net which might help explain the goal here, the vid quality is pretty crap but it's good enough to get the point. It's an example of rendering passes out of 3D studio max and compositing the layers in After Effects.

It's the link on the fifth box down posted by someone called 'scrimski'

Download it here:

http://www.vfxtalk.com/forum/compositing-3d-passes-and-z-t14441.html?t=14441

Let me know if it makes any sense

Actually, on the above link, pop onto the main forum index and there's a huge amount of useful stuff there for many people in the Fxhome community although me thinks that Visionlab should be in there in the 'Visual Effects Packages' section.

smile
Posted: Sat, 24th Oct 2009, 12:34pm

Post 10 of 23

GKDantas

Force: 475 | Joined: 17th Dec 2007 | Posts: 9

CompositeLab Pro User Windows User

Gold Member

Hi Paul, is good to see you here asking the same thing them I want to know. I only have Composit Lab, but looks that it can do something like what we want.
The problem here first is the blend mode for layers from Photoshop:

Multiply, Screen, Overlay and others (theres something like that in CLab?)

The other point is that Clab dont support PSD layered files, so we need to save everypass as a separeted file to work in Clab or Vlab.

Maybe you guys can take a look at this file:

http://www.euqfiz.com.br/PierView.zip (22Mb)

This PSD files contain almost all passes that we can have rendered in CarraraPro, so you guys can see how the layers work to control every pass.

Let me know if ou need mor einformation about this.
Posted: Sat, 24th Oct 2009, 3:56pm

Post 11 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

eek

Are you the same 'GKDantas' from the Carrara forums ????
Posted: Sat, 24th Oct 2009, 4:02pm

Post 12 of 23

GKDantas

Force: 475 | Joined: 17th Dec 2007 | Posts: 9

CompositeLab Pro User Windows User

Gold Member

Yes!!
Posted: Sat, 24th Oct 2009, 4:47pm

Post 13 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

well, hello there GKDantas

biggrin

Just getting around to putting some layers together to continue this thread and you beat me to it...*LOL*

Downloaded Gimp and opened the PierView.psd file....It sure is a good example of the available layers and quality produced but i did notice that the shadow pass is almost useless due to existing bug's within it. (better to turn off the shadow pass to see the composition correctly)

Not sure if CompositeLab or VisionLab will be most efficient using this .psd configuration or just to use seperate image streams, thats up to the fxhome developers to decide.

Tarn,

With not a lot of experience, gimp should hopefully be able to output each layer to seperate files so you can play with them to see what you can discover (i'd ignore the shadow pass as it's screwed - work in progress to eliminate the bug's at the moment !)

From what i can see by the layers, you can use the following:

Background
Diffuse
Specular
Global illumination
Sub-surface Scattering
Post Effect

Note:
If you were to composite this stuff on top of a background video layer as in 'CG on live footage' then it may be a little hard due to these layers actually being a whole scene, ill put together a quick layer set of a single character over the next couple of days which might help better.

I have just done a matchmove on the clip called 'Eyeline Interaction' near the bottom of the following page:

http://www.hollywoodcamerawork.us/greenscreenplates.html

I have done a simple CG character for this test scene with the goal of interaction between a live and CG character which i plan on loading up to youtube but ill attempt a couple of seconds of the individual layers at full resolution which i will try getting to you if your up for downloading the above test scene.

Until then....Bye all

Paul
Posted: Sat, 24th Oct 2009, 7:53pm

Post 14 of 23

GKDantas

Force: 475 | Joined: 17th Dec 2007 | Posts: 9

CompositeLab Pro User Windows User

Gold Member

Hi Paul, my idea here was to adjust the illustrattion that I do... but look thtat CLab dont support custom resolution. Some people use after effects to adjust the passes because its a non destructive way to work... you can always back and adjust again. In Photoshop you need to redo again and again, its a fixed effect...
Posted: Fri, 30th Oct 2009, 12:00am

Post 15 of 23

Chrispoole

Force: 1150 | Joined: 11th May 2009 | Posts: 1

VisionLab User VideoWrap User

Gold Member

Hi all,

I would really welcome this feature, there must be something about Carrara and Vision Lab Studio. The multi-pass option would certainly open a new revenue stream for FXHome with the power already available and the price. It would be so more attractive to people who never considered this package for the lack of multi-pass support.

Any chance of looking into a 'Velocity Pass' support as well and while I'm dreaming Z-Depth pass.

I'll follow this thread with anticipation.

