Return-Path: Received: from mail-ee0-f52.google.com ([74.125.83.52] verified) by media-motion.tv (CommuniGate Pro SMTP 4.2.10) with ESMTP-TLS id 5108525 for AE-List@media-motion.tv; Sat, 15 Jun 2013 21:12:10 +0200 Received: by mail-ee0-f52.google.com with SMTP id c50so978312eek.25 for ; Sat, 15 Jun 2013 12:22:11 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:x-gm-message-state; bh=z45Z5ecRuifvmuOgmGTxK3L3D7V7RBh7W02G7n96EJM=; b=RceqzkdBXX1w5IfURZ/K5hJWvKbt2rlDJCDKjlpjHoG12Cegs7JJZVguShmjwMhp80 RgNIFJF39p3cUueO2Kt+oC4iXP9RBEGJrrpeB1Gi4MO+xVvkh/8dlahQVt8iJjq3yj/r OMeghqYL72bVe2dSBSaAku+rxYwqXKv9bu8da7lvzSqAektdj+x78b5UCg6mPm+XZ2D+ LB3urXiN2kjRRhgq6gVJ5pQGnbbJPH5jOdrtkrWCFiA+X/NYw5qi0kTYz2pSlv9LyqqP FzfesL4yvOOKxahvrR8z4Eo9mj4kumnhECsaIFXiyBhgS6/zLQieVEan4Srbnp5OoKkK AyRA== MIME-Version: 1.0 X-Received: by 10.15.49.9 with SMTP id i9mr8986114eew.69.1371324131210; Sat, 15 Jun 2013 12:22:11 -0700 (PDT) Received: by 10.14.7.4 with HTTP; Sat, 15 Jun 2013 12:22:11 -0700 (PDT) In-Reply-To: References: Date: Sat, 15 Jun 2013 14:22:11 -0500 Message-ID: Subject: Re: [AE] How to identify motion artifacts not present in still? From: Brian Higgins To: After Effects Mail List , Stephen van Vuuren Content-Type: multipart/alternative; boundary=001a11c34282b6a3a004df3647e0 X-Gm-Message-State: ALoCoQm3U9vgjSNQv2ea6oehGR9TurIuSmVpSpwEzqO75FRV2N4T0ZMCrz6huycrSKiezREURHWC --001a11c34282b6a3a004df3647e0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: quoted-printable Something else you might try is throwing an color correction on the whole thing as an adjustment layer to massively raise the gamma and brighten the shadows. It might help showing you where the seam is. -bH On Sat, Jun 15, 2013 at 1:40 PM, Stephen van Vuuren wrote: > David,**** > > ** ** > > Thanks for thoughts =96 I had not thought of that approach and it might a= t > least help narrow it down. I use difference mode a lot in tracing down > errors but not thought of it like this =96 good idea. Will try this today= .** > ** > > ** ** > > Right now it=92s maddening. RAM preview, see it clearly. Stop, totally > vanished.**** > > ** ** > > *stephen van vuuren* > > 336.202.4777**** > > * * > > *http://www.insaturnsrings.com/*** > > *http://www.sv2dcp.com/* > > *http://www.sv2studios.com/* > > * * > > *A film is =96 or should be =96 more like music than like fiction. It sho= uld > be a progression of moods and feelings. The theme, what=92s behind the > emotion, the meaning, all that comes later.* > > =96*Stanley Kubrick***** > > ** ** > > *From:* After Effects Mail List [mailto:AE-List@media-motion.tv] *On > Behalf Of *David Simons > *Sent:* Saturday, June 15, 2013 3:18 AM > > *To:* After Effects Mail List > *Subject:* Re: [AE] How to identify motion artifacts not present in still= ? > **** > > ** ** > > Not sure this will work, but here=92s an idea to try: duplicate the layer= , > shift it in time by 5 or 10 frames, set the anchor point with the pan > behind tool to be the center of rotation (near the pole star), set the > blend mode to Difference, then rotate the layer until it lines up exactly > with the original layer. If the stitching error is local (i.e. the rest o= f > the layout is spatially rigid), you should see just the error and rotated > error, as everything else drops to black (modulo the lighting difference > due to sunrise). If the stitching error has introduced a global crimp (i.= e. > part of the image has moved closer/further from the rest of the image), > you=92ll only be able to make one side or other of the crimp drop to blac= k. > But that should also help identify the error on a still frame, as you can > see the edge of the black. > > This whole scheme depends on the stitching result being a uniform > projection. It might not be. Good luck! > > For a moment I thought there could be a role here for the new Warp > Stabilizer VFX in AE CC shipping on Monday. It has a reversible > stabilization mode. Good for locking down some movement, doing some pixel > surgery, and then de-stabilizing again (effectively tracking the surgery > into the shot). But in this case, your artifact is already stationary > relative to the frame, so locking down the content wouldn=92t help. Hmm, > except maybe it could repair a global crimp? No, I don=92t think so. Just > having a nice chat with myself now. > > -DaveS**** > ------------------------------ > > *From: *Stephen van Vuuren > *Reply-To: *After Effects Mail List > *Date: *Thu, 13 Jun 2013 11:58:32 -0400 > *To: *After Effects Mail List > *Subject: *[AE] How to identify motion artifacts not present in still? > > I=92ve got a sequence of my film that has bedeviled me as well as the > software developer of PTGUI (Panoramic stitching software). > > It=92s a 10K timelapse sequence shot over 11 days created from 5 stitched= 5D > Mark II raw files. The stitching is done in PTGUI, however it appears > impossible to fix according three of the best stitchers out there and the > software developer. You can see it raw test from 1 day here. It=92s reall= y > hard to see on web video but watch the upper part of the image and stars > blink in and out. It=92s a thin line or zone where star positions jump on= the > bad stitch line from the panoramic warp. > > https://vimeo.com/65194332 password is stitch > > I need to see if there is better method that trial and error to precisely > find the line as you can only see it and therefore mask it while the clip > is in motion. I=92m not aware of any way in AE to mask a moving clip nor = can > think of any way to simulate it on a still. As its stars at 10k, I need t= o > be nearly 1 pixel accurate in masking the region to apply the fix > (time-remapping layers) to fix this stitch error. > > > *stephen van vuuren > *336.202.4777 > * > **http://www.insaturnsrings.com/ > http://www.sv2dcp.com/ > **http://www.sv2studios.com/ > ** > **A film is =96 or should be =96 more like music than like fiction. It sh= ould > be a progression of moods and feelings. The theme, what=92s behind the > emotion, the meaning, all that comes later. > *=96*Stanley Kubrick > * > > **** > --=20 *brian higgins | creative director Sol Design 312.706.5500 higgins@soldesignfx.com soldesignfx.com * --001a11c34282b6a3a004df3647e0 Content-Type: text/html; charset=windows-1252 Content-Transfer-Encoding: quoted-printable
Something else you might try is throwing an color correcti= on on the whole thing as an adjustment layer to massively raise the gamma a= nd brighten the shadows. =A0It might help showing you where the seam is.
-bH


