From: "Roei Tzoref" Received: from mail-il1-f171.google.com ([209.85.166.171] verified) by media-motion.tv (CommuniGate Pro SMTP 6.1.0) with ESMTPS id 7491106 for AE-List@media-motion.tv; Sat, 11 Jan 2020 23:02:15 +0100 Received: by mail-il1-f171.google.com with SMTP id v69so4745581ili.10 for ; Sat, 11 Jan 2020 14:02:15 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=mime-version:references:in-reply-to:from:date:message-id:subject:to; bh=aknm/golpHgQE9pX2GO63TPZNlF1FW3MDGh2BhL3P5o=; b=Smos4CTSwO+YlJllec3sr+IInMaZq/zx+B1vAJCaCew2Q/1Y8q5CCzRWWAORlb9y2Z yAeHDQJm4ZZ3PSl/WblaOAB4SoycAn2gmyB6vGjalHSmfuvb+NOPo6ohbWCFead/nzLi WNPzkv8YiQyHJM7CNnrpm8eP0nfnWwxBPvYJ69qkpEBVEizGADEm16sO0UxRYisJlxXj YOr/mcMzm0qMQDaKxC2V/cfrWPzTrCtF9OfPNG2TagsQBGbDWvYKtZDg2Xd8kbVlVs7/ fP+rm1chdPzQ+y2FW1L/0H+z4BrWry6KgQfo/L8XPwd6x3aDNImjuM7w0CEiDfU7vfPA N12w== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to; bh=aknm/golpHgQE9pX2GO63TPZNlF1FW3MDGh2BhL3P5o=; b=rTGD9BC+fHHOKfMyGvrScqJsiAa6t33tJ24gY+JWLhHHDkuiKAcCzPI5yCF598Ln8L ofoAGrl3UQF2kDBnUuPcrWVZhnAFB+R0/+Z3/kqMHD0xdqfEBQYoerRJu1YoB9nSMqEy OOI2DOWn8GgorTEBTzG1KNbXZgNZnxDGLYM3OWEjivdPPpDjgvNV2/7DMj5lMVFmTHLv 5xH8eYI/ods/U7wQcuJBviw5Zj4/JyJ+SgeLjXRU38xua1Der24/2TyKy2Qm6fgQLgqt lszjYNmQy6glrD2e0M2YxnXQC70DTxWEcr9rZBzqGzipdna/cFQfbQRR+lPuG05ScIT3 OhDA== X-Gm-Message-State: APjAAAVspmkilFhWj+rntNz/xEL514yxelh2WzeWTfA5SNQvJdbE+SaX wuqfR/Rx/8njFmVvC6Vwmtj2rSRU36a7pAw7+WBRBg== X-Google-Smtp-Source: APXvYqwgdWd2VKjrtUF/WVTF/taIA/olzbwgryifcXd2rTVJms5FACbucXJHV76pZ72aFpjZoDonlQP2hPFvnfMt57g= X-Received: by 2002:a92:c990:: with SMTP id y16mr9148902iln.105.1578780133578; Sat, 11 Jan 2020 14:02:13 -0800 (PST) MIME-Version: 1.0 References: In-Reply-To: Date: Sun, 12 Jan 2020 00:01:37 +0200 Message-ID: Subject: Re: [AE] Keying with multiple masks and zero overlapping To: After Effects Mail List Content-Type: multipart/alternative; boundary="0000000000000a5c2d059be4654a" --0000000000000a5c2d059be4654a Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Thank everyone! examining a few approaches led me to these conclusions: A. linking splines in mocha and back to Ae works (until you need to update)... examining the mocha<->ae workflow, I wish mocha splines<->Ae masks workflow was better so Ross please consider these two features: 1. copy pasting Ae masks to mocha splines. this way, one could create the masks in Ae then import them to mocha and not just vice versa (I know mocha import script can do this, but I wish this could be native) 2. create Ae masks in the mocha plugin could use "update Ae masks" feature. so if I do an update to my splines in mocha and use "update Ae mattes" it would update the masks instead of creating new ones (say if they have the same name) B. using Ae's compositing options to mask different instances of the keying effect in the same layer works also and would solve my accumulating alpha issue but gets messy pretty quickly. C. converting the mattes to luma works to prevent accumulating alpha. my setup was like you said - use fill set to white, a black solid at the bottom of each layer using solid composite, then using the lighten blending and it works and gives me the max of the alpha area and not combining alpha. placing on top of the mattes an adj with set matte set to luminance to convert everything to alpha and then unmult effect gives me perfect alpha, and on top of that I place the footage with preserve underlying transparency and I get everything without precomposing :) On Fri, Jan 