Theory: Scan Data - The good, the bad, and the ugly...

Scanner makers have this idea that "precision" is the most important aspect of scan data PERIOD!!! nothing else matters.

At least in my experience working with frustrating data. Data with holes, spikes, blurps, and worse, BAD topology (like bad welded vertices). I'd like to suggest to them that a full and complete shell, at REALLY LOW accuracy is, at least sometimes, more important that only keeping PERFECT data.

When you photograph a portrait, you don't want all the pores, and pimples... a soft image is enough to get the picture, and is actually nicer.

When I'm scanning something, I sometimes just want a light-weight fast framework to model around. I want accurate-ish features to model around. But I build my models in pro/ENGINEER anyway, so it's all general... you build the features semantically correct, and then dial in the dimensions. That's the beauty of parametric feature based modelling. So why do I care if my data points are +/- 0.04617475455 <- yeah that says O POINT O A GNATS ASS - if yo don't see that - sorry, ponder it a second. :) anyway... point is, the dimensions could be off by huge amounts if all the features are there. I can always go back in later and tweak and dial the model in to match for precision.

So, another thing about more complete data... start off with a complete shell, and then TWEAK the model/data, as it gets better and better. like... as if you started with a blob roughly the shape and snapped all the points to the good data - marking the points that are better, and points that are worse, allowing the operator to use his/her processor(brain) to interpolate the data. Sometimes it take a half a day to get the scanner to see the data correctly, when I could smear it over in a mater of minutes (if I had a clean model to work with)

anyway ;) just a few thoughts on scan data... I'm sure there're reasons why they act like this about the data, but sometimes I wish they'd give me more control to visualize and manipulate the data myself.

So I've began reverse engineering the raw data from my scanner to see what exactly IS available... and honestly, it looks like they actually give me all the useable data when they give me the "good stuff"... as in, sheesh... anything that isn't good is looking HORRIBLY NOT GOOD!!! :) hah.

We'll hvve to see. I haven't got a white paper or anything on what the data is, or where it came from so I'm still piecing it together. -- will follow up when I've figured more out.

ntlthn cyal8r!