Simon Large wrote:
> 1. If the info pane is turned off and you select overlap mode, the blend
> slider is not shown. You have to show and hide the info pane.
I noticed that before I started on the new slider. I thought it a
little odd, but kept the old behaviour. One thing at a time >:)
> 3. The dual slider control is very clever, but in Rick's words it seems
> over-engineered. Why would you want to toggle between 2 custom blend
> values? I would be tempted to remove the two setpoints and just use the
> button to select between 0% 50% and 100% blend points (equivalent to
> up/right/down arrow keyboard shortcuts).
I really don't like the triple-toggle. It is *much* easier to detect
and compare differences when there's only two states, and I'm fairly
certain research on human sight backs me up here. Actually, the thing I
like the most about the new slider is that we can show the image at 50%
initially, which is an immediate indication to the user as to what the
overlap mode does, and then toggle back and forth between the two
extremes with the button, which are the two most useful positions on the
slider.
There's also some use to being able to toggle between 50% and one
extreme, since 50% is the closest you can get to a 'merged' image. The
up/left/down keys allow that, and so does the new control. The benefit
to giving the new control as an option is that users are far more likely
to discover what the markers do when they play with them (since there
*is* a visual cue) than they are to press random keys on their keyboard
until they find one that does something.
> 4. On a slow PC like my laptop, when dragging the blend point the system
> seizes up periodically to redraw. I would prefer the redraw to happen
> when I release the left mouse. Others may disagree - this laptop is
> about 5 years old, and it is probably not a problem on a modern machine.
Well, there is a 50ms delay before it refreshes while you're sliding.
Just pretend you're Keanu Reeves don't stop moving >:)
I dunno, 50ms is what it was at before I started looking at the code.
We could raise the timeout, but if we raised it as high as 1 second I
think it would feel unresponsive.
Getting rid of it completely isn't something I can get behind either,
because I do like to test the alpha value before committing to it. It
wouldn't be *so bad* now with the new alpha control, because you can now
simply click anywhere in the gauge to set the value, rather than having
to slide, so it wouldn't be as slow to test values out as it would with
the old slider, but even then it seems like it would be a little
uncomfortable.
We could add an option not to update the alpha until the value is
changed, but I think it would be confusing to most users. Ideally, it
would be best to simply make it not slow!
The blending will be done differently in the future when we add
different blending modes, though, so there's no point in optimizing just
yet. (Not that I would look forward to learning SIMD instruction sets
and then writing the same code five times for C++, MMX, 3DNow!, SSE, and
SSE2.)
By the way, what are the specs on your laptop?
-Rick-
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tortoisesvn.tigris.org
For additional commands, e-mail: dev-help@tortoisesvn.tigris.org
Received on Fri Feb 9 07:20:52 2007