Re: Axes Ticks Positioning in 3D plots (FLTK)

[Ben Abbott]

On Jan 9, 2011, at 9:35 PM, Ben Abbott wrote:

> On Jan 9, 2011, at 9:27 PM, Ben Abbott wrote:
> 
>> On Jan 9, 2011, at 8:29 PM, Jordi Gutiérrez Hermoso wrote:
>> 
>>> On 9 January 2011 17:07, logari81 <address@hidden> wrote:
>>>> In the attached patch I have modified the positioning of axes in 3D
>>>> plots, in order to make it more compatible with ML.
>>> 
>>> Is this "compatibility"? I always thought of compatibility of just
>>> making sure code written for Matlab also work in Octave, not so much
>>> about superficial cosmetic differences. Will someone really complain
>>> about how incompatible Octave is because it displays ticks at top
>>> instead of bottom when you rotate to look at the bottom? I like how
>>> the ticks without your patch are always jutting out. With your patch,
>>> the ticks are either coming out of the screen or going into the screen
>>> depending on the elevation from which you're looking at the figure.
>>> 
>>> 
>>> On 9 January 2011 17:38, Ben Abbott <address@hidden> wrote:
>>>> The axis box and grid lines appear to create a 3D cube whose
>>>> contents are hidden. Thus the globe appears (to me) to be outside
>>>> the axis limits.
>>> 
>>> You appear to be a victim of this:
>>> 
>>>   http://en.wikipedia.org/wiki/Necker_Cube
>>> 
>>> I agree with Ben here on what it appears to look like. I don't like
>>> this Matlab behaviour. Having the ticks on top seems to give a sort of
>>> visual cue that the figure is inside the cube, not on top of it.
>>> 
>>> Or it could be just a case of the spinning dancer, everyone sees it
>>> one way and exactly one way only and can't agree with people who see
>>> it the opposite way.
>>> 
>>> - Jordi G. H.
>> 
>> I've applied the patch and run the ellipsoid demo.
>> 
>>      demo ellipsoid
>> 
>> Using the mouse to rotate, the result looks very nice to me.
>> 
>> My apologies to Jordi, but after applying the patches and experimenting, I 
>> like the result.
>> 
>> I've uploaded a couple of mpg videos so others can compare Kostas' 
>> implementation with what I see with Matlab 2010b on MacOS X.
>> 
>> The Octave version is  ...
>> 
>>      https://files.me.com/bpabbott/zahp7n
>> 
>> The Matlab one is ...
>> 
>>      https://files.me.com/bpabbott/k0xzj8
>> 
>> Ben
> 
> I think I've misapplied the patch. If I remember what I did the Octave 
> version is the results *before* Kostas' changeset was applied.
> 
> I'll apply the changeset and produce another video.
> 
> Ben

With the patches correctly applied Octave behaves like Matlab does.

After playing with both implementations, can't say I favor one over the other 
... I'm fickle if nothing else ;-)

Ben