-
Three-Pass Method for KeyLight – Best way to do multiple passes
I’m trying to use a three-pass method for keying that is described Mark Christiansen. My goal is to improve the quality of my key by isolating the edge of the foreground in order to fine tune it’s quality. I’d like to verify whether I am going about the process correctly. Also, I wonder if this is the best way to attain a good edge on the foreground. I am using KeyLight in AE. With just doing one pass in KeyLight I am often left with jagged edges and fringing–a thin dark outline around the edge of the foreground.
Working with an actor shot on greenscreen. The lighting was good, but on the wide shots (actor framed head to toe) there is variation in the luminance on the background. That makes using just one pass in KeyLight an unfavorable option.
I have searched the forums and I have had some success achieving a good edge by using the ‘Super Tight Junk Matte’ tutorial in addition to duplicating the source layer and using mattes for doing multiple keys on the same take–one layer for the torso, one for the head and one layer for legs of the actor. Also combining the resulting key with light wrap instructions by Darby Edelen (found in these forums – https://forums.creativecow.net/readpost/2/928509).
I was curious if a 3-pass method might be a better way to go about getting the best edges out of KeyLight. So I tried to emulate the instructions I found for the “Three Pass Method,” I am following the instructions found at this URL:
https://www.peachpit.com/articles/article.aspx?p=1278847&seqNum=3My adjusted workflow:
First I create comp with the footage and apply Keylight. I make an OK key using colour picker (selecting a highly saturated pixel from the greenscreen) that drops out most of the background, and using clip black and white.Second, I create a precomp containing the footage layer with Keylight applied. I call the layer in the precomp Edge. I duplicate that layer and name it Background. I place the background layer at the top. On the Background layer (in Status view) I adjust the clip black setting until all gray is removed from the background. Then I lower the clip white to just above the clip black number (Ex. if Clip black is 30, then white is 31). I expand the matte so it extends beyond the edge. I switch to Final Result and set the Background layer’s blending mode to Stencil Alpha.
Third, I duplicate the Edge layer again and call the new layer Core. The Core layer is in the middle–under the Background layer and above the Edge layer. On the Core layer I lower Clip White until the foreground is opaque, and raise the Clip Black just below that value. I then shrink the matte so it is within the edge and does not touch the edge. I switch to Final Result and change the layer’s blending mode to Silhouette Alpha.
The Edge layer is set the Normal blending mode. I am now left with the edge of the foreground visible. I can go into this layer and refine the keylight settings to get a good quality edge.
After getting the edge layer to an acceptable point I change the Core layer’s blending mode to Normal for the final viewing.
-
1 Reply
-
I did forget to list the specs of the footage. We discussed the color limitations of HDV in pre-pro and chose to shoot with the native DVCPRO HD 720p30. In addition for our workflow, we transcoded the footage needed for compositing into Sheer RGB 10bit in order to retain the color and gamma when going from a FCP to AE (PC) and back to FCP.
Reply to this Discussion! Login or Sign Up