10, 2020 at 2:06 AM Robert Kjettrup wrote: > oh... yes. An idea to get an Max combined matte as Brian mention, is to > fill all your layers RGB with white and the transparent with black (Solid > Composite set to black), use the Lighten blend mode (works like an Max > operator) on all these layers to get your new combined B/W matte. Add an > adjustment layer over all these layers with an Unmult effect (get the > Mekajiki version), then lastly add your original layer on top over all > these set to Preserve Underlying Transparency. > > > Den fre. 10. jan. 2020 kl. 00.11 skrev Ross Shain >: > >> If you happen to have Primatte Studio (in Continuum) - this is a great >> keyer that can handle transparencies and includes the Mocha >> masking internally - so you could mix and masks layers with >> varying opacity in the one layer. >> >> When I read this thread, I was reminded of my earlier days in node-based >> compositing, where it is very useful to pipe your mattes into math comps= to >> combine and keep separate from your fill, then bring it all back togethe= r. >> Node-based systems like NUKE or Fusion really are great for this type of >> compositing. I am sure you could emulate this with AE and channel your >> masks/mattes separately in their own pre-comps and than combine it all >> together. >> >> On Thu, 9 Jan 2020 at 15:41, Robert Kjettrup >> wrote: >> >>> If the mask work is not too complicated to quickly do again in mocha, >>> then you could do all the masks with linked masks there, and via the Mo= cha >>> plugin get the animated mask paths to copy to the masks on the original >>> layers? >>> >>> but 2 antialiased edges right next to eachother will still not get you = a >>> 100% tight edge, 50% alpha + 50% alpha does not equal 100% alpha. But t= hen >>> you maybe could use the Add Alpha blendmodes to get better merge of the >>> masks. >>> This is a case where it actually would be useful to have the option to >>> render masks an un-anti-aliased (guess that is called aliased :-P) to g= et >>> 100% pixel perfect merge of mask alphas. >>> >>> I dont know if it can be used for your case, but maybe a bit of >>> rethinking of the setup and you could do the keying on 1 layer but usin= g >>> the masks as Effect Masks instead? >>> >>> Robert >>> >>> >>> Den tor. 9. jan. 2020 kl. 10.28 skrev Roei Tzoref < >>> AE-List@media-motion.tv>: >>> >>>> Thx Mike, Having a stencil from them for my RGB doesn=E2=80=99t solve = my Alpha >>>> channel problem. >>>> >>>> *Let me explain it in another way:* >>>> >>>> I am creating different mattes with semi transparent areas in the >>>> overlapping and that causes a problem because the create multiplicatio= n of >>>> the alpha information >>>> >>>> The ideal way to fix this is if the mask could be perfectly adjacent t= o >>>> one another the same way a mask on =E2=80=9Cadd=E2=80=9D and its dupli= cate on =E2=80=9Csubtract=E2=80=9D >>>> work. Only that I have more than 2 masks. >>>> >>>> On Thu, 9 Jan 2020 at 9:41 Mike Abbott wrote= : >>>> >>>>> Could you add a further copy of the original footage on top of all >>>>> your keyed layers and enable ' preserve underlying transparency' ? >>>>> >>>> >> >> -- >> >> *Ross Shain* >> Chief Product Officer / Business Development >> Boris FX | borisfx.com >> mobile: +1.646.298.6458 >> >> >> >> >> >> >> >> >> --0000000000000a5c2d059be4654a Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Thank everyone! examining a few approaches led me to = these conclusions:=C2=A0

