[Bf-vfx] Bf-vfx Digest, Vol 17, Issue 4
David McSween
3pointedit at gmail.com
Sat Aug 23 12:10:07 CEST 2014
In a similar vein to the recent Microsoft video of hyperlapsing gopro
vission there was this thread at BA
http://blenderartists.org/forum/showthread.php?346230-spatial-temporal-(3D)-blender-stabilizer-Proof-of-Concept&p=2709869#post2709869
It seems that a projected stich of panned frames would be nice in 3d view.
Not sure its applicable to stabilising contra movement or moving objects in
the frame.
David
On 23 Aug 2014 20:00, <bf-vfx-request at blender.org> wrote:
> Send Bf-vfx mailing list submissions to
> bf-vfx at blender.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> http://lists.blender.org/mailman/listinfo/bf-vfx
> or, via email, send a message with subject or body 'help' to
> bf-vfx-request at blender.org
>
> You can reach the person managing the list at
> bf-vfx-owner at blender.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Bf-vfx digest..."
>
>
> Today's Topics:
>
> 1. Re: Bf-vfx Digest, Vol 17, Issue 3 (David McSween)
> 2. Re: Improving the 2D stabilization tool :: Demo Video
> (Sebastian K?nig)
> 3. Re: Improving the 2D stabilization tool :: Demo Video
> (Sean Kennedy)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Fri, 22 Aug 2014 20:34:02 +1000
> From: David McSween <3pointedit at gmail.com>
> Subject: Re: [Bf-vfx] Bf-vfx Digest, Vol 17, Issue 3
> To: bf-vfx at blender.org
> Message-ID:
> <CAKm_dB=
> 17BNwK7tJXdLFm6SHfKLuAXvqd22Uf4CrD4tMHrrxgA at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> This problem reminds me of sound baking to an fcurve. At least you're
> propogating an editable result. Blending between stabilisation weights
> would be neat.
> David
> On 22 Aug 2014 20:00, <bf-vfx-request at blender.org> wrote:
>
> > Send Bf-vfx mailing list submissions to
> > bf-vfx at blender.org
> >
> > To subscribe or unsubscribe via the World Wide Web, visit
> > http://lists.blender.org/mailman/listinfo/bf-vfx
> > or, via email, send a message with subject or body 'help' to
> > bf-vfx-request at blender.org
> >
> > You can reach the person managing the list at
> > bf-vfx-owner at blender.org
> >
> > When replying, please edit your Subject line so it is more specific
> > than "Re: Contents of Bf-vfx digest..."
> >
> >
> > Today's Topics:
> >
> > 1. Re: Improving the 2D stabilization tool :: Demo Video
> > (Sean Kennedy)
> >
> >
> > ----------------------------------------------------------------------
> >
> > Message: 1
> > Date: Thu, 21 Aug 2014 15:26:05 -0700
> > From: Sean Kennedy <mack_dadd2 at hotmail.com>
> > Subject: Re: [Bf-vfx] Improving the 2D stabilization tool :: Demo
> > Video
> > To: Blender motion tracking & VFX <bf-vfx at blender.org>
> > Message-ID: <COL127-W11AAA01427E05F464FC048A3D30 at phx.gbl>
> > Content-Type: text/plain; charset="iso-8859-1"
> >
> > That's all very exciting! Glad to hear you have ideas and intentions to
> > keep improving it. :)
> >
> > > Date: Thu, 21 Aug 2014 03:25:54 +0200
> > > From: prg at ichthyostega.de
> > > To: bf-vfx at blender.org
> > > Subject: Re: [Bf-vfx] Improving the 2D stabilization tool :: Demo Video
> > >
> > > On 21.08.2014 01:19, Sean Kennedy wrote:
> > > > I was up in Vancouver at Siggraph last week, so I've only gotten a
> > chance to
> > > > watch this video today.
> > >
> > > > Really nice! Very impressive work, and Blender has definitely been
> > needing
> > > > some kind of scaling stabilization since stabilizing was introduced.
