I started by scanning the original glasses' lenses on a flatbed scanner, that gave me some pretty good measurements in 2 dimensions.
I lack the measuring tools to get better precision, so in the spirit of rapid prototyping I quickly estimated an inner and outer tolerance by dropping an image of the scanned lenses into the model in SketchUp. I traced the image of the lens, and used the offset tool to copy the outline of the lenses and create a concentric inner shape that was small enough to mask all the inner edges of the lens.
After removing the lens holes from the original model, I was left with a flat plane to drop my new lens outline on. At this point, I'm only working with one half of the model, because I know I'll need to make some adjustments after the first print.
The push pull tool was used to carve the new lens hole into the old model (that had its lens holes removed), The inner outline was punched all the way through, while the outer outline was pushed only halfway into the model. After this, the first test was printed.
The depth of the lens holders turned out to be too shallow on the first test. the lens also needed some resistance in the corners to avoid rocking around. I also needed to greatly increase the length and depth of the tab that would hold the lenses in. I increased the depth of the the entire frames portion of the glasses to accommodate the lenses better and bevelled the corners of the lens holes to fit the lens curvature better.
The second test print was optimistic, so I printed both halves of the glasses (by mirroring my working 1/2 glasses model), and the stems. The stems came out well, but the parts of the hinges on the frames were too close together for the height of the part of the hinge on the stems, and the frames needed more help before they would hold the lenses.
The third test had the added addition of more space for the stem hinge, and some reinforcement of the hinges on the frames. I also added a slight bump to the front of the glasses along the bridge of the nose so they wouldn't be completely flat on the front. This test was once again only a print of 1/2 of the frames, (I concluded that my previous optimism was overoptimistic.) The third test turned out great, so I mirrored the half model back to wholeness and printed the ones you see on my wife's face.
Currently, the hinge is secured with part of a nail. I'll have to find the right screws to make it nice. For now, the old frames are holding up, but these will be ready for their lenses when they finally give out.
I'm told they are blurry around the edges, so the next revision will have a bend in the bridge that causes the lenses to sit at the same angle as the previous frames.