Antlab wrote:
Exactly

Maybe you use that sort of attention called "selective", let'see:
-for you the success of Microsoft is ONLY related to DOS and text based appilcations (I use your exact words), for the rest of the world the history did not stop in 1990, and the substantial monopoly of Windows (GUI here, attention) is a well known fact.
You not only seem to be extremly stupid (or else you would have known what I told you several times and I am really getting fed up with it), you are also a liar because I've never written anything that you state. It's not even close to what I wrote! Actually you are twisting words to the opposite without facing the facts I named you. You didn't get the whole point at all and I tried it quite politely in the beginning. So get your head out of your rear before making false quotes of things I obviously have not said.
- your two images are VERY interesting. In the second, about Maya, you ONLY look at the far right, an impartial observer should ALSO notice, well...
entire rows and one column full of.... ICONS, for every sort of function (those designed by Azrael are nicer though

). And what I wrote some posts ago? That using icons to access more detailed panels is more rational respect to the famous enormous matrix of text buttons directly on the screen. So, if you calm down and look with more lucidity at your own examples, you could easily agree with my positions (I think quite similar to those of Sabrina and many others).
Again: What you call an "enourmous matrix of text buttons" is exactly the same that is on the far right of Maya or any other package I know of - even the layout is similar though the Blender widgets are quite different in their appearance. So if you are comparing things then they should be of the same kind ("My lear jet moves a lot faster than your living-room standard lamp!")! I even showed you that (visually) and you still don't understand. It won't go away even there would be 1,000 icons on the screen! My conclusion is now that you have never used a 3D package before and you don't know what you are talking about! To the icons I already have made a proposal based on existing code which Matt Ebb has already written. If you actually would be able to read then you certainly would have understood it. Azrael got it, I'm sure and it seemingly made sense to him. Anything else that you write does only show an extreme lack of knowledge, lots of ignorance and plain stupidity: So please simply shut up and do something useful!
. o O (We need an ignore list on this forums, geez!)
---
Now to something more productive (I really had the hope that this thread could have become that - so I don't give up):
I made a few little tests on efficiency and economy in the usage of 3D modellers. It's something more for people who like indices and statistics but perhaps interesting. I for my part was surprised with some results.
I used following programs with the following types of usage (all free software - I don't have any Windows installation anymore - so I couldn't install the last Maya version I own which also is quite old - I used K-3D instead which uses similar concepts):
K-3D - using only icons + using the shortcuts
Wings3D - haven't used any shortcuts because I haven't set up good shortcuts
Blender - traditional usage + using the manipulator widget
The tests consisted in creating a cube, doing some few modifications like extrusion and some basic transformations like translation, scaling and rotation. For simplicity I only worked in face selection mode on all programs. But this doesn't matter because selections are quite identical on all. The efficiency was calculated using a simple GOMS model (like described in "The Human Interface" - thanks - I had a reason to get it out of the shelf again). So I don't take ergonomics, user fatigue or other factors into account. The same goes for the time invested in learning or the fact that the programs have different complexities - though it doesn't matter in this simple test.
The overall efficiency (for the same simple modelling task, of course) resulted into the following (rounded):
K-3D using the icons: 30%
K-3D using shortcuts: 47%
Wings3D: 30%
Blender: 55%
So seemingly not a big difference between Blender and K-3D. But this was a very small test and probably not representative. So I decided to analyse the basic actions which are the most used - there I had the first surprise:
K-3D using the icons: 36%
K-3D using shortcuts: 71%
Wings3D: 36%
Blender traditional usage: 100%
Blender using the widget like in K-3D without shortcuts: 36%
Blender using the widget with shortcuts: 60%
This means that Blender only needs the minimal information that is required to perform simple tasks like transformation, scaling, rotating and the like. These actions are actually the most used, which means that the limit for an overall efficiency would be 100% - though never reaching it. The same would apply to the other overall efficiencies. This was a big surprise for me because anything over 60% is good and I thought that 100% would actually be impossible.
So I tested what could affect the overall efficiency. The result for adding a mesh cube (100% would be a single keypress which would be the minimal amount of user action that accomplishes the task):
K-3D if the tab is visible: 63%
K-3D if the tab is hidden: 32%
Wings3D: 39%
Blender using the spacebar menu: 16%
Blender using SHIFT-A: 24%
This was the second surprise regarding Blender. Though it could be raised up to 32% (25%) if the NUMKEYs would affect menu item selection. Using the menubar would even be worse for the user. So it would make sense to have an object palette for people who are constantly adding new objects into the scene. But I guess that this doesn't happen that often to justify less space on the screen. But it's clear information.
Then I took a look at the mouse movement because it is a factor in efficiency that I did not take into account. I used Kodo to get the way the mouse pointer went. This is not really accurate but should be enough to show the tendencies. I calculated some kind of index which show the relations between the best value and the other values (1 is the most economic, 0 the worst):
K-3D only using icons: 0.42
K-3D using shortcuts: 0.75
Wings3D: 1.0
Blender traditional usage: 1.0
So Wings3D and Blender need quite exactly the same amount of mouse movement and K-3D needs more. This means that Wings3D and Blender are more economic than K-3D. Wings3D and Blender work similar while performing transformations: There is no need to move the cursor to the widget and point exactly on an axis (or the midpoint) of the widget and dragging it around. This leeds to less mouse movement. Using Blender the same way K-3D is used (means "using the widget") would decrease the economic factor to the same amount. In addition there is no need to hold the mouse button pressed with Wings3D and Blender which would certainly fatigue the user after a few hours resulting in less accuracy and less speed - or even a strained and injured arm.
That was it. I certainly will do further investigation when I've got the time for it.
What does it mean for a future interface and for migrating from another package to Blender? The Blender interface is extremly efficient in the way it works. This is the biggest point for a migration. It was a inhouse software with the most possible productivity in mind. It is solid, reliable and mature - though not easy to learn. The biggest obstacle in a software migration is mostly the need to (re-)learn the new program (relearn in the way that people who master some software have the notion that they already know everthing - which is natural but often not true). So one point would be to minimize the need to relearn the already known. This can be achieved by giving the possibility to adapt the program to the habits of the user - means: A first step has been already made with the transformation widget. An absolute must would be to change keyboard layouts. This could be either used for setting up a mimic of another interface with a loss in efficiency or for just assigning shortcuts to the traditional way Blender works and having a 100% efficiency on most-used basic operations. This is on the way with the event system refactor. A second thing to be done could be having (like in Matt's proposal) the possibility to create toolpanels inside a view and give some setups for several tasks/ views along with Blender, including for example basic modelling operations like deletion, extrusion, mirroring, knife tool/ loop cut and the like - and in addition a panel with some basic objects since the analysis has shown that having icons for that would be more efficient and also users of other packages might need them. For the latter I would remain only on mesh objects because anything else isn't used that often. If someone needs it then the possibility to add it still remains. This also avoids clutter in the viewport. The advantage of doing it this way instead of creating a new window type is clear: When maximizing a 3D window the panels are still there and can be used. These actions taken would reduce the need for basic instruction quite a lot as it would reduce the need to break with habits (the way of doing things or screen layouts for example) and so making a migration a lot easier. In addition they wouldn't break the common Blender workflow and keeping the efficiency and economic mouse usage and could even bring a better way of working into existing workflows of people that haven't used Blender before.
Rests the question about existing data and integration with other tools which also counts a lot. But this isn't a interface issue.