View Full Version : iDof 3 available for purchase

03-22-2006, 12:02 AM
You want iDof? Go get it. (or rather, pay me, and I'll email it to you).


03-22-2006, 07:55 AM

Looks like you like to composite.
Got any ideas on where to get a multipass plugin that will do separate passes(diffuse, color, spec, shadow) for each light separately and automatically?

03-22-2006, 08:06 AM
Unfortunately I don't. Surpasses is the only Multipass tool i know of, and Dodgy's LayersMC a close second, but neither do what you're asking that I know.

Personally I think LW's strength of being able to do it all in-camera is also a weakness, because a lot of users never realize the power and control of the compositing process. Thus iDof's recent makeover. Also, now that I've figured out how to get TGA output from an image filter, I'm thinking through other possibilities right now.

Now, if you did the separate passes for diffuse, spec, color, and shadow, you'd need either a flat grey geometry to receive it (so the geometry's colors don't get doubled in comp when the spec is screened over), or you'd need a diffuse pass, then a diffuse with a specular and have the specular areas be the alpha as well (so you could just lay it over the top of the diffuse without blending addition or multiplication and have the alpha differenciate where it goes), etc.
Which are you thinking?

03-22-2006, 08:40 AM
actually, you don't need light COLOR, as you can change the white to any color you want in POST.

Right now I do, starting from bottom layer: "RAW SURFACE COLOR,normal" a "DIFFUSE,multiply" "SPECULAR,screen" and "REFLECTION,screen"

This is good but you don't have control over each light's diffuse/spec...

But I don't see why each light could not be rendered out separately, automatically with a plugin...

03-22-2006, 08:41 AM
Also, how in the world do you get acceptable results of DOF blur from a compositor? I get horrible looking edges where one "depth" meets another...

03-22-2006, 09:51 AM
i see what you're asking for now, as far as the buffer export. Let me mull it over a bit.

As far as those lovely edges, I usually have my scene broken up into layers enough that when the layer is blurred, there's info behind it and it cuts down on those edges.
For a good commercial post app, LensCare (http://www.frischluft.com/) is good, and cuts down a bit on those edges.
I've heard good things about Depth of Field Pro (http://www.richardrosenman.com/dofpro.htm) but it's a batch image processor at best.

FG out of focus elements are almost always in a separate layer for me. What compositor do you use most?

And FWIW, iDof Post Blur tries to account for the edges a bit (little or no halo on in-focus objects, at least).

03-22-2006, 05:06 PM
One thing that actually works really well. In your compositor. Slice your image up into about 30 layers, based off the depth channel. (depth channel allowing you to get different depth slices) then gradually blur each slice/layer a little more than the previous. You can get incredible results with this if I remember correctly. At one point I made an AE project like this, set up with expressions to control the blur point. If I remember it took FOREVER to render but was very accurate...

03-22-2006, 05:15 PM
I suppose so. But if you render into layers anyway, you save yourself a step. And Lenscare and the like adjust to prevent Halos. So render to as many layers as you like, then apply lenscare to the entire composition. If you need to, you can have extreme FG elements in a different composition and blur those independently, then layer the FG comp over the main comp. (this being in AE, of course).

04-02-2006, 07:06 AM
I guess the spread is like where the focus is? Would be nice if this was controlled by a third, middle null.

04-02-2006, 09:15 AM
black point is the 'middle null' or where the focus is. white point controls the distance it will take to fade to white (both in front of, and behind the black point, unlike the default depth matte). spread is where that fade begins.

with 100% spread, the fade begins immediately and is a gradual fade from the black point to the white point.
with 75% spread, the black area is extended 25% of the way to the white area before starting to fade.
with 25% spread, the black area is 75% of the way to the white area before starting to fade to white.
with 0% spread, the black point is pushed all the way to the white point with no fade. you have a hard line between them.

04-02-2006, 01:21 PM
Oh, now I get it :), havent tried it yet... Good work :thumbsup:

04-06-2006, 10:22 AM
hmm. haven't encountered that. just curious...do you have depth buffer AA on?

04-08-2006, 04:34 PM
Haha. Wow, I didn't see the test scene post. Sorry mate. You know, I do have an email for this type of stuff. I'm pretty good with getting back to people. Feel free to try that before the 3 thumbsdowns next time.

04-08-2006, 04:43 PM
I assume you hid the skydome for your depth renders?

04-08-2006, 06:44 PM
To be honest it's very subjective, and every compositor will tell you that their way is the right way.

Personally, I render most of my stuff at HD or half HD and scale down anyways. Not just for DOF purposes, but for clarity and in case I need to deliver in a different format, I won't lose quality.

Then I render my scenes in multiple passes and apply blur to my environment and/or distant characters together, then for foreground elements I apply blur to each pass separately based on the depth matte. If ever it's yielding results that aren't good, or I'm getting halos/sharp edges, I apply a lens blur on that layer only and eye it. It happens.
Another fix for depth matte problems that I do is use it as a matte for a blurred version of the footage. It's incorrect physically but works in a lot of cases. And since it's a matte or mask, you can blur it's edges there in comp.

LW's depth mattes need work, honestly. As a matter of fact, all the problem's with iDof Channels come from the fact that it pulls depth information from the existing depth channel info and gives new visual results. It reworks the data in a form more suitable for DOF in comp. Unfortunately it's based on the same old data. When you see an odd rim or AA error in the DOF depth matte, that was already there. It's just more obvious when you start changing pixel colors.