Hi Leonard,
Leonard Mada wrote:
Hi Armin,I have expanded the ideas surrounding some of your concerns. Indeed, there is NO method perfect for every situation.
Hmmm. Maybe there is one and we still have to find it. That's what i hope for.
Before presenting my new ideas, lets get a step back and see, where is this *selection* useful. When I select a number of objects, there are a number of common scenarios:1. I wish to move the selection somewhere:- border match: I wish to move the selection in such a way, that the border of one object is matched to that of a different object -colour match: the move should position one coloured object from selection in apposition to a specific colour of a non-selected object2. I wish to make some global changes on all objects (or other possibilities) - won't be discussed hereSo, in some situations I might want to have the border clear, while in others I wish the best view of the colours. In the first scenario, a border might be very counterproductive, while in the second, even an overlay will make colour finding/ adjusting/ visualizing a hard thing to do. These issues should always be kept in mind, when deciding how the selection should look like.
You can simplify this to one case: The User makes a selection to transform object properties. This may be geometrically (move, scale, transform somehow) or graphically (line, fill, shadow, transparence, all...). Since simple things are always better (to understand and to handle) i would prefer to not handle selection for different cases (and have to guess what the user wants to do next) but in a single case. I would only make an exception if it is really unavoidable.
Armin Le Grand wrote:> method I: solid/ dashed borderCON: With non-dashed, the current line style (if hairline) will not be WYSIWIG -> hard to changeCON: When hairline, hard to see at all (if not at least animated) That's why i extended the line width and pained it transparent.I would like the plain (little bit thicker, 0.04'') line. I would use this method on objects where the border is non-relevant. When the border is relevant, NO border-method will be optimal.
You do not know if the border will be relevant. You cannot guess if the user wants to work with the border or something else. Forcing him to choose some 'mode' and thus giving information about his attentions makes things unnecessarily complex.
Leonard Mada wrote: > 1.1 SPECIALIZATION of BORDERS: > - see the 3D effect I created (in my previous post) > - DO NOT draw a complete border, BUT ONLY 40-70% of the borderDo You talk about transparence or dashed? I already offered those.NO. Draw a normal border, BUT not covering the border on its whole length. Draw it just over 40-70% of the border-zone, that is some of the border will remain visible as is. It might be useful, if I need to position 2 objects based on their borders, yet other methods are too complex (complex objects). I created this in my test case as if dropping a shadow (3D effect), but it is really about drawing only a portion of the border, NOT on the whole length.
So which part to leave off? If i understand You right here, it's again something to 'guess', here which 60-30% to leave out. This is not a deterministic approach and will need some KI ('guessing code') again?
> method III: arrows to selected objects; more complex> - one can more easily distinguish the boundaries of the objects because> they are NOT overlapped by borders or ghostedPRO: Maybe a good way to do something with the whole selection, e.g. grab the point where all the arrows emerge and drag. There may be more useful. CON: There may be technical paintings which already have a lot of arrows, not good to distinguish CON: The arrow head is intended to point to the center of the selected object. Think about a shape in the form of a big "C" or "O" -> pointing to something completely different. Finding a good point for the arrow head will be hard (some KI ? :-))Here is where I developed a nice solution. See the appended drawing, where I used some semicircular lines around the original line / border to draw attention to the selection. It looks really good. This is how you would do a *professional drawing* anyway. There remains the limitation of technical paintings with a lot of arrows, but as I said, for every method there are technical limitations.
I looked at the drawing.Question to II: Why is it possible here to 'place selection point' ? I thought, the selection point is the common point in a multi-selection and lies in the center of gravity? If this is the case, You cannot 'place' it here as needed.
This leads to another problem: In single object selection, the center of gravity and the selection point (arrow head) would be the same point -> only a point to visualize a single selected object?
Question to IV: How is the position for the circle on the rectangle 'guessed' ? How is it positioned? This also seems not deterministic to me.
All in all this leads to a lot of special cases for selection handling. It's not only to create the overlay objects for visualizing the selection, it's also each time another specialized 'controller' to handle inputs. Makes things more complicated and is no uniform handling.
> 2. FILL: change fill colour; there are various flavours, eg.I would generally not change object props at all, may it be line or fill style or something else. I think we should agree in keeping the object visualisation as it is, keeping WYSIWIG. Let's look at the selection as extension of that visualisation -> Overlay. So i would prefer to 'Overlay' with something transparent or 'checkered' or so where the WYSIWIG object shines through.Even with overlay, the colours are strongly distorted. So, if the user really needs the colour, this is as bad as a full fill.
No, i think this is formulated too universal. If You use something merged with a decent transparency, You still have a good impression of the object. But i agree here that this is an argument to prefer fat outlines over filled overlays.
For very complex selections, having a full fill (*OR REVERSE THIS*) is quite powerful. I used it a lot in Photoshop (although for more complex drawings).
> 4. SELECTION MASK (ala Photoshop)I would not offer the user another kind of view, the selection-mask-view. There are already enough views, for many users OOo still looks complex. We putted (and put) a lot of work to make things easier, so i dont think there is demand for a special view for working with the selectionPlease let me tell you, that selection *MASKS* are one of the most complex things in drawings. In Photoshop, there are specialized plugins (quite expensive ones) to allow different masks. *You never have enough mask views!* simply because there is always a situation where all existing masks fail and working there is a hell. (and sometimes they fail grandiosely)
As CL already said, we are not a pixel-oriented application. In those, it is clearly necessary to work interactively with masks to change their alpha and other things to get 'unsharp' selections with alpha support on pixel base.
We are talking about a clearly distinguishable selection of vector objects.Another aspect is that PhotoShop offers that more complex mask methods for professional users, but definitely tries to hide that complexity from the casual i-use-it-from-time-to-time user. That hints to not offer complex, specialized modes for selections for Joe-User if not needed.
Just my two cents, Leonard Mada
Thanks for that ideas and specialisations, Leonard. They are welcome and lead to more refined aspects, like the 'prefer fat outlines over filled' thing.
I took Your example from the last posting (the hard case wit the hidden, selected circle) and experimented with it with my suggested selection methods. I will attach the document (rgnf3, page2).
Please take a look if You find time and comment. I think 'fat transparent outline, merged' is the best compromise up to now.
- it's uniform for all selections (filled and non-filled) - it solves the problem of hidden objects- it does as much WYSIWIG as possible (filling is not disturbed, line still visible. Line fill mode is less important as area fill, anyways)
- it can be created deterministically- it supports different selection colors -> high contrast support, system selection color support
- supports big and very small objects- the selection generates attention and a good overview on the first look (this point is opinion-based :-))
I think the ideas get too complex when they need specialisations. Maybe we should concentrate to a more abstract view and shed some light on completely new ideas (if we find them).
I want to ask all people reading this to post their ideas. This will raise possibilities to find something completely different that might be better. Dont' be shy!
------------------------------------------------------------------------ --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
-- Greetings, Armin Le Grand [EMAIL PROTECTED]
rgnf3.odg
Description: application/vnd.oasis.opendocument.graphics
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
