[Bf-committers] multi-resolution compositing and preview (was: render percentage and compositing...)

[David Jeske]

On Thu, Jun 27, 2013 at 11:17 AM, Sergey Sharybin <sergey.vfx at gmail.com>wrote:

> > Currently render-percentage and the compositor is very broken. It only
> > "works" if you are only compositing 3d renders. If you add any video or
> > image sources, the output is very very broken and requires manually
> > adjusted scale nodes after every image/video source.
> >
>
> Percentage is another story, let's don't mix it into current question ok?


I understand and agree -- let's talk only about multi-resolution
compositing preview discrepancies for now and remove render-percentage and
scene-render-resolution. I can do the coding, if I can only understand what
is broken! :)  Perhaps later I can look at render-percentage.

When multiple resolution inputs are involved (even all images, no 3d
render), I'm seeing the following issues...

1) The composite-node preview-thumbnail does not match the F12-render output

2) No way to backdrop something that matches F12-render-output, because a
viewer sees the data uncropped by render-resolution.

3) discrepancies between the preview node-thumbnails and final-render
node-output, due to differences in cropping and centering during preview vs
final..

Here is a screenshot which I think will help this discussion...

http://www.pasteall.org/pic/54391

This shows issue #1, where none of the node-previews actually look like the
F12-render output.

Is it safe to change only the composite node thumbnail to always match the
F12-render output?

Well, kind of yes. At least in case viewer is connected to the same node as
>  Composite Output. In other cases you still could want to see original
> image.
>

I see, that is tricky. I think it's confusing for the viewer to behave
differently based on whether or not composite output is connected to the
same output. However, it is desirable to be able to have a backdrop which
matches F12-render. Here are two ideas, do you like either of them?

(a) make the composite node also a "viewer", in that you could click on it
and have the backdrop pull from it. (this backdrop should 100% match
f12-render output)

(b) give the composite node an "output", which could be hooked to a viewer
to backdrop the composite/f12-rendered output.

I prefer "a", because it seems strange to me that the composite node can't
be a backdrop, and because it seems simpler. However, either seem fine. Any
reason not to do one of these?

..... on to #3... Here is a screenshot...

http://www.pasteall.org/pic/54401

In this screenshot you can see another discrepancy. The add and mix nodes
have the same inputs, but their thumbnails are not the same size or aspect
ratio. Swapping the inputs on both nodes changes their thumbnail
size/aspect-ratio but they still don't match. Is there any reason for this
to be the case?