Adjustment layers for VSE!

Look at this cool commit:
http://lists.blender.org/pipermail/bf-blender-cvs/2011-May/035920.html

Revision: 36716
http://projects.blender.org/scm/viewvc.php?view=rev&root=bf-blender&revision=36716
Author: schlaile
Date: 2011-05-16 17:14:47 +0000 (Mon, 16 May 2011)
Log Message:

[PATCH] == Sequencer ==

This patch adds adjustment layer tracks to the sequencer and does some cleaning
up of the code.

What’s an adjustment layer?

Think of it as an effect track, which takes no explicit input, but alters
the output of everything down the layer stack.

So: you can add several stages of color correction with it.

And: you can even use it with metastrips to group several adjustments together.

Could be that my tutorials are going to be redundant soon :wink:

Very cool commit indeed, I remember asking Peter about this in a bug report a year or so ago when the colour balance tool in the VSE was recoded and no longer worked the same when applied to a color strip, then found a transform strip would do the job but restricted to video strip length.

And I’d imagine many people familiar with AE have asked over the years before that.

May be a curve tool will follow perhaps, a luma mask and gradient tool to make more use of it. :slight_smile:

Time to update that excellent tutorial 3point. :slight_smile:

Nice! I love to see sequencer upgrades!

sorry but What is VSE ?

Video sequence editor in blender

ouch! thankyou.

yes, welcome to the Dark Side MmAaXx…

Yellow, porting those functions from nodes would be SO cool! Curves alone would increase the value of the VSE 10 fold. (assuming that you could move values cleanly)

It certainly would, when Xat did the VSE Filter Stack patch there was discussion between Matt & Xat about the methods to combine Nodes code with VSE, new shared code for both with regard to image processing. With the isolated upgrade of Comp Nodes to Cuda/OpenCL that’s probably farther away than ever.

With the new adjustment layer it would great to be able to drop a comp node group onto a adjutsment strip as a filter, Cuda/OpenCl accelerated it’d be more useable realtime.

Being able to use numerous Node Groups with simple feed in from Image or Movie and including feed in from marked frame sequences IN/OUT points from VSE strips, then out of the node group back onto the VSE including adjustment layers would help fill the gap with lack of VSE filters, as someone would have to code those filters, when non coders could create many filters by using node groups, sort of visual filter coding. :slight_smile:

After a quick look at this Code layout http://www.blender.org/bf/codelayout.jpg posted by Ton, you can see that Nodes and VSE reside in completely unrelated places with unrelated calls to each other. :frowning:

So I see Ton’s point that, sadly, Nodes probably won’t be integrated into the VSE not even as groupings of code. Unless there is a pressing need in the new movie. But I doubt that the Foundation will need any effects work in the VSE as it’s strength is supposed to be in simple editing only.

GEGL does GPU for image processing…

Will be interesting to see the direction taken with the new movie.

Reading between the GSOC lines and the deliberation over matching/libmv, it seems that the workflow will be (as it is now) very modular. Footage in > edit > track > 3D generation > Grade > Comp > re-edit > render.

As a spoilt user of modern NLEs I love being able to do so much right there in the timeline. Alas right now the timeline itself requires much TLC before dreaming of other features.

Still a nice find there, and it will be nice to get to try this out

Footage in, edit, retime slowmo shots = VSE + Optical Flow + proxies
Frame sequences for VFX out to Nodes from VSE timeline for tracking and 3D.
Comp = Nodes / VSE
Redit / Retime slow mo shots = VSE + Optical Flow + Proxies
Primary Grade for consistency = VSE
Grade for Looks over movie as a whole = VSE
Render = VSE

:slight_smile:

As a spoilt user of modern NLEs I love being able to do so much right there in the timeline. Alas right now the timeline itself requires much TLC before dreaming of other features.

Yeah, simple editing requires a bit more than the VSE offers right now.

Underestimating the need and flexibility of a timeline approach would be a grave error.

Wait till they start looking at Timecode! Arrrrgh they will run for the hills.

hehe :slight_smile: Joking apart if this next movie is to be ‘Live Action’ then dealing with the assests from multiple cameras, multiple shots, differing camera setups/ frame rates etc is a task in itself. VSE lends itself to that, nodes don’t. :slight_smile:

Would be great if they used DSLRs (like mine) and had to solve all those h264 issues (to transcode or not, that is the question-GPU?). I must get some matte painting/compositing tutorials out of my system, Blender would be handy for that (even now).

Can’t help thinking DSLR’s would be terrible for live action due to rolling shutter, well hand held anyways, maybe on a rig it would tame it a bit.

I think your ‘transcode or not’ will be solved with Peters proxies patch you mentioned recently, with regard to playback and hopefully the code added a while ago to check for full range levels will stay, so your files should go into blender without messing with the levels.

Certainly seems to work for h264, the only codec that does stay put into blender, certainly DV and HDV gets the levels messed with.

If OpenColorIO gets implemented that could improve things, rather than using ffmpegs swscale.

I must get some matte painting/compositing tutorials out of my system, Blender would be handy for that (even now).

Excellent. :slight_smile:

I think that one of the targets for the matchmoving project is rolling shutter correction, (just like avisynth? deshaker) you shouldn’t need it on to many shots but I get the point.

Back on topic, I just tried making a LUT for my Technicolor Style footage and then correcting it. It’s great for this! However I noticed that after adding the Adjustment Strip in the middle of my clip I made the clip freeze.

I did this by stretching the Adjustment Strip’s head back to the beginning of the underlying clip. Doing this repeated the frame from the origination point. Weird huh?! I fixed it by simply dragging the whole Adjustment Strip back to the start of the underlying strip and stretching the tail back out.