Cheers
Chris
Posted: Fri, 30th Oct 2009, 12:16am

Post 16 of 23

GKDantas

Force: 475 | Joined: 17th Dec 2007 | Posts: 9

CompositeLab Pro User Windows User

Gold Member

Not only for Carrara users, all 3D software now can save multi pass render. And like I said if it can support other custom size files and Photoshop layer blending, it could be used to edit illusration to with the help of multipass render.
Posted: Mon, 23rd Nov 2009, 5:21am

Post 17 of 23

Divide

Force: 0 | Joined: 23rd Nov 2009 | Posts: 1

Member

I am interested in this program. I know a little about 3d stuff.
if you can import video(with alphas) and add adjustments to it like color adjustment, contrast, whatever i guess it would work. does this program allow you to do that? if so how many layers? is it like photoshop?
You would need to render out a z-depth pass so you can add depth of field. Ambient occlusion layer also. a few others.
I dont suppose theres any camera tracking with this program, if not then you are basicaly stuck with fixed camera .
Posted: Tue, 24th Nov 2009, 4:32pm

Post 18 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Divide wrote:

I am interested in this program. I know a little about 3d stuff.
if you can import video(with alphas) and add adjustments to it like color adjustment, contrast, whatever i guess it would work. does this program allow you to do that? if so how many layers? is it like photoshop?
You would need to render out a z-depth pass so you can add depth of field. Ambient occlusion layer also. a few others.
I dont suppose theres any camera tracking with this program, if not then you are basicaly stuck with fixed camera .
Hello Divide,
Visionlab does do what your first point refers to and a great job it does too, from what i can see there is no limit to the amount of layers you could use in a composite although the boy's at FXhome would need to confirm that, it is like a 'photoshop in motion' which is the best way i can describe visionlab.
Currently, visionlab does not support the use of z-depth passes for automatic depth of field effects nor can it use a velocity pass for auto motion blur either, but, if we are lucky in the future, maybe visionlab will support these things (both of which are actually elements of a multipass rendered output).
There is 2 types of motion tracking - 3D and 2D, 3D tracking is in the realms of 3D applications mainly and tracking software like syntheyes and matchmover pro provide the 'link' with the goal of producing a 3D environment from the origonal footage for CG integration. 2D tracking is where things like visionlab and after effects live and the main goal here is to track a point(s) through the clip which can produce a horizontal and vertical position for the object frame by frame (and rotational data too) where items/effects can be assigned/attached to take away the nessecity of doing it manually which takes ages and ages.....Unfortunatly, visionlab does not do 2D tracking either at this time but many users hasstle them all the time about features like this.

Visionlab actually does a very nice job with the ambient occlusion, shadow, specular and diffused passes as it already contains the required blend modes they need.

I continue to experiment when i get time

"Still no expert to enlighten us all ???" sad

Take care all
Posted: Thu, 3rd Dec 2009, 6:09pm

Post 19 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Message for Tarn.....

I have some passes for you to play with, if you PM me an email addy i will send them to you, probibly the easiest and quickest as it just a set of images not video's at this time.

Cheers
Paul
Posted: Thu, 3rd Dec 2009, 6:31pm

Post 20 of 23

B3N

Force: 3081 | Joined: 26th Feb 2006 | Posts: 1534

VisionLab User VideoWrap User FXpreset Maker Windows User MacOS User

Gold Member

You can send your attachments off to team@fxhome.com where Tarn can download them when he gets back to work tomorrow smile
Posted: Thu, 3rd Dec 2009, 7:27pm

Post 21 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

B3N,
Ok.....Thanks for that, ill do this some time tomorrow, aint got time now, also at the same time ill post in this forum explaining a few things ref these images etc.

Thanks again
Posted: Sat, 5th Dec 2009, 3:20pm

Post 22 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Tried emailing to team@fxhome.com and it failed to go

I'm sending it from a hotmail account, could this cause the failure


Any other ideas ?
Posted: Fri, 11th Dec 2009, 1:14pm

Post 23 of 23

pscamm

Force: 2225 | Joined: 5th Nov 2006 | Posts: 300

VisionLab User VideoWrap User Windows User

Gold Member

Hello,
Attempted to resend these passes to team@fxhome.com, this time split into 2 seperate emails. This time i have not recieved a failure notice so i can only assume they have gone through.

Both emails are headed in the subject line as :

ATT: TARN.......Multi-Pass Renders

The first email has all the passes attached and the second one contains the main fully rendered image (for reference) and 2 others the same which include a straight alpha and a premultiplied alpha.

If you were interested at any point, i can produce the following passes as does almost all major 3D applications:


Position
Object Index
Velocity
Surface Coordinates (UV)
Normal Vector
Depth
Fragment Colour
Fragment Coverage
Background
Ambient
Diffuse
Specular
Shadow
Glow
Refraction
Transparency / Refraction
Global Illumination
Caustics
Subsurface Scattering
Atmosphere
Volume Primitive
Volume Effect
Post Effect
Ombient Occlution
Material Diffuse
Material Specular

Happy Playing and if you need anything else then dont hessitate to contact me

Cheers
Paul