> > > > I'd love to see these tools get into trunk. It's great that you took
> > some
> > > > good, real world video to demo on, video with all the problems
> typical
> > > > stabilization projects run in to.
> > >
> > > thanks, I really appreciate your understanding, and indeed, the
> > real-world usage
> > > situation can be rather insidious at times. Doing this example showed
> me
> > that
> > > it is possible to get acceptable results with what I've implemented
> thus
> > far,
> > > which makes me think it's the right moment to aim for review and
> > inclusion
> > > of this version, to get it in the hands of more people.
> > >
> > > But it also showed me that the process is still laborious and the tool
> > could
> > > support the process of stabilising footage better in several ways; I'd
> > like to
> > > follow up on these, while I really hope to get more feedback to shape
> > those
> > > next ideas. To some degree, modest additional "automagic" could help,
> > but I am
> > > reluctant to engage in developing an elaborate automatic algorithm
> which
> > turns
> > > out to be not so helpful when put under real-world stress.
> > >
> > > Bjorn (sunboy) commented already on the review page, and I will
> > definitively
> > > have a closer look into his add-on; a lowpass looks like one of these
> > ideas
> > > to be explored further -- a physical steadycam is likewise some kind
> of a
> > > lowpass after all
> > >
> > >
> > > > Are the keyframes you set in the Expected position tool panel
> viewable
> > in the
> > > > curve editor? I imagine they are, just want to make sure.
> > >
> > > In theory yes. In practice no.
> > > Let me explain. The animation system of blender is entirely generic,
> > thus you
> > > can animate (almost) all parameters found in the data blocks making up
> a
> > blender
> > > scene. Thus, if we use a persistent parameter and hook it up into the
> > GUI, you
> > > can press the "I" key and animate it through keyframes.
> > >
> > > In fact, this system turns out to be more generic than what the GUI for
> > > animation curve editing assumes. Meaning that this curve editor and
> > animation
> > > sequence handling makes some seemingly innocuous assumptions as to
> > *where*
> > > the animation data is attached -- and these turn out to be too limited.
> > > Thus, if you have animation data attached to a property which belongs
> > directly
> > > to the 3D model, then the animation editor can find this data by
> > descending
> > > top level into the current scene. Unfortunately, our animation data is
> > attached
> > > to the post processing stages of the output graph (more precisely, to
> the
> > > movie clip editing data) and from the looks the current setup of the
> > > animation editors simply don't have this possibility "on the radar"
> > >
> > > I haven't investigated that problem yet (and I'd appreciate any hint of
> > the
> > > more experienced blender coders); but off the top of my head, I can't
> see
> > > why this should not be doable.
> > >
> > > Cheers,
> > > Hermann
> > > (Ichthyo)
> > >
> > > _______________________________________________
> > > Bf-vfx mailing list
> > > Bf-vfx at blender.org
> > > http://lists.blender.org/mailman/listinfo/bf-vfx
> >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
> >
> http://lists.blender.org/pipermail/bf-vfx/attachments/20140821/7f2b1a16/attachment-0001.htm
> >
> > ------------------------------
> >
> > _______________________________________________
> > Bf-vfx mailing list
> > Bf-vfx at blender.org
> > http://lists.blender.org/mailman/listinfo/bf-vfx
> >
> >
> > End of Bf-vfx Digest, Vol 17, Issue 3
> > *************************************
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://lists.blender.org/pipermail/bf-vfx/attachments/20140822/640eda09/attachment-0001.htm
>
> ------------------------------
>
> Message: 2
> Date: Sat, 23 Aug 2014 04:23:38 +0200
> From: Sebastian K?nig <sebastiankoenig at posteo.de>
> Subject: Re: [Bf-vfx] Improving the 2D stabilization tool :: Demo
> Video
> To: Blender motion tracking & VFX <bf-vfx at blender.org>
> Message-ID: <etPan.53f7fb2a.1f16e9e8.41e at agent-cooper-3>
> Content-Type: text/plain; charset="utf-8"
>
> Hey!
> I?didn?t follow that closely yet, but found the video quite interesting!
