Return-Path: Received: from mail-lb0-f169.google.com ([209.85.217.169] verified) by media-motion.tv (CommuniGate Pro SMTP 4.2.10) with ESMTP-TLS id 4890825 for AE-List@media-motion.tv; Sun, 04 Nov 2012 23:45:02 +0100 Received: by mail-lb0-f169.google.com with SMTP id k6so3399275lbo.28 for ; Sun, 04 Nov 2012 14:48:01 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type; bh=Bk6SwMUyFEqqFxC1iGO8EmpebDellaggd/lvHJTP5Vo=; b=DcEEuk3YmFhdoI0ts1x8gBP/VDMhnWRvHsvIDRY1/Juvwda1vgeRTVk/eEbF3cFDCF JBhwzbbx+Cga1vuyGylWW/IiYQGaEirkZXyGRrw3CgiSEhBZz8oUSZhT4XAxNP+xLocr fEaRIV/E/85CMUNPgGAQSJqImdB6jE7vMBOzoSrYe3PRJafI5dxNHePnIdVc+sQujqFG MD5kAEIJNCpWZ7HjE5vTcPnv7/+B2gkySPfxdMZanRQMzjIh8wC5ABU7AL3lfLjM9IRq AQDFrqHJ6oF2Thh11KhDu0/DDyFWcSjtRJt35eseZrMMAzMaLHl+QvOj0xY9ClaKby5J tzMQ== Received: by 10.112.50.106 with SMTP id b10mr3272885lbo.122.1352069280826; Sun, 04 Nov 2012 14:48:00 -0800 (PST) MIME-Version: 1.0 Received: by 10.112.44.67 with HTTP; Sun, 4 Nov 2012 14:47:20 -0800 (PST) In-Reply-To: References: From: Teddy Gage Date: Sun, 4 Nov 2012 17:47:20 -0500 Message-ID: Subject: Re: [AE] pulling a better difference matte? To: After Effects Mail List Content-Type: multipart/alternative; boundary=f46d0401fe5932044b04cdb3298b --f46d0401fe5932044b04cdb3298b Content-Type: text/plain; charset=ISO-8859-1 yeah, I'm totally hosed. It's an Arab-American actor in a brown shirt and khakis against wood grain shelves, brown bottles and heavy motion blur in totally flat lighting. Thanks for the roto brush tip but even it is super confused. The movement is too fast with too little color difference. Going to try to boost the levels a bit. But it may still be faster to tweak the roto brush than frame-by-framing it. thank god for the wacom anyway -TG On Sun, Nov 4, 2012 at 5:31 PM, Brian Higgins wrote: > Difference mattes have gotten more useful with the advent of high-quality > digital cameras, but still rarely work. It's easy for your brain to say > "look, the actor is in one plate but not the other," but the computer can't > tell what's an object and is only looking at differences in color. If your > actor's skin tone is similar to the shelves behind him, you're hosed. I > would look into the roto brush like Jack said, and maybe give the Furnace > tools from the Foundry if you have access to Nuke. If not, I'd bite the > bullet and start rotoing. If your actors are *really* flailing around, > hand-painting the mattes may be quicker than yanking splines around. > > $.02, > Brian > > On Sun, Nov 4, 2012 at 4:19 PM, Teddy Gage wrote: > >> Hello, I'm working on an indie film and they have presented me with >> some "interesting technical challenges," to put it diplomatically. The >> current shot I'm working on is two actors getting riddled with bullets, >> with no practical effects or greenscreen. I am adding blood, bullet holes, >> dust, smoke and glass. they are flailing about like rag dolls then go down. >> It lasts about five seconds. There are rows of shelves behind them with >> glass bottles that stay intact throughout the shot. In a separate shot, the >> director then pulled the bottles off the shelves using strings, and the two >> shots need to be composited together. If only the actors had been shot on >> greenscreen, this would be easy. But there are enough changes between shots >> that a difference key only gets me about 50% of the way there. However this >> ancient effect is not 32-bit, has awful controls, and it's almost >> impossible to get a clean difference matte; half the background is pulled >> into the effect and I cannot cleanly separate the actors. >> >> Of course, I could roto actors out of the scene, then lay them on top of >> the glasses breaking, but I'd like to avoid this. Are there any tools that >> are better for pulling out a motion matte from a static background other >> than difference matte? Or is there another way to do this I'm not thinking >> of? >> >> The approach I currently have going is to pull a difference key of the >> bottles flying off the shelves from the static shot, then to roto the >> actors' arms etc. in just the regions where they intersect with the >> bottles. This way it appears the bottles are being shot up behind the >> actors, as they should be. This at least saves me from having to do a full >> roto of each actor against the background. Would love to hear any other >> suggestions, thanks >> TG >> >> -- >> Animator & Editor >> www.teddygage.com >> Brooklyn >> >> > > > -- > *brian higgins | creative director > Sol Design > 312.706.5500 > higgins@soldesignfx.com > soldesignfx.com > * > > > > -- Animator & Editor www.teddygage.com Brooklyn --f46d0401fe5932044b04cdb3298b Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable yeah, I'm totally hosed. It's an Arab-American actor in a brown shi= rt and khakis against wood grain shelves, brown bottles and heavy motion bl= ur in totally flat lighting.

