regressions transform tool

[Scott Petrovic]

Thanks for checking that out Tim. Just to keep this conversation in scope
for everyone, we are referring to the scale and position sliders for the
free transform tool.

For the position that I updated based off our conversation, it does not use
relative position. It still uses absolute position x/y values. It just
calculates the minimum and maximum values for the slider after it knows
what the current position is. It uses the current position as the center.
This concept might be too clever for people to figure out if it it doesn't
make sense. Position would probably be better as a numeric input.

The scale maybe the same. You mentioned the use case of flipping the
selection horizontally or vertically (-100%). That seems nice. Maybe we
could even add a couple buttons next to the scale inputs to automatically
do that. That seems somewhat common and would eliminate typing.

Technically we can do whatever, but I think it would be the best to design
things around the most common use cases for inputs. When painting, when is
exact pixel control important? I never use the position and scale sliders
because I have never had a use for them while painting. I always think
visually while painting, so I always use the handles for everything.

Scott

On Wed, Nov 12, 2014 at 3:48 AM, Timothée Giet <animtim at gmail.com> wrote:

>  Hey Scott,
>
> Thanks, I gave a try to your last changes, but it still is a big usability
> regression to me..
>
> As you say, " The sliders are faster, while the input allows for greater
> control.", thing is usually when needing fast transform, one would use
> handles rather than the docker, and only use dockers numerical input for
> precise transforms. (at least that's how I use it). So I don't really see
> the point in making controls in the docker faster if it makes them (very)
> restricted compared to handles. It looks prettier at first sight, but it is
> actually less usable when thinking about what they are used for.
>
> The sliders min/max are ok for a small number of actions, but make any
> out-of-range transform much more tedious as they have to be done in several
> steps, and doing boring calculations to split a big transform in several
> ones, or rely on imprecise transform (which is not an option most of the
> time when I need precise transforms).
> Also changing position to relative values instead of absolute makes it not
> really possible anymore to place directly a selection at a precise pixel
> position in the picture.
>
> I forward the discussion on the mailing list, so we can have more opinions
> on this problem.
> (to the list people: read previous emails below for context ;) )
>
> Timothée
>
> Le 12/11/2014 01:16, Scott Petrovic a écrit :
>
>  hey Tim,
>
>  I modified the the tools a bit and commited them to master for you to
> check out.
>
>  For the position
> -----------------------
> The position limits now change based off the current position. This is
> much nicer.
>
>  For the scale
> ------------------------
> I made it so you can add negative values up to -100%. You're 'base' is at
> 100% now since that is that is what you should start out at. This makes
> flipping something vertically or horizontally much easier. If you need
> really large scaling, you can always use the handles. Let me know if this
> is better or if you have a different idea.
>
>  For typing in values outside the slider limit
> -------------------------
> I don't see any way currently to programmatically allow values greater
> than the slider limits.  If the slider only allows up to 300, you can't
> type in 350. There might need to be some compromise with this. The sliders
> are faster, while the input allows for greater control.
>
>  let me know your thoughts on what I have done.
>
>  Best,
> Scott
>
>
>
>
> On Tue, Nov 11, 2014 at 1:30 PM, Timothée Giet <animtim at gmail.com> wrote:
>
>> Hi Scott,
>>
>> I just noticed a bad regression in transform tool caused by your UI
>> changes:
>> now it is not possible anymore to enter negative values for scale, which
>> is very useful when needed.
>>
>> In my case here I needed to use scale -100% to mirror an object
>> precisely, which I do quite often, and discovered it doesn't work anymore.
>> Of course it still works when grabbing the anchors, but I need to can enter
>> numeric values for precise transformations.
>>
>> Please fix this by moving to 0% at the center and allowing negative
>> values on the left side, like you did for position.
>>
>>
>> Also, on the same topic, I really think limiting position values to
>> 3000px is not enough.. also ideally, if techically possible, it should be
>> possible to enter manually a value that's above or below the slider limits.
>>
>> Cheers,
>> Timothée
>>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kimageshop/attachments/20141112/1b778bd9/attachment.html>