Return-Path: Received: from mail-ig0-f175.google.com ([209.85.213.175] verified) by media-motion.tv (CommuniGate Pro SMTP 4.2.10) with ESMTP-TLS id 5398524 for AE-List@media-motion.tv; Wed, 05 Mar 2014 19:23:56 +0100 Received: by mail-ig0-f175.google.com with SMTP id ur14so8761751igb.2 for ; Wed, 05 Mar 2014 10:24:26 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:content-type:message-id:mime-version :subject:date:references:to:in-reply-to; bh=dxkePh4BEmsI67e7Xk6iGBrjO6aCdF/wkjeZUfTlQtY=; b=PtIxfGk9qvPFOAB5XUAW3eSL0tXhqzUiy+UF/+9743sT6tVU3nWXZ/gzpLdX5peUlQ +fDJeasx5Te37ooqB5cbHEvRy7w2WrPc4K7Ucd6V9mQ0RSRlRLCC71jbfYoRjvdoyCf2 6YSpbia5RcGoAmhk36/O7uKnxPXR/U9l1sGxa747uPa0PpuUxJlLuh9NaAgKqzyDZ/CW xc8tmOVwPLz4tzr0q+BxRI8SvX/JwU9RDw0SaEhPutUDHT9kkN4pMCSSIgfHpjummd4K msojJlwCsMONyhLe7HkqqkXd697kxNbkHySkjCfx89/LaW+9hKCVUbCxVUW9RHtWJa4p qgBQ== X-Gm-Message-State: ALoCoQmMXPfqkWkbMyM8PV2OEXiQt1X7sHzNv3eMRu0QHtvCQm+skaAn7xe3Jec/JFLn+TUgPest X-Received: by 10.43.145.137 with SMTP id ju9mr6087814icc.36.1394043865885; Wed, 05 Mar 2014 10:24:25 -0800 (PST) Return-Path: Received: from [192.168.1.2] (c-24-118-184-204.hsd1.mn.comcast.net. [24.118.184.204]) by mx.google.com with ESMTPSA id pn6sm15692733igb.4.2014.03.05.10.24.23 for (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128); Wed, 05 Mar 2014 10:24:24 -0800 (PST) From: Carey Dissmore Content-Type: multipart/alternative; boundary="Apple-Mail=_6DEE7510-1C1A-4F39-BF51-6D64346E6461" Message-Id: <9CB144F2-2354-4F8F-B80B-B7A2EBC0467D@imugonline.com> Mime-Version: 1.0 (Mac OS X Mail 7.2 \(1874\)) Subject: Re: [AE] Using optical flow to speed up 3D renders Date: Wed, 5 Mar 2014 12:24:18 -0600 References: To: After Effects Mail List In-Reply-To: X-Mailer: Apple Mail (2.1874) --Apple-Mail=_6DEE7510-1C1A-4F39-BF51-6D64346E6461 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=windows-1252 Optical flow works best when the foreground is moving and the background = is static.=20 Optical flow works worst when both are moving, particularly bad if the = pixel motion of foreground/background are moving opposite directions. FWIW, Carey On Mar 5, 2014, at 11:39 AM, Teddy Gage wrote: > I use AE with depth passes to simulate DoF in 3D scenes, that is a = major time saver as true 3D DoF can triple render times. In my = experience using reelsmart motion blur, a vector pass is definitely = required for anything that's going to be seen close up. Neat video is = great but there is little substitute for properly-tuned GI. Might be = worth investing more time to learn the renderer than trying to cheat it = - Vray can produce super clean results quickly if you invest the time = in tweaking the settings and using cached light maps, etc. >=20 > Overall, I think there would be too much judder for anything with = reasonable motion to it with an optical flow plugin replacing frames. = However it could be great for slow-moving or stable 3D objects like a CG = set extension or other BG replacements. Something that is a principle = onscreen like a character or CU of a vehicle etc, you may not be able to = get away with, as you could have artifacts and / or framerate issues. >=20 > In my experience RSMB, even with good vector maps, is still not that = close to photorealistic motion blur, and it operates on the same = principles. >=20 > The hours you spend looking for a simple AE cheat may end up better = spent on just renting some distributed render farm hours to render = everything at full res, prices are pretty reasonable these days. I've = had good experiences with renderrocket.com Just my 2c > -TG >=20 >=20 >=20 > On Wed, Mar 5, 2014 at 6:48 AM, Chris Zwar = wrote: > I have this vague recollection that optical flow technology (e.g. = twixtor and reel smart motion blur) was originally developed to help = speed up 3D rendering workflows. I think the idea was that 3D animators = could render out every 2nd or 3rd frame, and use an optical flow plugin = like twixtor to create the in-betweens. Even if that's not strictly = correct, the potential is there. Some of the photo-realistic projects = I've worked on have had 3D render times of 3 - 5 hours a frame. = Rendering every 2nd frame is effectively halving the overall render = time, which can be a massive saving. Even a slow After Effects plugin = is usually only seconds per frame, not hours. >=20 > So I was wondering if anyone has actually done this, or tried using = other 2D techniques to help speed up 3D rendering. >=20 > I can think of 3 ways in which slow 3D renders can be compensated for = by faster compositing techniques: >=20 > 1) Up-resing. For example rendering at 720p instead of 1080p and = scaling up the finished renders. If compositing multiple passes, only = the slow renders need to be smaller and scaled up. > 2) De-noising. Forgive me for not knowing the correct terminology, = but when rendering with global illumination it seems that there's an = overall quality setting that directly determines both the speed of = rendering and the noisiness of the image. Rendering with a lower = setting can make renders noisier, but a de-noising plugin such as Neat = Video can fix this. > 3) As stated above, rendering every 2nd or 3rd frame and using = something like twixtor to create the missing frames. A motion vector = pass would make this more accurate. >=20 > So I'm familiar with 2 of those 3 approaches - I have worked in = situations where 3D passes are rendered at smaller sizes and then scaled = up. It works very well and the time savings can be dramatic when = dealing with very long renders. 3D renders can be so clean that they = scale up very well. >=20 > I have also worked in situations where the neat video de-noiser was = used to compensate for noisy GI renders, and again the savings can be = dramatic - in some cases this can almost half 3D rendering times. Neat = video seems to be an incredible plugin, so much faster and so much = better than the AE equivalent. >=20 > So that leaves the optical flow technique as the one I haven't tried = yet. Has anyone done this? I'd love to hear from real-world examples = where people were able to render every 2nd or 3rd frame. Is a motion = vector pass essential for it to work properly? >=20 > Any other thoughts or insight welcome=85 >=20 > -Chris > +---End of message---+ > To unsubscribe send any message to >=20 >=20 >=20 > --=20 > _____________________________ > VFX & Motion Graphic Artist > teddygage dot com --Apple-Mail=_6DEE7510-1C1A-4F39-BF51-6D64346E6461 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=windows-1252 Optical flow works best when the foreground is = moving and the background is static. 
Optical flow works worst = when both are moving, particularly bad if the pixel motion of = foreground/background are moving opposite = directions.