On Sat, Jun 15, 2013 at 1:40 PM, Stephen van Vuuren = <stephen@sv2studios.com> wrote:

David,=

=A0<= /p>

Thanks for thoughts = =96 I had not thought of that approach and it might at least help narrow it= down. I use difference mode a lot in tracing down errors but not thought o= f it like this =96 good idea. Will try this today.

=A0<= /p>

Right now it=92s madde= ning. RAM preview, see it clearly. Stop, totally vanished.

=A0<= /p>

stephen van vuuren

336.202.4777=

=A0

http://www.insaturnsrings.com/

http://www.sv2dcp= .com/

http://www= .sv2studios.com/

=A0

A film= is =96 or should be =96 more like music than like fiction. It should be a = progression of moods and feelings. The theme, what=92s behind the emotion, = the meaning, all that comes later.

=96Stanley Kubrick=

=A0

From: After Effects Mail List [mailto:AE-List@media-motion.tv] On Behalf Of Da= vid Simons
Sent: Saturday, June 15, 2013 3:18 AM

To: After Effects Mail List
Subject: Re: [AE] How = to identify motion artifacts not present in still?

=A0

Not sure t= his will work, but here=92s an idea to try: duplicate the layer, shift it i= n time by 5 or 10 frames, set the anchor point with the pan behind tool to = be the center of rotation (near the pole star), set the blend mode to Diffe= rence, then rotate the layer until it lines up exactly with the original la= yer. If the stitching error is local (i.e. the rest of the layout is spatia= lly rigid), you should see just the error and rotated error, as everything = else drops to black (modulo the lighting difference due to sunrise). If the= stitching error has introduced a global crimp (i.e. part of the image has = moved closer/further from the rest of the image), you=92ll only be able to = make one side or other of the crimp drop to black. But that should also hel= p identify the error on a still frame, as you can see the edge of the black= .

This whole scheme depends on the stitching result being a uniform proje= ction. It might not be. Good luck!

For a moment I thought there coul= d be a role here for the new Warp Stabilizer VFX in AE CC shipping on Monda= y. It has a reversible stabilization mode. Good for locking down some movem= ent, doing some pixel surgery, and then de-stabilizing again (effectively t= racking the surgery into the shot). But in this case, your artifact is alre= ady stationary relative to the frame, so locking down the content wouldn=92= t help. Hmm, except maybe it could repair a global crimp? No, I don=92t thi= nk so. Just having a nice chat with myself now.

-DaveS

From: Stephen van Vuuren <stephen@sv2studios.com>= ;
Reply-To: After Effects Mail List <AE-List@media-motion.tv>
Date: = Thu, 13 Jun 2013 11:58:32 -0400
To: After Effects Mail List &= lt;AE-List@med= ia-motion.tv>
Subject: [AE] How to identify motion artifacts not present in still?=

I=92ve got a sequence of my film that= has bedeviled me as well as the software developer of PTGUI (Panoramic sti= tching software).

It=92s a 10K timelapse sequence shot over 11 days created from 5 stitch= ed 5D Mark II raw files. The stitching is done in PTGUI, however it appears= impossible to fix according three of the best stitchers out there and the = software developer. You can see it raw test from 1 day here. It=92s really = hard to see on web video but watch the upper part of the image and stars bl= ink in and out. It=92s a thin line or zone where star positions jump on the= bad stitch line from the panoramic warp.
=A0
https://vim= eo.com/65194332 password is stitch
=A0
I need to see if there is = better method that trial and error to precisely find the line as you can on= ly see it and therefore mask it while the clip is in motion. I=92m not awar= e of any way in AE to mask a moving clip nor can think of any way to simula= te it on a still. As its stars at 10k, I need to be nearly 1 pixel accurate= in masking the region to apply the fix (time-remapping layers) to fix this= stitch error.
=A0

stephen van vuuren
336.202.4777

<= /span>http://www.insaturnsrings.com/
http://www.sv2dcp.com/=
http://www.sv2studios.com/ <http://www.sv2studios.com= />

A film is =96 or should be =96 more like music than like fiction= . It should be a progression of moods and feelings. The theme, what=92s beh= ind the emotion, the meaning, all that comes later.
=96Stanley Kubrick
<= span style=3D"font-size:11.0pt;font-family:"Calibri","sans-s= erif"">



--
brian higgins | creative director
Sol =A0Design
312.706.5500
higgins@soldesignfx.com
soldesignfx.com


--001a11c34282b6a3a004df3647e0--