On 9 Apr 2002, Sven Neumann wrote:

> Someone wanted a way to a enter zoom value numerically, so we might want 
> to use an entry/spinbutton instead.

Seems to me that an entry/spinbutton has all the advantages of the other 
ui proposal with the extra advantage that you can quickly and easily go to 
magnifications other than 100%.
> - A split mode would be handy:
>     -------------------
>     |        |        |
>     | Before |        |
>     |        | After  |
>     |        |        |
>     ------------------- 
>   Alternative setups like top/bottom, diagonal split have been
>   suggested. Difficult to get this integrated into the GUI w/o 
>   cluttering it with buttons. Could be in a right-click menu, but
>   it would probably not be found there.

It seems to me that the best way would be to have a "paned" view. Since it
doesn't seem useful to have the pane in any other positions, the pane bar
would "click" to the mid-point and to the left edge.  I could even make a 
case for clicking to the right edge (showing the before image).

Also, since left-to-right language speaker's eyes tend to move to the 
upper left and lower right corners of a dialog, it would be a little more 
usable to have the "after" (which is looked at more frequently) on the 
left and the before on the right (assuming the preview is placed on the 
left top as it is in all of the plugins I can think of).  Not a big deal 
either way, though.

> - Alternative preview in image window would be nice to have.
> - Provide an API that allows the plug-in developer to use the same
>   function for manipulating the image as well as the preview. The
>   preview would have to provide a drawable API and pixel-regions
>   etc. in order to achieve this goal. 

There is a serious problem here: what if two plug-ins are open at the same 
time and want to draw on the same image?  We wouldn't just need tile-level 
locking but layer or image-level locking as well, and the preview widget 
would have to gracefully fall back or force the other plugin to give up 
its hold on the display. You could run into serious UI issues here.

> [pie menus]
It's really hard to design a useful pie menu for a large number of menu 
items, especially in this case where you can add and remove items by 
adding and removing plugins.  Perhaps a pie-menu enthusiast could give us 
a mock-up.

> - Make docks scrollable.

Eeek!  Bad idea.  But several people around here have suggested to me that 
image windows be dockable.  Honestly, I can't see the point in that, but 
there is no harm either, and they seem to think it's more convienent to 
have everything in one big window.

> - Make layers resizable/scalable in by adding handlers to the layer
>   boundaries.

Sounds good, but I would make it a separate tool.  Otherwise they would 
get in the way -- you would go to draw in a corner area and resize the 
image instead.

> - Effect layers. Hasn't been discussed much, probably not for 1.4.

Perhaps more powerful would be to have effect layer _modes_ instead of 
effect layers. It would integrate with what gimp currently better as well. 
I don't really think it's a 1.4 thing.

> - Replace canvas XOR drawing by something nicer. We looked at some
>   commercial apps and found they all get problems if the background
>   color matches the color used to preview lines/beziers etc. GIMP has
>   this problem with gray backgrounds. 

I suggest the color that is (1-H, 1-S, 1-V) (where HSV are the hue,
saturation, and brightness of the pixel on a 0...1 scale). There will
always be some pathological case where any algorithm will produce almost
invisible lines, but this one should only be a problem if you have ugly
high-frequency high-contrast lines in your image and you are selecting
parellel to them.  (bleck!) It should be very visable in almost all other

On second thought it might be better to use (1-H, S, 1-V).  We'll have to 

> - Find a better heuristic to place pasted selections/layers. Right
>   now they appear in the center of the image which is often far
>   away from the spot where one needs them. Using the center of the
>   display could be a better choice.

Might I suggest we use the same UI that "Get It" does in the Self IDE? 
(http://research.sun.com/self/ -- original MacOS and Solaris versions,
http://www.gliebe.de/self/download.html -- Linux port)

For those who haven't had the unique pleasure of programming in Self, the
Get It function is a lot like those window managers that make you place a 
new window whenever it pops up.  But it's not like that in two important 

It's not annoying. The window manager mouse placement thing is annoying 
since windows pop-up unexpectedly at time (like netscape errors) and you 
have to go to all the trouble of placing it before you see what the 
windows is.  This wouldn't be annoying since you don't paste layers 
unexpectedly, and the layer would be displayed instead of moved in outline 

So, when you press the paste key combo or select it from the menu, it 
the pasted layer would be "stuck" to the mouse until you press the mouse 
button again. You would place the layer by moving the mouse or using the 

> Text Tool
> ---------
> - Clicking somewhere into the image and starting to type should
>   create a new text layer the size of the text's bounding box.

We could make it so that you type directly onto the image. You could set 
font, size, etc, in the tool options box.


Gimp-developer mailing list

Reply via email to