> So yesterday I built with the patch and it works quite well. I think the
> workflow can be improved still, but I should test a bit more for that. The
> thing that is really nice is to be able to ?counteranimate? the expected
> location, which I?believe would also be what you?d do in other compositing
> apps (that have a ?canvas?). Since we don?t have a canvas in compositor
> yet, your approach seems reasonable and the clip editor also seems like a
> good place for that. UI wise it would be nice to have some kind of handle
> for that. Maybe when using stabilization it could just have a point to grab
> and move, just as with the new tweaking center for masks?
> But yeah, I should test this a bit more when I find the time.?
> Nice work so far, and thanks for putting your time into this!
>
> Cheers,
>
> Seb
> ?
> --?
> Sebastian K?nig
> K?rnerstra?e 28
> 04107 Leipzig
> Germany
> sebastiankoenig at posteo.de
> www.blendFX.com
> www.3dzentrale.com
>
> Am 22. August 2014 bei 00:26:05, Sean Kennedy (mack_dadd2 at hotmail.com)
> schrieb:
>
> That's all very exciting! Glad to hear you have ideas and intentions to
> keep improving it. :)
>
>
> > Date: Thu, 21 Aug 2014 03:25:54 +0200
> > From: prg at ichthyostega.de
> > To: bf-vfx at blender.org
> > Subject: Re: [Bf-vfx] Improving the 2D stabilization tool :: Demo Video
> >
> > On 21.08.2014 01:19, Sean Kennedy wrote:
> > > I was up in Vancouver at Siggraph last week, so I've only gotten a
> chance to
> > > watch this video today.
> >
> > > Really nice! Very impressive work, and Blender has definitely been
> needing
> > > some kind of scaling stabilization since stabilizing was introduced.
> > > I'd love to see these tools get into trunk. It's great that you took
> some
> > > good, real world video to demo on, video with all the problems typical
> > > stabilization projects run in to.
> >
> > thanks, I really appreciate your understanding, and indeed, the
> real-world usage
> > situation can be rather insidious at times. Doing this example showed me
> that
> > it is possible to get acceptable results with what I've implemented thus
> far,
> > which makes me think it's the right moment to aim for review and
> inclusion
> > of this version, to get it in the hands of more people.
> >
> > But it also showed me that the process is still laborious and the tool
> could
> > support the process of stabilising footage better in several ways; I'd
> like to
> > follow up on these, while I really hope to get more feedback to shape
> those
> > next ideas. To some degree, modest additional "automagic" could help,
> but I am
> > reluctant to engage in developing an elaborate automatic algorithm which
> turns
> > out to be not so helpful when put under real-world stress.
> >
> > Bjorn (sunboy) commented already on the review page, and I will
> definitively
> > have a closer look into his add-on; a lowpass looks like one of these
> ideas
> > to be explored further -- a physical steadycam is likewise some kind of a
> > lowpass after all
> >
> >
> > > Are the keyframes you set in the Expected position tool panel viewable
> in the
> > > curve editor? I imagine they are, just want to make sure.
> >
> > In theory yes. In practice no.
> > Let me explain. The animation system of blender is entirely generic,
> thus you
> > can animate (almost) all parameters found in the data blocks making up a
> blender
> > scene. Thus, if we use a persistent parameter and hook it up into the
> GUI, you
> > can press the "I" key and animate it through keyframes.
> >
> > In fact, this system turns out to be more generic than what the GUI for
> > animation curve editing assumes. Meaning that this curve editor and
> animation
> > sequence handling makes some seemingly innocuous assumptions as to
> *where*
> > the animation data is attached -- and these turn out to be too limited.
> > Thus, if you have animation data attached to a property which belongs
> directly
> > to the 3D model, then the animation editor can find this data by
> descending
> > top level into the current scene. Unfortunately, our animation data is
> attached
> > to the post processing stages of the output graph (more precisely, to the
> > movie clip editing data) and from the looks the current setup of the
> > animation editors simply don't have this possibility "on the radar"
> >
> > I haven't investigated that problem yet (and I'd appreciate any hint of
> the
> > more experienced blender coders); but off the top of my head, I can't see
> > why this should not be doable.