FWIW,
Carey<= /div>

On Mar 5, 2014, at 11:39 AM, Teddy Gage <teddygage@gmail.com> = wrote:

I use AE with depth passes to simulate = DoF in 3D scenes, that is a major time saver as true 3D DoF can triple = render times. In my experience using reelsmart motion blur, a vector = pass is definitely required for anything that's going to be seen close = up. Neat video is great but there is little substitute for = properly-tuned GI. Might be worth investing more time to learn the = renderer than trying to cheat it -  Vray can produce super clean = results quickly if you invest the time in tweaking the settings and = using cached light maps, etc.

Overall, I think there would be too much judder for anything = with reasonable motion to it with an optical flow plugin replacing = frames. However it could be great for slow-moving or stable 3D objects = like a CG set extension or other BG replacements. Something that is a = principle onscreen like a character or CU of a vehicle etc, you may not = be able to get away with, as you could have artifacts and / or framerate = issues.

In my experience RSMB, even with good vector maps, = is still not that close to photorealistic motion blur, and it operates = on the same principles.

The hours you spend = looking for a simple AE cheat may end up better spent on just renting = some distributed render farm hours to render everything at full res, = prices are pretty reasonable these days. I've had good experiences with = renderrocket.com Just my = 2c
-TG



On Wed, Mar 5, = 2014 at 6:48 AM, Chris Zwar <chris@chriszwar.com> wrote:
I have this vague = recollection that optical flow technology (e.g. twixtor and reel smart = motion blur) was originally developed to help speed up 3D rendering = workflows.  I think the idea was that 3D animators could render out = every 2nd or 3rd frame, and use an optical flow plugin like twixtor to = create the in-betweens.  Even if that's not strictly correct, the = potential is there.  Some of the photo-realistic projects I've = worked on have had 3D render times of 3 - 5 hours a frame. =  Rendering every 2nd frame is effectively halving the overall = render time, which can be a massive saving.  Even a slow After = Effects plugin is usually only seconds per frame, not hours.

So I was wondering if anyone has actually done this, or tried using = other 2D techniques to help speed up 3D rendering.

I can think of 3 ways in which slow 3D renders can be compensated for by = faster compositing techniques:

1) Up-resing.  For example rendering at 720p instead of 1080p and = scaling up the finished renders.  If compositing multiple passes, = only the slow renders need to be smaller and scaled up.
2) De-noising.   Forgive me for not knowing the correct = terminology, but when rendering with global illumination it seems that = there's an overall quality setting that directly determines both the = speed of rendering and the noisiness of the image.  Rendering with = a lower setting can make renders noisier, but a de-noising plugin such = as Neat Video can fix this.
3) As stated above, rendering every 2nd or 3rd frame and using something = like twixtor to create the missing frames.  A motion vector pass = would make this more accurate.

So I'm familiar with 2 of those 3 approaches - I have worked in = situations where 3D passes are rendered at smaller sizes and then scaled = up.  It works very well and the time savings can be dramatic when = dealing with very long renders.  3D renders can be so clean that = they scale up very well.

I have also worked in situations where the neat video de-noiser was used = to compensate for noisy GI renders, and again the savings can be = dramatic - in some cases this can almost half 3D rendering times. =  Neat video seems to be an incredible plugin, so much faster and so = much better than the AE equivalent.

So that leaves the optical flow technique as the one I haven't tried = yet.  Has anyone done this?  I'd love to hear from real-world = examples where people were able to render every 2nd or 3rd frame. =  Is a motion vector pass essential for it to work properly?

Any other thoughts or insight welcome=85

-Chris
+---End of message---+
To unsubscribe send any message to <ae-list-off@media-motion.tv>



--
_____________________________
VFX & Motion Graphic Artist
teddygage dot com

= --Apple-Mail=_6DEE7510-1C1A-4F39-BF51-6D64346E6461--