[ImageJ-devel] [ome-devel] ROI Display Options

[dzmacdonald]

Hi Curtis,

We've just moved the description of the ROIShape from the  
ShapeDisplayOptions onto shape, there are several reasons for this:
* It helps on the server side, simplifying the queries for retrieving  
shape info (no joins on a potentially large table).
* One of the reasons we've got display settings on shape rather than  
ROI is to compensate for the lack of ROI-ROI links, so an ROI could  
define the change of state of an object via text attribute
* We also are looking at using ROI/ROIShape to do more than just  
define ROI on an image, but to store other UI geometries.
* It was the way I coded up the measurement tool :)

This proposal is definitely not set in stone, I can see it being more  
advantageous to move the display settings to ROI and therefore make  
ROIShape deal strictly geometry when we get ROI-ROI Link.
I am quite dubious about  moving DisplayOptions to a separate table to  
allow more displays to store separate display options for ROI, as the  
trade of off performance (joins on table) vs use-case is debatable.

We are going to have a minigroup to discuss ROI and modeling, it's  
provisionally set for Friday (28th May) at 2pm, anyone who's  
interested is welcome to join us via Skype, Teamspeak.

Also  If any one is coming to Paris we can certainly discuss use-cases  
and talk about any changes that maybe required.

BTW, You may not be aware of the proposed changes to ROI to  
accommodate 3D ROI, http://ome-xml.org/browser/Documentation/Graphics/ROI/2010-09/Ome-ROI-overview-Proposal-2010-09.png 
.

Regards

D.

On May 18, 2010, at 7:21 PM, Curtis Rueden wrote:

> Hi everyone,
>
> Today the ImageJDev team met to discuss regions of interest in  
> ImageJ, and as part of that discussion, we were looking at the  
> 2010-09 proposal (http://www.ome-xml.org/wiki/ROI/ProposalSeptember2010 
> ). Barry suggested that the "Display Options" section is distinct  
> from the rest of the ROI model, in that it describes how to render a  
> ROI, rather than its geometric representation in model space.
>
> Would it make sense to break out the Display Options into a separate  
> element (e.g., ROIDisplay), linked back to the ROI? (This could be  
> similar to the old "DisplayOptions" element, but for ROIs.) Doing so  
> would allow two or more displays to store separate display options  
> for the same ROI.
>
> -Curtis
>
> _______________________________________________
> ome-devel mailing list
> ome-devel at lists.openmicroscopy.org.uk
> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://imagej.net/pipermail/imagej-devel/attachments/20100525/721688de/attachment.html>