> >
> > Cheers,
> > Hermann
> > (Ichthyo)
> >
> > _______________________________________________
> > Bf-vfx mailing list
> > Bf-vfx at blender.org
> > http://lists.blender.org/mailman/listinfo/bf-vfx
> _______________________________________________
> Bf-vfx mailing list
> Bf-vfx at blender.org
> http://lists.blender.org/mailman/listinfo/bf-vfx
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://lists.blender.org/pipermail/bf-vfx/attachments/20140823/4d244c1b/attachment-0001.htm
>
> ------------------------------
>
> Message: 3
> Date: Fri, 22 Aug 2014 21:30:46 -0700
> From: Sean Kennedy <mack_dadd2 at hotmail.com>
> Subject: Re: [Bf-vfx] Improving the 2D stabilization tool :: Demo
> Video
> To: Blender motion tracking & VFX <bf-vfx at blender.org>
> Message-ID: <COL127-W31F34AE8BCC0808DDD9725A3D10 at phx.gbl>
> Content-Type: text/plain; charset="windows-1252"
>
> When you say you'd like to have a handle for animating the main camera
> moves back into the stabilized footage, I agree with you totally. It would
> even be cool to see the animation path you are creating by counteranimating
> right there in the MCE viewer. It's been very helpful in other apps when
> doing this kind of thing on a real world production.
> Another way the canvas would come in handy would be another technique
> using this kind of stabilization. A few times on shows, I've had to
> stabilize a big, panning camera move (like the footage Hermann used in his
> demo), and then create an extremely large plate of the entire scene. I
> would create a gigantic canvas, then lay out the frames in a long image,
> like a giant panoramic image. Once that's laid out, it's very easy to
> recreate the camera move manually and roto back in the moving objects in
> the scene. Not sure if I'm explaining this in a way that's understandable,
> but it would be relatively easy with these great stabilizing tools and the
> ability to set a huge canvas size in the compositor. (Actually, I suppose
> you could set up the "canvas" on an image in the 3d view and recreate the
> camera move with a 3d camera.)
> Anyway, just babbling now....
>
> Date: Sat, 23 Aug 2014 04:23:38 +0200
> From: sebastiankoenig at posteo.de
> To: bf-vfx at blender.org
> Subject: Re: [Bf-vfx] Improving the 2D stabilization tool :: Demo Video
>
> Hey!I didn?t follow that closely yet, but found the video quite
> interesting! So yesterday I built with the patch and it works quite well. I
> think the workflow can be improved still, but I should test a bit more for
> that. The thing that is really nice is to be able to ?counteranimate? the
> expected location, which I believe would also be what you?d do in other
> compositing apps (that have a ?canvas?). Since we don?t have a canvas in
> compositor yet, your approach seems reasonable and the clip editor also
> seems like a good place for that. UI wise it would be nice to have some
> kind of handle for that. Maybe when using stabilization it could just have
> a point to grab and move, just as with the new tweaking center for
> masks?But yeah, I should test this a bit more when I find the time. Nice
> work so far, and thanks for putting your time into this!
> Cheers,
> Seb --
> Sebastian K?nig
> K?rnerstra?e 2804107
> LeipzigGermanysebastiankoenig at posteo.dewww.blendFX.comwww.3dzentrale.com
> Am 22. August 2014 bei 00:26:05, Sean Kennedy (mack_dadd2 at hotmail.com)
> schrieb:
>
>
>
>
>
>
>
>
> That's all very exciting! Glad to hear you have
> ideas and intentions to keep improving it. :)
>
>
>
>
> > Date: Thu, 21 Aug 2014 03:25:54 +0200
>
> > From: prg at ichthyostega.de
>
> > To: bf-vfx at blender.org
>
> > Subject: Re: [Bf-vfx] Improving the 2D stabilization tool ::
> Demo Video
>
> >
>
> > On 21.08.2014 01:19, Sean Kennedy wrote:
>
> > > I was up in Vancouver at Siggraph last week, so I've only
> gotten a chance to
>
> > > watch this video today.
