More About Alpha Masks

We were talking tech in the DAZ3D forums one day and the subject of blur and masks came up when talking about this picture:

What Eats the Monsters?

What Eats the Monsters?

I had said: Using Semidieu’s Advanced Render Setting script, I got the normal render, a depth cue render (alpha), and two alpha mats done in under 3 hours. …

The response was: Aha, I believe you mentioned Semidieu’s script in an earlier thread. Do you recommend getting the whole package or just the alpha’s? (They sell the alpha’s script separately as well and, given the March Madness, my budget is pretty much maxed.)

If you have the $20 and use Poser 7 or 8, Pro or Pro2010, I’d get the Advanced Render Settings from Runtime DNA, which gets you a whole slew of good tools in one go. Tools you will use. Especially if you have Poser 7 Pro or Pro2010. There is a version especially for the latter.

Or you can spend $10 and get passes and alphas (you’ll want both) and spend a lot more time fiddling with them. This didn’t work quite as satisfactorily as the bigger kit, for me.

I then said about this picture: The motion blur was done and then composited so that it didn’t rub out the Oni’s details too much.

The reply: Also, I may be a bit thick this morning, but did you do the motion blur in postwork as well or in poser? If it is done in poser and I know it can be done, can somebody explain how? If it was done in postwork, then also, how?
I feel like such a tube/newbie. (Which in fairness I still am, have only really done this for a year now.)

Motion blur in postwork:

You can do motion blur in Poser, but it’s not as easy to control and things like rendering depth of field and motion blur slow your render down dramatically if you have your render set for high/final quality. I have Photoshop and have used it for many years, so I’m quite comfortable with doing postwork; your experience my differ.

How I do motion blur in Photoshop varies. In this case, I recall this was about what I did:

  1. Use alpha mat for the Oni figure to create a selection and copy of him out of the render layer.
  2. Save that as new layer directly above.
  3. Use motion blur filter on that new layer. Hmm, can’t see his face that well now, details too blurred.
  4. With the blur layer still selected, I used the alpha mat to create a selection and this time feathered the selection some.
  5. Deleted within that selection. Now I can see him better!
  6. May have fiddled with blur layer opacity and blend mode after that.

Addendum about alphas masks:

In Photoshop, these are grayscale images that are put into the Channels palette below the usual R, G, B & composite channels. White is “show” and black is “hide”. The Alpha mask renders that you get out of Poser are about 1 pixel too large (you get ugly edges unless you contract them by 1 pixel), but they are far, far more accurate and time-saving than trying to make a selection yourself.

You click the selection icon in the Channels palette and the white parts become a selection. The beauty of this lies in the use of grayscale in these channels, and that you can paint or modify them the way you do layers to a large degree, thereby giving you a LOT of control over your selections and masks. Want soft edges? Apply a Gaussian blur. Need something to fade gently? Apply a gradient.

(Pro2010 will render you a Z-depth mask as a standard option–this is HUGE! and os useful! The depth cue image is grayscale, however you get it—the Render Passes script will also do it for you if you are working with Poser 7. In Pro2010, it comes in as a layer above your rendered layer when you save your render as a Photoshop document. Grab that layer and paste it into a fresh channel in your Photoshop doc and label it “Z-depth.”)

Now, go back to the layers palette, make a new layer of your render (saving the old one, just in case) and use the Lens Blur filter, you would identify that depth cue channel as your mask. Now you can adjust how much blur, how much noise and such in real-time. Awesome! It’s like fixing the F-stop on the fly. How good is that? :D That’s why I do that. No need to take forever re-rendering because the depth cue wasn’t what you wanted.

Earlier Discussions:

Leave a Reply




You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

A sample text widget

Etiam pulvinar consectetur dolor sed malesuada. Ut convallis euismod dolor nec pretium. Nunc ut tristique massa.

Nam sodales mi vitae dolor ullamcorper et vulputate enim accumsan. Morbi orci magna, tincidunt vitae molestie nec, molestie at mi. Nulla nulla lorem, suscipit in posuere in, interdum non magna.