Originally Posted by Mr. Foo
I took a longer look tonight at the 1080i deinterlace tests that I had at my disposal, specifically the Vertrez and the Vertrez motion tests. These tests can be found here
, including a master file with all the tests that can be used to burn your own HD-DVD test disc.
I also went back and looked at an old post
discussing these tests to ascertain what exactly to look for in these two tests and have the following results:Vertrez (static pattern) deinterlace test:-------------------------------------------------
What does the test look like?: Test video (static) contains blocks of alternating black and white lines of various thickness (e.g. each block has a set of lines of a given thickness). The area of interest is the topmost block of lines (1 pixel in height). HERE
, you can reference a picture of the test screen that I took when I ran it on an A2000 set.
Want: alternating black and white lines in topmost block to be visibly distinct, which indicates that the TV is doing a proper weave deinterlace and preserving the full 1080 lines of resolution.
Mode 0: All lines visibly distinct.
Mode 1: Block will flicker fast or slow depending on pallette settings. Setting Reality to 25 and Clarity to 100, completely removes the flicker and all lines appear distinct.
Mode 2: Block will flicker fast or slow depending on pallette settings (though not as severe as with mode 1). Setting Reality to 1 and Clarity to 1, completely removes the flicker and all lines appear distinct.
So, in summary it appears that proper weave deinterlace is possible in all modes, depending on how you set the DRC palette. I would assume that the above results can be assumed to be the desired settings (this is where I think I will be leaving them). Another thing to note here that is nice is that it seems that the TV remembers your palette settings for the given mode. Like, if you have different palette settings for mode 1 and 2 (which appears to be very significant) and you switch from 1 to 2 and back to 1, it will remember what you had mode 1 set at. I'll assume that the TV saves settings per mode for each input, though I haven't tested this out yet.Vertrez Motion test:-------------------------------------------------
What does the test look like?: Same as the static Vertrez test, but also containing a rotating line/blade that rotates clockwise through the image (think airplane propeller)
Want: alternating black and white lines in topmost block to be remain distinct at all times and the line to be a single, solid image as it sweeps through the block. This would indicate a per pixel motion adaptive deinterlace
Results: Block of horizontal lines remain distinct if the aforementioned mode-specific palette setting are used. The picture shows a double image of the blade that are both solid in color, which, from what I read may indicate motion adaptive blending. Note the following regarding the color of the lines and the coloring around the lines, as noted per mode (keep in mind clockwise direction when reading terms such as preceding and trailing to help get a mental picture of what is going on):
Mode 0: Trailing line of the pair changes from white to black over the region sweeping through the topmost group of lines. The leading line remains white. There also appears to be a region of white in between the two lines and preceding the leading line when it sweeps through the topmost block.
Mode 1 and Mode 2: Trailing line of the pair also changes from white to black over the region sweeping through the topmost group of lines. The leading line also remains white. There again appears to be white in between the two lines, BUT there is a black region preceding the leading line when it sweeps through the topmost block.
My use of the term region above really almost appears as a shadow that traces the rotating lines and that which is only significantly present in the topmost 1-pixel width set of lines.Jay Leno Test:-------------------------------------------------
I also viewed the Tonight Show monologue in HD, per your suggestion earlier as another quick way to try and assess video deinterlacing capability. I didn't notice any changes at all in sharpness or resolution in any of the buildings or set background set as he was moving around on stage. Clear as a bell really.
So, to sum up - this all seems like good news to me, though I can't quite fully interpret what is going on exactly with the motion tests. Perhaps, its some kind of region adaptive blending going on that, after performing the Leno, may work out just fine.