custom zmi interface & hidden fields

Hi,

I'm thinking of implementing a custom graphic conten object (or already begun 
implementing it), that allows to embed graphics into texts as usual, but also 
hooks them up to a lightbox-like display.

As extra bonus, I want to offer a cropping/display-rectangle-option that 
allows it to specify a certain portion of the uploaded image for the 
thumbnail. Think of the way galleries are exposed in spiegel online - the 
article shows a few images, always squares, that are a partial view of the 
whole image you see when clicking to it.

Now I have two questions:

 1) I created my own thumbnail-creation-external method (wasn't satisfied with 
the build-in), and store the thumbnail as property of the object.

But I didn't find a way to hide the thumbnail - I can only create a publically 
visible object property.

Is there a way to circumvene that and create "private" properties?

 2) For the region selection, I need of course a rather sophisticated JS-based 
thing. Is there a way to create custom ZMI interface components for 
content-objects? I'm happy to hack ZMS itself, only wonder if & where to 
start.

Diez