>
> >
>
> > > Really nice! Very impressive work, and Blender has
> definitely been needing
>
> > > some kind of scaling stabilization since stabilizing was
> introduced.
>
> > > I'd love to see these tools get into trunk. It's great
> that you took some
>
> > > good, real world video to demo on, video with all the
> problems typical
>
> > > stabilization projects run in to.
>
> >
>
> > thanks, I really appreciate your understanding, and indeed,
> the real-world usage
>
> > situation can be rather insidious at times. Doing this example
> showed me that
>
> > it is possible to get acceptable results with what I've
> implemented thus far,
>
> > which makes me think it's the right moment to aim for review
> and inclusion
>
> > of this version, to get it in the hands of more people.
>
> >
>
> > But it also showed me that the process is still laborious and
> the tool could
>
> > support the process of stabilising footage better in several
> ways; I'd like to
>
> > follow up on these, while I really hope to get more feedback
> to shape those
>
> > next ideas. To some degree, modest additional "automagic"
> could help, but I am
>
> > reluctant to engage in developing an elaborate automatic
> algorithm which turns
>
> > out to be not so helpful when put under real-world
> stress.
>
> >
>
> > Bjorn (sunboy) commented already on the review page, and I
> will definitively
>
> > have a closer look into his add-on; a lowpass looks like one
> of these ideas
>
> > to be explored further -- a physical steadycam is likewise
> some kind of a
>
> > lowpass after all
>
> >
>
> >
>
> > > Are the keyframes you set in the Expected position tool
> panel viewable in the
>
> > > curve editor? I imagine they are, just want to make
> sure.
>
> >
>
> > In theory yes. In practice no.
>
> > Let me explain. The animation system of blender is entirely
> generic, thus you
>
> > can animate (almost) all parameters found in the data blocks
> making up a blender
>
> > scene. Thus, if we use a persistent parameter and hook it up
> into the GUI, you
>
> > can press the "I" key and animate it through keyframes.
>
> >
>
> > In fact, this system turns out to be more generic than what
> the GUI for
>
> > animation curve editing assumes. Meaning that this curve
> editor and animation
>
> > sequence handling makes some seemingly innocuous assumptions
> as to *where*
>
> > the animation data is attached -- and these turn out to be too
> limited.
>
> > Thus, if you have animation data attached to a property which
> belongs directly
>
> > to the 3D model, then the animation editor can find this data
> by descending
>
> > top level into the current scene. Unfortunately, our animation
> data is attached
>
> > to the post processing stages of the output graph (more
> precisely, to the
>
> > movie clip editing data) and from the looks the current setup
> of the
>
> > animation editors simply don't have this possibility "on the
> radar"
>
> >
>
> > I haven't investigated that problem yet (and I'd appreciate
> any hint of the
>
> > more experienced blender coders); but off the top of my head,
> I can't see
>
> > why this should not be doable.
>
> >
>
> > Cheers,
>
> > Hermann
>
> > (Ichthyo)
>
> >
>
> > _______________________________________________
>
> > Bf-vfx mailing list
>
> > Bf-vfx at blender.org
>
> > http://lists.blender.org/mailman/listinfo/bf-vfx
>
>
>
>
>
> _______________________________________________
>
> Bf-vfx mailing list
>
> Bf-vfx at blender.org
>
> http://lists.blender.org/mailman/listinfo/bf-vfx
>
>
> _______________________________________________
> Bf-vfx mailing list
> Bf-vfx at blender.org
> http://lists.blender.org/mailman/listinfo/bf-vfx
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://lists.blender.org/pipermail/bf-vfx/attachments/20140822/04f0cac8/attachment-0001.htm
>
> ------------------------------
>
> _______________________________________________
> Bf-vfx mailing list
> Bf-vfx at blender.org
> http://lists.blender.org/mailman/listinfo/bf-vfx
>
>
> End of Bf-vfx Digest, Vol 17, Issue 4
> *************************************
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/bf-vfx/attachments/20140823/b78efb77/attachment.htm
More information about the Bf-vfx
mailing list