Rendering with Alpha Channel
Posted: Sat, 16th Oct 2010, 9:11pm
Post 1 of 6
I searched for an answer for this issue and didn't find it, so apologies in advance if this has been asked and answered before.
I'm creating an layered effect using a fire, a keyed-out head, and a preset-composite effect. Everything looks great in VisionLab, but the problem seems to be that when I export with alpha channels, I lose the alpha channel... but only if I include the preset composite, even though I'm using an object mask based on the keyed-out head.
Things are layered like this...
4. Preset Diffusion Glow (using object mask of layer 3)
3. Keyed-out Head (.mov footage using green screen)
2. Preset Stepladder Fireball
1. Background image (for testing... turned off when rendering)
If I remove (or turn off) layer 4, the alpha channel renders perfectly. However, when I add it back in, the alpha channel is lost when rendered.
I had the same issue when I manually added a composite effect IN layer 3's settings (same settings as the Diffusion Glow preset... just in the footage settings.
So it looks like any time I'm adding anything under the heading of "Composite," I lose the alpha channel on render.
I'm assuming that I'm just overlooking a simple setting, but can't, for the life of me, figure out what it is. Any pointers would be extremely appreciated! Thanks!
Posted: Sat, 16th Oct 2010, 9:31pm
Post 2 of 6
You've pretty much caught on correctly; Any time you apply a filter under the Composite heading, you can't render with an alpha channel.
As best I know, the reason for this is that the Composite Filters all function by taking the layer they are applied to, and combining it mathematically with the footage underneath. This isn't something that can be retained in a rendered video file. The glow, in this case, dynamically controls how the foreground and background interact, based on the color of each pixel of each layer on each frame. So any time the color of any pixel on any layer changes, the filter has to recalculate the results.
The ability to recalculate those numbers can't be stored in a clip of rendered footage, and it would have to be for the alpha to properly function. As a result, any time you implement a Composite Filter, the final composite will have to be done within your FXhome software.
Now, if you can explain why you need the alpha for this outside of VisionLab, we may be able to offer suggestions of ways to work around this, and still get the final result you are after.
Posted: Sat, 16th Oct 2010, 10:44pm
Post 3 of 6
Wow! First off, thanks for replying so quickly! Second, I'm shocked it wasn't something *I* did wrong. LOL!
So what I'm trying to do is create a short video of a man with the old "angel on one shoulder and devil on the other" effect. I'm just using angel and devil heads, though, so I have the angel and devil head video shot against a green screen. The main character is shot normally.
Since I have a lot of cutting and sound-synching to do, I was going to render the angel and devil videos with the alpha channel intact, so that I could cut and overlay them using my video editor (Premiere Elements) as needed. Depending on the shot, the "head" videos will be zoomed out and/or positioned differently throughout the video, which is why I thought pre-rendering that footage with all the effects would be easier than working the entire thing in VisionLab.
The glow effect I'd chosen just added a nice tint/blur to the entire effect of the devil head.
Hopefully, that's enough information to suggest a better work flow! I know I can do it all in VisionLab and will happily do that if need be. I just figured it would be easier (for ME, of course!) to do the editing/sound all in one step in Premiere.
Posted: Sun, 17th Oct 2010, 2:37am
Post 4 of 6
Depending on the look you are after, if some transparncy on the floating heads is acceptable, then you could just render them on a black background, with the glow composite intact. Then, in Premiere, change the composite blend mode to Add or Screen to remove the black.
Posted: Mon, 18th Oct 2010, 10:41pm
Post 5 of 6
Are you editing the video first?
You should render out the video in small increments after you edit it in your video editor the way you want or need it to be, get all your timeing just right... import those clips one at a time into Visionlab so you can add all your layers and effects... you should be fine with synching up the new footage with your audio if you first edit it.
Hope this helps.
Posted: Wed, 20th Oct 2010, 12:23am
Post 6 of 6
Thanks, spydurhank. I think that sounds like the best plan. I am going to edit first, so that should work for me.
I don't consider this a problem. It's just something that requires a different workflow.
I appreciate the input from all!