Stephen Marsh wrote:

> When timing a first run of a filter, say gaussian blur for example, it may
> take say 12 seconds. An undo or stepping back in history and then
> reapplication of the filter showed a significant decrease in time for the
> re-application of the filter on the original data, say 5 seconds for
> example. The only difference seemed to be that the filter was a rerun and
> not the first application?

I already observed that, when I suggested the experiment - that was why
I specified doing the filter and then do a reapplication of the filter
using a keystroke before timing the result

seems clear to me that the GUI menu and all the mechanism and overhead
for getting rid of the preview window was enough that one got differing
results each time depending on how it was done


I dont think it was the lack of science, consistency and all the rest
which caused some unease in the group  - more that the fear that the
thread could take up more space than it deserves

philip
===============================================================
GO TO http://www.prodig.org for ~ GUIDELINES ~ un/SUBSCRIBING ~ ITEMS for SALE

Reply via email to