Improve transform display in SceneInspector

[johnhaddon]

Display the world matrix and the local matrix. Possibly also show decompositions into translate/rotate/scale?

[johnhaddon]

What does everyone think about this one? We're now showing the world matrix and local matrix, but not any decompositions into translate/rotate/scale. While I think they would be handy, I think they'll also lead to bug reports like "scene inspector rotate doesn't match node editor rotate", where the decomposition of the matrix doesn't give the same result as the rotation that went in to it. Do we omit them on those grounds, or can we do something in the UI to make it clear that this is a decomposition, and the original values are not always to be expected? Any thoughts/ideas?

[andrewkaufman]

@carstenkolve @edmondeng @japmohican @thiasbxl @bentoogood

Are any of you really keen on the decompositions? I think it's fine as is, but maybe I'm too used to looking at matrices...

[bentoogood]

I don't think you'll find many lighters or td's who are so comfortable
reading matrices so as to be able to isolate the r/t/s at a glance.
I think I'd find them useful.

With regard to john's point - I'm not totally sure I follow. is the concern
that people might not get that the displayed r/t/s are the result of
concatenated transforms?

On Wed, Jul 23, 2014 at 5:05 PM, Andrew Kaufman notifications@github.com
wrote:

@carstenkolve https://github.com/carstenkolve @edmondeng
https://github.com/edmondeng @japmohican https://github.com/japmohican
@thiasbxl https://github.com/thiasbxl @bentoogood
https://github.com/bentoogood

Are any of you really keen on the decompositions? I think it's fine as is,
but maybe I'm too used to looking at matrices...

[image: xforminspect]
https://cloud.githubusercontent.com/assets/3979682/3675872/606021da-1282-11e4-8037-61ca7add041a.jpg

—
Reply to this email directly or view it on GitHub
#896 (comment).

ben tooogood
vimeo https://vimeo.com/user9291410 | linkedin
https://www.linkedin.com/in/bentoogood

[andrewkaufman]

I think his concern is that you might set one rotate value on the plugs, and see a different rotate value in the diff, because matrix decomposition is non-deterministic.

[johnhaddon]

The simplest example is when a rotation in a single axis goes past 360 - there's no representation for that in the matrix, so 361 degrees would decompose the same as 1 degree. So what you see in your scene inspector is unlikely to be what you typed in to your node editor.

I do think having translate/rotate/scale would be handy, I'd just like to find a way of presenting it that doesn't lead to unfulfillable expectations.

[bentoogood]

got you.

i'd say given that people should understand that the contents of the scene
inspector is a state dump resulting from lots of combined processes (both
by the user and by the gaffer internals), that sort of mismatch would be
expectable and understandable... if documented.

and maybe the labels to the values could be something like 'resultant
translate' 'resultant rotation'... (bit of a mouthful but some indicator
there might be handy)

?

On Wed, Jul 23, 2014 at 5:22 PM, John Haddon notifications@github.com
wrote:

The simplest example is when a rotation in a single axis goes past 360 -
there's no representation for that in the matrix, so 361 degrees would
decompose the same as 1 degree. So what you see in your scene inspector is
unlikely to be what you typed in to your node editor.

I do think having translate/rotate/scale would be handy, I'd just like to
find a way of presenting it that doesn't lead to unfulfillable expectations.

—
Reply to this email directly or view it on GitHub
#896 (comment).

ben tooogood
vimeo https://vimeo.com/user9291410 | linkedin
https://www.linkedin.com/in/bentoogood

[johnhaddon]

Cool - I'll stick it in then. Any other thoughts on naming? "Resultant rotation", "Rotation component", "Extracted rotation", "Rotation from matrix"?

[johnhaddon]

For what it's worth, I think my vote would be for "Rotation component"...

[yutapple]

Maybe improve this WHEN Ben is back at IE?
As serious as I am about above, I do agree with Ben that as long as it explains somewhere and we all are aware then I think it is fine :) As far as naming goes...I like "Rotation from matrix" but I don't really have strong preference.

thanks!

[bentoogood]

'Rotation from matrix' stood out to me too actually.

if you call it that, I will come back to IE for sure!

On Wed, Jul 23, 2014 at 5:41 PM, yuta shimizu notifications@github.com
wrote:

Maybe improve this WHEN Ben is back at IE?
As serious as I am about above, I do agree with Ben that as long as it
explains somewhere and we all are aware then I think it is fine :) As far
as naming goes...I like "Rotation from matrix" but I don't really have
strong preference.

thanks!

—
Reply to this email directly or view it on GitHub
#896 (comment).

ben tooogood
vimeo https://vimeo.com/user9291410 | linkedin
https://www.linkedin.com/in/bentoogood