Gray Scale Imagen For Z Composition

[user_deleted]

I'm trying to compose in AE a Z depth effect i have the Z information render in tiff gray scale image how to compose to take the right DOF effect ? its possible ?

[joojaa]

Dont render the z out as a grayscale use the native z depth (its free of render time sice its done use it or not) it has more resolution and since its the real floating point data (its not grayscale its jut one way of showing it to human), otherwise you end up with normalization problems. Offcourse you need to have ae pro.

[user_deleted]

QUOTE(Joojaa @ 10/31/07, 12:39 PM) [snapback]275695[/snapback]
Dont render the z out as a grayscale use the native z depth (its free of render time sice its done use it or not) it has more resolution and since its the real floating point data (its not grayscale its jut one way of showing it to human), otherwise you end up with normalization problems. Offcourse you need to have ae pro.

but in cant read the iff Maya format in AE exist other format ho contain the Z information and work with AE?

[joojaa]

QUOTE(MasterFreak @ 11/01/07, 10:33 PM) [snapback]275764[/snapback]
but in cant read the iff Maya format in AE exist other format ho contain the Z information and work with AE?

RLE? but last time i saw it does support iff.

[y030052]

There are different schools of thought about the best way to do DOF, so take some of the advice here with a grain of salt. You'll have to figure out the best approach for your needs.

First of all, last I heard AE did not support the depth channel of iffs, but can read them from RLA files.

However, if you have your depth rendered into the RGB I would suggest that you buy the Lenscare Filter from Frischluft. This is extremely easy to use and allows you a lot of control over your depth of field effects.

There are many advantages to NOT using the native depth channel in file formats such as iff, the most obvious one being that you are limited to 8 bits of information and thus the number of depth values that can be rendered.

One workaround for this is to render at twice your resolution and then downrez in compositing.

However, personally I MUCH prefer the RGB approach -- for one thing it allows you to see right within Maya where your black and white points are going to fall.

[joojaa]

QUOTE(stevedavy @ 11/02/07, 12:39 AM) [snapback]275780[/snapback]
There are many advantages to NOT using the native depth channel in file formats such as iff, the most obvious one being that you are limited to 8 bits of information and thus the number of depth values that can be rendered.

Neither is true the z in iff or rla is its actually floating point which is better than what a 16 bit format can do, AE just turns it ito a 8 bit to show it to you (howevere many of the operations bundled in the software does take this info at full value of depth). But it can use the full floating range. But its a bit bady designed! So you cant pull out a deeper field out in AE so a plugin thats not aware of z can't utilize the range. (this is MAJOR design flaw in ae that supposedly been fixed in cs3)

Now adobe in its infinite wisdom keeps pople loicked in lets do it the hard way land.

So that would be a blind leading teh blind kindof statemant. Its been while ive used AE for this tough. Ist a bit painfull to say the least bacause AE is made to work in any case and shotcuts on data manipulation in many places.

But they have upgraded the z workflow in cs3 i hear. And cs 3 accoring to my information does read maya iff files with z and can use it floating.

BUt i mean if you compare the workflow of shake or even better digital fusion youll soon understand why AE users have so much problems. I mean fusion uses zdepth totaly trasparently as if it was nay other channel, zblurs very well. But also uses coverage automatically in z comp which is a bit of hustle in shake. BUt comparing the one node in place solutions of BOTh this software the way you need to do this in AE is just insane. Bottom line is that Adobe programmers didnt jhjust think of graphics originally the way 3d people do. Wich allmost dorpped them of the race.

[y030052]

If AE CS3 supports Z-depth in iffs then I stand corrected. In so far as my comment about bit depth, this certainly used to be the case and often led to nasty aliasing around the edges of objects where the adjacent depth values were too different. One production solution to this was to render at double rez and then downsize in post, effectively antialiasing the depth channel.

However, Joojaa how about some useful advice for the original poster, instead of using this as a forum for complaining about After Effects? People come here to get solutions, not a bunch of holier than thou preaching about how rubbish the program they're using is. And, use the spell check, huh?

FYI, I've had very good depth of field results with the techniques I outlined.

[joojaa]

Yes, sure. The aaing has nothing to do with the fact that the z channel is inaccurate. But rather that z is a measure of distance. Wich is not a continious filed of data (a z image has no and should not have a aa because objects dont mysteriously merge together in images but rathjer they remain discontinuous)

However each pixel is a amalgame of many distances the z just happens to be one value it is either the closest or mo the furthest visible part of the pixel. Because rendering at 2 res is a bit heavy (it takes 4 times the time plus 4 times the sam etime to comp also), most renederers nowadays can output either the unfiltered super sample field or what is known as coverage both at minimal render time penalty (both cost some extra memory tough).

The coverage is used to reintroduce aaing fixes to the discontinuous field. It represents a sort fo local aa where it tells how many percent of the pixels contribution actually hit the primary object. This enables you to separate the parts of a pixel. Off course for dof this is unneccesery, the quality of a 2d dof is so coarse thet it couldn't account for this very well in any case.

If you need a local sharpening you isolate the areas presumably by id tags and diallate the result by one pixel to compensate the neibouring coverage pixel for a accurate map.

The really highend are possibly moving away form this tough for not so complex shots. What they are doing is to save out deep maps where each pixel has multiple depth values and alphas thisway each pixels dof is jut a sort fo pick and choose operation.

[qcoe]

Few weeks ago,I had 2 do some rain effects with particles in Fusion.Anyway,to make reasonable good rain isn't the easiest task2do regardless of which software Y use.However,what helped me 4 this rain 2 happen is the flexebility of Z depth manipulation in F.Adding some depth blur & a little bit of fog,got the impression of not just flat image plane of rain in the FG,but also a good depth impression in the BG.

[jsahn]

so how to use a separate z-depth layer in AE? or is it possible?

h.