Rendering & Compiling Multiple Passes

[JamesTaylor]

I'm looking at render out a number a passes for the animation i'm currently working on, most likely these will be the main, complete, image and then additionally a shadow pass and a reflection pass - however this is the 1st time i've done it and hoping to get the heads up on the do's ands don't of the process as i won't have to much time to test the procedure.

 

I'm going to rendering via Vray and Viz and have access to both After Effects and Combustion (my working knowledge of AE is considerably better than that of Combustion even tho i've recently had combustion training) and there will be approx 4000 frames to render.

 

So what do you guys think... best file format to render to (file size isn't really an issue but network load may be problamatic as sizes get higher)? preferable methods to use in AE or Combustion? Things to consider prior to rendering to make life easier later? Things to do / not do in Post Production??

[Stephen Thomas]

I too would appreciate a few pointer in this area having experimented with this in the past. However I have noted from previous non-responses to questions that this forum is a bit of a black hole. We could use a moderator who really knows his stuff. Anyway, a few pointers in the right direction which may help you out.

 

There is a tutorial here which shows a working example of a scene built up in this way and how to combine the layers.

http://evermotion.org/?unfold_exclusive=157&unfold=exclusive

 

Autodesk also has some short videos on bringing render elements from 3ds max into combustion as well as an intro to RPF files here. They are quite small on screen, but if you've got the Combustion Unplugged bonus DVD you can find them on there too.

http://usa.autodesk.com/adsk/servlet/item?siteID=123112&id=5574967

 

If you want to use 3d post effects such as fog or dof etc. the Combustion help files tell you which g-buffer elements you require for each effect. Then either save everything to jpeg sequences (or targa if you need alphas) and combine using g-buffer builder, or write to RPF file sequence (these can become very large though). Finally layer up your additional passes and set your layer modes accordingly (multiply for ao passes, add/screen for others) and adjust the layers opacity to suit.

 

To those more experienced please feel free to add to this or contradict me, I don't mind!

[JamesTaylor]

Thanks for the reply Stephen, i think your right its certainly seems to be somewhat of a black hole area in comparision to the wealth of knowledge CGA members can provide in our more traditional subject areas. Hopefully this will change given a bit of time as its becoming a much more integral part of the industries workflow and will continue to do so in the future i feel.

 

anyways... i notice in the RenderElements section that there seems to be 2 versions for each available channel - 1 is simply the vrayShadows / Reflection and the other is vrayRawShadows / RawReflections.

 

can anyone shed some light on which version is preferable to use and perhaps some insight as to why there are 2 versions?

[Stephen Thomas]

[quote name=anyways... i notice in the RenderElements section that there seems to be 2 versions for each available channel - 1 is simply the vrayShadows / Reflection and the other is vrayRawShadows / RawReflections.

 

can anyone shed some light on which version is preferable to use and perhaps some insight as to why there are 2 versions?[/quote]

Sure can.. Raw means that it has not been multiplied by the diffuse surface colour. I assume this is so you could change the colour of an object without retaining a tint in the scattered light from the original colour. Probably not needed, but this will shed some more light on it.

http://www.spot3d.com/vray/help/150R1/render_elements.htm

[Nils Norgren]

Either After effects or Combustion would work for combining the passes to re-create the "Beauty Pass" (Vray calls it the "RGB color") for compositing information there are many useful books. There are general ones, and ones that are specific to applications.

 

There is a more fundamental question, which is why use all the passes to re-create what is already created by default? The main reason to combine all the passes is to get control over each and every piece of the render in post. This is often (in my opinion) overkill, if a compositor has analyzed a shot and realized that there are going to be some problems in the shadows, a shadow pass will allow it to be isolated, vector passes allow post processed motion blur, and z-depth will allow depth of field or atmosphere to be added and adjusted in post.

 

In the help files it states (at the top)

 

"the Raw GI and Raw Lighting elements must be multiplied by the Diffuse element before they are added to the final image. "

 

The version at the bottom of the page reads

 

"Self-illumination + Direct lighting + Global illumination + Reflection + Refraction (all added); the result is the same as the Real RGB Color channel"

 

At the bottom of the page are examples showing that the Raw GI + Diffuse will look like the GI pass, the Raw lighting + Diffuse will look like Direct Lighting only, the method of combining these passes is shown between the words in this case they are all Multiplied, (*) to get the result in AE you would put the passes in a comp, and the top one would get a "Multiply" transfer mode. The main passes (to achieve the "Beauty pass") would all need to be "Added" (using the "Add") transfer mode. As for the other "Raw" passes I suspect that they are the same deal, all of this information is calculated by vray, therefore (again I suspect) that Vlado and the other developers enabled the ability to export the information as separate channels. The only way to figure out if you need them is to work up a test comp, and see if you need the extra control (or can even get it to work).

 

All of this I should re-iterate is somewhat speculation, as I have not used any of these specific passes. We generally only render out main passes and sometimes a z-depth, shadow or diffuse for further control in the comp.

 

Generally all transfer modes translate to simple math, in these cases literally what they are called, Multiply or Add, each pixel's value in the RGB are either "Added" or "Multiplied" over the pixel(s) that are under them.

 

One shortcoming is (at lease in older versions) Photoshop does not have an "Add" transfer mode, "Screen" is fairly close, but it is not the same. Screen is actually an invert of both multiplied and the result inverted, for a bit of obscure transfer mode math.

 

Hope this helps, again there are many books that get very deep on the subject. Some more theoretical and some more practical.

 

The ultimate in geeky details is Ron Brinkman's book (Text book on the subject) but there are some that are less about the under the hood details and more about the real-world uses.

 

http://www.amazon.com/Science-Compositing-Kaufmann-Computer-Graphics/dp/0121339602/ref=pd_bbs_1/002-3206390-3118425?ie=UTF8&s=books&qid=1191933411&sr=8-1

 

This is an excellent book with many cool AE tips and tricks, but it takes some mental leaps to apply it to our industry.

http://www.amazon.com/DV-Rebels-Guide-All-Digital-Approach/dp/0321413644/ref=pd_sim_b_4_img/002-3206390-3118425

 

Our Demo reel: http://www.neoscape.com/clients/demoreel07/

 

-Nils

[JamesTaylor]

Thanks Nils, i kinda makes sense on 1st read! i'm not quit at the point of having a go with building up a test comp but should be in the next day or two, so i'll be re-reading your post a few times i think!

 

i'm not planning on rendering every available channel to build up the Beauty Pass, i was planning on rendering the final image as a whole and then using a selection of passes to tweak certain attributes. For example i know the shadows will need adjusting and the likelyhood is that the client will want to continally tweaking this element until they are happy so i figure doing it in post is going to be much more flexible and less demanding than re-rendering 4000 frames.

 

i'm sure more questions will arise as i jump in and have go, but your reply is much appreciated...

 

from experience have you found using any particular file type to say out to works better than others??

[Nils Norgren]

from experience have you found using any particular file type to say out to works better than others??

 

My preference is .exr images set to Half float (16 bit per pixel) it is fairly small with lots of latitude for post. Passes can be whatever, although having z depth in 16 bit is good, at 8bit things can get posterized.

 

.png files can also be 16 bit,

 

Good luck

 

-Nils