Hi Rogier,

On December 30, 2010 05:12:26 am Rogier Wolff wrote:
> It seems we're pretty much fully in disagreement on this issue...

... and it is not a problem at all for us to keep working together and 
collaborating on Hugin.

The development process and repository structure I have put in place enable 
just that:

We have 'default' that is the integration codeline.  In there go feature that 
are declared mature according to the maturity criteria set out in [0].  There 
is the only point where we need to reach consensus.

From there out we branch into our different interests.  People who want to add 
new features branch out a development codeline.  People who want to polish an 
existing set of features work on the a release codeline until they declare it 
final.  You can even branch out your own personal flavor - other projects like 
VLC do this too.  Once the branch is your, you can do whatever you want with 
it.  If others will like it, it will be added to the integration queue and go 
into mainstream.


> Hugin is a user-friendly open-source solution allowing stitching of
> multiple photo images into a single output image. From stitching three
> six-megapixel images together into a wide-FOV panorama to
> multi-gigapixel HDR panos.

Nothing prevents you from branching out any time in the project history and 
transform Hugin into what you believe fits the above vision.

 
> Because of a good software engineering design, hugin allows several
> steps of the workflow to be dispatched to subprograms that can easily
> be replaced, compared, extended and researched.

And nothing prevents people interesting in adding features to Hugin to do 
this.  In fact, what you are writing above has just been improved with the new 
Makefile lib.


> As a side note: Maybe Hugin has too many controls for the first-time
> user. But because of reasonable GUI design with an "assistant" a first
> time user can achieve results.

This is actually inconsistent with the feedback we're getting.  Users (and 
what is even more worrying: not only first-time users)  run into optimization 
problems of all sorts.  That's not reasonable GUI design.  A reasonable GUI 
design would prevent that.

If somebody is interested into researching and developing a proper GUI design 
for such a complex multi-faceted problem, here is a very interesting 
challenge.  And it would not yet be a finished product.  In 2007 Ippei did 
such a research, and the idea was to switch to a Qt based UI.  The branch 
dried out after GSoC 2007 and things have moved on with the existing wxWidgets 
UI.  The refactoring of the code however helped bringing more clarity and a 
better architecture on which the current Hugin builds, a few iteration and 
many massive changes later.

Yuv


[0] 
<http://wiki.panotools.org/Development_of_Open_Source_tools#Maturity_Criteria>

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to