A. linking splines in mocha and back to Ae = works (until you need to update)...=C2=A0 examining the mocha<->ae workflow, I wish mocha splines<->Ae ma= sks workflow was better so Ross please consider these two features:
1. c= opy pasting Ae masks to mocha splines. this way, one could create the masks= in Ae then import them to mocha and not just vice versa (I know mocha impo= rt script can do this, but I wish this could be native)
2. create Ae mas= ks in the mocha plugin could use "update Ae masks" feature. so if= I do an update to my splines in mocha and use "update Ae mattes"= it would update the masks instead of creating new ones (say if they have t= he same name)=C2=A0=C2=A0

B. using Ae's compositing options to m= ask different instances of the keying effect in the same layer works also a= nd would solve my accumulating alpha issue but gets messy pretty=C2=A0quick= ly.

C. converting the mattes to luma works to prevent accumulating a= lpha.=C2=A0 my setup was like you said - use fill set to white, a black sol= id at the bottom of each layer using solid composite, then using the lighte= n blending and it works and gives me the max of the alpha area and not comb= ining alpha. placing on top of the mattes an adj with set matte set to lumi= nance to convert everything to alpha and then unmult effect gives me perfec= t alpha, and on top of that I place the footage with preserve underlying tr= ansparency and I get everything without precomposing :)




On Fri, Jan 10, 2020 at 2:06 AM Robert Kjettrup <AE-List@media-motion.tv>= ; wrote:
oh... yes. An idea to get an Max combined matte as Brian mention,= is to fill all your layers RGB with white and the transparent with black (= Solid Composite set to black), use the Lighten blend mode (works like an Ma= x operator) on all these layers to get your new combined B/W matte. Add an = adjustment layer over all these layers with an Unmult effect (get the Mekaj= iki version), then lastly add your original layer on top over all these set= to Preserve Underlying Transparency.


Den fre. 10. jan. 2020 = kl. 00.11 skrev Ross Shain <AE-List@media-motion.tv>:
If you happen to ha= ve Primatte Studio (in Continuum) - this is a great keyer that can handle t= ransparencies and includes the Mocha masking=C2=A0internally - so you could= mix and masks layers with varying=C2=A0opacity=C2=A0in the one layer.=C2= =A0

When I read this thread, I was reminded of my = earlier days in node-based compositing, where it is very useful to pipe you= r mattes into math comps to combine=C2=A0and keep separate=C2=A0from your f= ill, then bring it all back together. Node-based systems like NUKE or Fusio= n really are=C2=A0great for this type of compositing. I am sure you could e= mulate this with AE and channel your masks/mattes separately in their own p= re-comps and than combine it all together.=C2=A0

On Thu, 9 Jan 2020 at 15:41= , Robert Kjettrup <AE-List@media-motion.tv> wrote:
If the mask work is not t= oo complicated to quickly do again in mocha, then you could do all the mask= s with linked masks there, and via the Mocha plugin=C2=A0get the animated m= ask paths to copy to the masks on the original layers?

b= ut 2 antialiased edges right next to eachother will still not get you a 100= % tight edge, 50% alpha=C2=A0+ 50% alpha does not equal 100% alpha. But the= n you maybe could use the Add Alpha blendmodes to get better merge of the m= asks.
This is a case where it actually=C2=A0would be useful to ha= ve the option to render masks an un-anti-aliased (guess that is called alia= sed :-P) to get 100% pixel perfect=C2=A0merge of mask alphas.
I dont know if it can be used for your case, but maybe a bit of= rethinking of the setup and you could do the keying on 1 layer but using t= he masks as Effect Masks instead?=C2=A0

Robert


Den tor. 9. jan. 2020 kl. 10.28 skrev Roei Tzoref <AE-List@media-moti= on.tv>:
<= div>
Thx Mike, Having a stencil from them for my RGB doesn= =E2=80=99t solve my Alpha channel problem.=C2=A0

Let me explain it in another way:

I am creating different mattes with semi transparent ar= eas in the overlapping and that causes a problem because the create multipl= ication of the alpha information

The idea= l way to fix this is if the mask could be perfectly adjacent to one another= the same way a mask on =E2=80=9Cadd=E2=80=9D and its duplicate on =E2=80= =9Csubtract=E2=80=9D work. Only that I have more than 2 masks.

On Thu, 9 Jan 2020 at 9:41 Mike Abbott <AE-List@media-motion.tv>= ; wrote:

Could you add a further copy of= the original footage on top of all your keyed layers and enable ' pres= erve underlying transparency' ?



--

Ross Shain
Chief Product Officer / Bus= iness=C2=A0Development
Boris FX |=C2=A0borisfx.com
mobile: +1.646.298.6458








--0000000000000a5c2d059be4654a--