[Bf-vfx] Footage proxying

[François T.]

>
> But there are such ideas: use use proxied footage for display and
> non-proxied for footage, so solving would use "final" image sequence and
> would be accurate.


Oh ok, sure !

So maybe it should be more like a mipmaps

Sounds great


2011/6/7 Sergey I. Sharybin <g.ulairi at gmail.com>

>  Hi, François!
>
> Your explanation is quite clear.
>
> But there are such ideas: use use proxied footage for display and
> non-proxied for footage, so solving would use "final" image sequence and
> would be accurate.
>
> Problem with non-proxied footage is that it'll be scaled anyway so the
> whole frame would be visible in the screen. So maybe it should be more like
> a mipmaps -- space clip would use the smallest mipmap for displaying.
> Smallest in meaning scale factor would be the minial and there would be no
> scaling up (increasing resolution from mipmap). So, when you're zooming up
> mipmaps with higher levels (smaller resolution) would be used, but when
> you're zooming in mipmaps with lower levels (higher resolution) would be
> used.
>
> Not sure if it'll give speedup -- maybe OGL is fast enough to display 4K
> pixel buffer in realtime?
>
> François T. wrote:
>
> I'm using DJV a lot which does have this kind of on-the-fly proxy stuff you
> can set.
> Nuke also have this ability where you can set the resolution of the proxy
> you want to.
>
>  But for tracking, as a rule of thumb, we never downsample the footage,
> but track on the one which would be final (even if it's 4k and very slow
> :p). The reason for that is because downsampling could cause some slight
> shifting in the pixels which would cause a mismatch when applying the solve
> (which I've been done on the downsampling) onto the final 4K. Might not make
> a huge difference for most people, but it usually does for the one who is
> working on it, and we usually try to have a subpixel accurancy.
> Actually even the inverse (resample to create a better resolution) is
> available in most the matchmove packages (Syntheyes for instance has it). I
> only use it on really poor footages or fast moving camera, because those
> kind of resample would create thick edges which usually doesn't match the
> original footage.
> On a 64bit computer with 4/6 gb of memory, you can put a few seconds in RAM
> which is usually enough for tracking.
> I'm sorry if that doesn't sounds very clear :( . tell me if you need more
> explanation.
>
>  Cheers,
>
>  F.
>
> 2011/6/7 Peter Schlaile <peter at schlaile.de>
>
>> Hi Sergey,
>>
>> you might want to take a look at the vse-proxies branch, which I want to
>> merge very soon into trunk.
>>
>> http://lists.blender.org/pipermail/bf-committers/2011-May/032152.html
>>
>> It's working on ImBuf-level, so you could use it directly in other parts
>> of blender, if you like.
>>
>> Cheers,
>> Peter
>>
>> > I'm still working under camera tracking integration and need some help.
>> >
>> > One of goals is to be able to deal with 4K footage. But to deal with it
>> > fine, there should be some stuff which would reduce resolution to make
>> > playback faster. Such transparent resolution changing is called proxying
>> in
>> > our development stuff.
>> >
>> > So, questions are:
>> > - How proxying should happen? Should it be slider to specify quality of
>> > footage or it should be determined automatically using current zoom
>> level,
>> > space clip size and so?
>> > - Should it be able to specify tracker which footage to use: original 4K
>> or
>> > proxied?
>> > - Should user really care about this proxies or they should be totally
>> > transparent for him?
>> >
>> > --
>> > With best regards, Sergey I. Sharybin
>>
>>  ----
>> Peter Schlaile
>> _______________________________________________
>> Bf-vfx mailing list
>> Bf-vfx at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-vfx
>>
>
>
>
> --
> ____________________
> François Tarlier
> www.francois-tarlier.com
> www.linkedin.com/in/francoistarlier
>
>
> _______________________________________________
> Bf-vfx mailing listBf-vfx at blender.orghttp://lists.blender.org/mailman/listinfo/bf-vfx
>
>
>
> --
> With best regards, Sergey I. Sharybin
>
>
> _______________________________________________
> Bf-vfx mailing list
> Bf-vfx at blender.org
> http://lists.blender.org/mailman/listinfo/bf-vfx
>
>


-- 
____________________
François Tarlier
www.francois-tarlier.com
www.linkedin.com/in/francoistarlier
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/bf-vfx/attachments/20110607/350d307b/attachment-0001.htm