Christian Froeschlin <chr### [at] chrfrde> wrote:
> gregjohn wrote:
>
> > Q: What tricks would you use in povray to make a great 3D rendering "fax-machine
> > proof"?
>
> I have no idea what they did but the images look dithered / color
> reduced. So possibly it would be better to provide "predithered" images
> in b/w or with only 16 fixed gray values (0,16,32,64,...) instead of
> arbitrary ones. However, for best results you should then also know
> the dpi resolution they're going for (when in doubt it is probably
> better to use a lower resolution, because scaling up has less
> terrible artefacts than scaling down).
>
> But a nightmare for debugging if you have to invent a new
> patentable gadget for every test run ;)
>
> BTW is the limit of your invention a space-filling curve?

In effect, you have a certain volume available to you in the chip, and your goal
is to maximize the number of discreet elements that can be tested in that
volume. So in that sense, it is space-filling. But when one individual (or
thousands) go bad, you have to be able to find at least one individual for TEM /
SEM analysis. The trick here was to be able to capitalize on defect isolation
techniques that may have very low resolution, to allow regionalities in two
different-space filling curves to get you to the right neighborhood.

Post a reply to this message