[Bf-committers] Bf-committers Digest, Vol 72, Issue 38

François T. francoistarlier at gmail.com
Fri Jul 23 18:24:51 CEST 2010

answering in the mail with [FRANCOIS] tag

> > First of all the Color Wheel :
> > - it is really jerky (not smooth)
> do you mean it moves to fast? (I was using 1/10th by default but
> brecht and ton didnt like this), This could also be because big
> calculations are going on as the color is changing (threaded UI of
> some sort could solve too but probably not happening any time soon)

[FRANCOIS] it does look more like a threading issue. moving > update >
moving again. seems like the compute slow down the color wheel

> > - the brightness sidebar is affected (doesn't do that in MBL)
> Agree, we could correct the UI to compensate for this.
> Sidebar changing, for gain & gamma (not lift), MBL keeps the color's
> combine magnitude but your right that it doesn't move the value bar.
> Keeping the magnitude gives good results IMHO, it allows changing the
> color without darkening the image. So we could have this bar show the
> combine value of RGB and it would behave as you'd expect. but I dont
> see this as highly important either.

[FRANCOIS] I didn't meant any of this was highly important, I was just
giving my feedback. In color process, you usually set the brightness and
then move into this space to pick the color. makes adjustment just easier.

> - brightness only go up to 2 (I believe MBL is 10)
> This is very easy to change, I didnt get useful results with values
> above 2 so I set this, but if there is some example of how its useful
> we could change.

[FRANCOIS] it really depends how wide the range is. In case you want to burn
out your blacks you might need something about 2. also if you previously low
down your blacks and they don't get clipped (there are not supposed to right
? ) might be usefull to bring them up back. But if I found the example I'll
send it to you. That's true in the absolute we don't need it.

> > - cubic color selection is something good, but since the second
> modification
> > to it, it does feel less natural
> See what you mean, the precision is kept high in the middle rather
> then moved to the location of the mouse click. (indecently float
> buttons in 2.4x had higer precision around the initial location), I'd
> not be against this different behavior but just see this as a
> different way to get higher accuracy when selecting color.
> Having the precision in the middle is mainly because you often want to
> selected a subtle shade of white.

[FRANCOIS] true, but since the brightness is also changed as you move, you
will want to go out of the middle pretty often no ?

> - releasing the color wheel do not move the mouse to the new position,
> which
> > make fine tuning from this release point very hard
> > but I must say the finest selection with SHIFT is awesome :)
> Agree, this could be added, however with subpixel precision the second
> click would be different, unless you add compensation eg: <1 pixel
> float xy offset which is re-applied on top of the mouse location.
> Would be a nice touch, again not high priority for me now.

[FRANCOIS] at least is will be much closer, because right now it's really
frustrating to get your old color back if you are not careful

> > for the Color Balance part, I don't quite understand what as been change?
> I
> > did some test before with the exact same value on both side (AE MBL &
> > Blender's Color Balance) and I had the exact same result. As far as I was
> > setting both with the same Color space environment (linear or
> non-linear).
> > I didn't push the test as far as before, but it doesn't feel the same, I
> > can't quite understand what it is yet but it's feeling funny. Am I the
> only
> > one having this sensation ?
> The change I made was with lift, it was being added, now its like an
> inverted gain (scales color around white), with some subtle changes
> for values above 1.0 (like MBL, which tends to give more useful
> results, adding was clipping and looked ugly)

[FRANCOIS] the ugly part is exactly the same in MBL. Have you tried it with
linear color space in AE (it is not set as defaut like in Blender) ? I asked
Stu (the guy who design the application with Red Giant Software) about it,
and he told me it was normal and LGG was supposed to work with non-linear
(gamma corrected) image. So maybe the node should re-gamma correct the input
when blender is set to CM, then re-linearize the output.
Would that work ?



More information about the Bf-committers mailing list