Thanks for the roto brush tip but even= it is super confused. The movement is too fast with too little color diffe= rence. Going to try to boost the levels a bit. But it may still be faster t= o tweak the roto brush than frame-by-framing it. thank god for the wacom an= yway

-TG

On Sun, Nov 4, 2012 at 5:31 PM, B= rian Higgins <higgins@soldesignfx.com> wrote:
Difference mattes have gotten more useful with the advent of high-quality d= igital cameras, but still rarely work. =A0It's easy for your brain to s= ay "look, the actor is in one plate but not the other," but the c= omputer can't tell what's an object and is only looking at differen= ces in color. =A0If your actor's skin tone is similar to the shelves be= hind him, you're hosed. =A0I would look into the roto brush like Jack s= aid, and maybe give the =A0Furnace tools from the Foundry if you have acces= s to Nuke. If not, I'd bite the bullet and start rotoing. =A0If your ac= tors are *really* flailing around, hand-painting the mattes may be quicker = than yanking splines around.

$.02,
Brian

On Sun, Nov 4, 2012 at 4:19 PM, Teddy Gage <<= a href=3D"mailto:teddygage@gmail.com" target=3D"_blank">teddygage@gmail.com= > wrote:
=A0=A0=A0=A0 Hello, I'm working on an indie film and they have presente= d me with some "interesting technical challenges," to put it dipl= omatically. The current shot I'm working on is two actors getting riddl= ed with bullets, with no practical effects or greenscreen. I am adding bloo= d, bullet holes, dust, smoke and glass. they are flailing about like rag do= lls then go down. It lasts about five seconds. There are rows of shelves be= hind them with glass bottles that stay intact throughout the shot. In a sep= arate shot, the director then pulled the bottles off the shelves using stri= ngs, and the two shots need to be composited together. If only the actors h= ad been shot on greenscreen, this would be easy. But there are enough chang= es between shots that a difference key only gets me about 50% of the way th= ere. However this ancient effect is not 32-bit, has awful controls, and it&= #39;s almost impossible to get a clean difference matte; half the backgroun= d is pulled into the effect and I cannot cleanly separate the actors.

Of course, I could roto actors out of the scene, then lay them on top o= f the glasses breaking, but I'd like to avoid this. Are there any tools= that are better for pulling out a motion matte from a static background ot= her than difference matte? Or is there another way to do this I'm not t= hinking of?

The approach I currently have going is to pull a difference key of the = bottles flying off the shelves from the static shot, then to roto the actor= s' arms etc. in just the regions where they intersect with the bottles.= This way it appears the bottles are being shot up behind the actors, as th= ey should be. This at least saves me from having to do a full roto of each = actor against the background. Would love to hear any other suggestions, tha= nks
TG

--
Animator & Editor
= www.teddygage.com
Brooklyn




--






--
Animator & Editor
www.teddygage.com
Brooklyn

--f46d0401fe5932044b04cdb3298b--