On Sat, 2004-05-22 at 01:56, Steven M. Schultz wrote: > On Sat, 22 May 2004, Dan Scholnik wrote: > > > > > You might even try running y4mspatialfilter before and after y4mdenoise > > > > in case the latter introduces any high-frequency artifacts. > > > You probably would get the same 3% by running y4mspatialfilter twice in > > a row before y4mdenoise, or by slightly lowering the bandwidths (which > > Maybe - but your suggestion was that running it after y4mdenoise > would remove any high-frequency artifacts that were introduced > during the denoising.
Right, but only 3% gain didn't suggest there was much of that, is all. > How would lowering the bandwidth (or running y4mspatialfilter twice > before y4mdenoise) remove any hf artifacts caused by denoising? It wouldn't. > Keeping the same bandwidth for the 2nd run shouldn't find anything > to remove/rolloff - should it? "bandwidth" is a tricky notion, because Fourier theory tells us that no finite length filter can roll off instantly. So usually the stated bandwidth of a filter is where it has rolled off by some particular amount, typically 1/sqrt(2) or 1/2. Cascade 2 identical filters, and the result rolls off earlier than either alone. The bandwidth in y4mspatialfiler isn't quite that well defined, but the result is the same. Try running about 10 y4mspatialfilters in a row and my guess is you'll be able to see the difference. (Ick. just tried it, and there is a pronounced green color shift that accumulates. Looks like I need to fix a little bug somewhere...) So, to sum up, I was just speculating that the extra 3% came not from removing y4mdenoise artifacts, but from the slightly heavier effective spatial filtering resulting from two calls to y4mspatialfilter. To find out for sure, run the 2 y4mspatialfilter calls back-to-back before y4mdenoise. Dan ------------------------------------------------------- This SF.Net email is sponsored by: Oracle 10g Get certified on the hottest thing ever to hit the market... Oracle 10g. Take an Oracle 10g class now, and we'll give you the exam FREE. http://ads.osdn.com/?ad_id=3149&alloc_id=8166&op=click _______________________________________________ Mjpeg-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/mjpeg-users
