Originally Posted by susanstone
has anybody tried deinterlacing test pattern? can nvidia or ati handle it?
See the "Tested and Failed: throwing away half the Resolution by Gary Merson" threads:http://www.avsforum.com/avs-vb/showthread.php?t=608670http://www.avsforum.com/avs-vb/showthread.php?t=608998
Here's some patterns with 3/2 pulldown that aren't hyperlinked on the main page.
1) email@example.com bitstream of rotating "helicopter" bar. The source is 72 frames of a solid (progressive) bar in both fields that advances 360/72 or 5 degrees per frame. However, 3/2 pulldown has been turned on at the source. If you examine each frame, you'll see the bar as solid in 3 frames and then interlaced in 2 frames (in a repeating sequence).http://www.w6rz.net/vertrezivtc.zip
2) firstname.lastname@example.org bitstream with inverse telecine flags of the same source. Unfortunately, the encoder I used is not able to stay is film mode when the bar goes horizontal. This is due to the encoder being a 3 chip hardware encoder where each chip does a strip of the image (top, middle and bottom). When the bar goes horizontal, it's only in the portion of the image that's being processed by the middle chip. The top and bottom chips don't see the repeated field, and the encoder drops to video for a few frames.http://www.w6rz.net/vertrezivtcflags.zip
3) email@example.com bitstream. A progressive encode of the 72 frames without 3/2 pulldown. I tossed this one in for folks to see if their decoder can properly decode 24p bitstreams. On a good decoder, all three bitstreams should act the same.http://www.w6rz.net/vertrez24p.zip