@rendalunit

Hi ren,

I tried the LanczosResize on "The Green Mile", and the result looks worse than BilineraResize 8O . Maybe the added sharpness around the edges makes this difference. I didn't like the result, compared to the test sample that came out with BilinearResize. Looks more like the original LBR sample. Maybe Mr. Lan C. Zos ( :lol: ) doesn't like low bit rates :twisted:

-kwag

Take a look :). This sample includes, at the beginning, about the worst Gibbs I could find in the whole clip...

That looks VERY good for 2 hours 8O
I viewed it with WMP, so I guess fully stretched in my TV it will look even better :D
I can see the Gibbs effect, but it's very little and blurred. I rather see this encode on my TV than the same thing encoded at 352x240. Look at the horses on the far background. You can even see some detail! At 352x240 they would look like "Something" moving in the background :lol:

-kwag

Yep :). I decided to take a chance with x3 given the Cinemascope aspect ratio. I think had it been regular 16:9 I probably would've had to use too low a CQ_VBR for one disc. However, I won't know until I try it... :D

Exactly. I'm willing to put up with a little Gibbs (a little, mind you ;)) and encode at 528x480. I find the sharpness very pleasing. That said, even going from 352x240 to 352x480 makes a huge difference for me. I think it's that doubling of the vertical resolution that's the most important part.

Overall it looks like this encode came out well, and I can't wait to try it on the TV when the kids have gone to bed :).

Hi guys,
Just got up and back to business. Yes, I did use FitCD to resize, and yes I did set Block TV overscan to 2, and yes I did use the new GOP (1-12-2-1-24). I used Flux, LegalClip, and Blockbuster as well. I just don't know what went wrong.

So, is the prediction script in FitCD still valid or we can now switch to:

IL = Framecount / 50
SL = round(Framerate) * 2
SelectRangeEvery(IL, SL)
[/quote]

Tell me about it 8O , I've been waiting for the TV to test my burned CD-Rs of "The Green Mile" and "Red Planet". My wife and my brother in law are watching a movie on the HDTV, and my kids are on the 60" old Magnavox playing Dreamcast :? :x So I'm 3'rd in queue :P

-kwag

@SansGrip,

Tmpgenc's Clip Frame mask is top=73, bot=78, left=19, and right=17 for
resize 496x448 with borders (16,16,16,16) :)

-black prince

Are these values valid for pal too?
I want to encode Brain Dead :twisted:
Movie lenth is 99 min

Is this ok?:
IL = Framecount / 30 # interval length in frames
SL = round(Framerate)*2 # sample length in frames
SelectRangeEvery(IL,SL)

And what is the constant now? 0.89?

If that's the size of the borders before resizing then here's the script you can try:

Those are mighty big borders that you weren't cropping. No wonder the file size is so different :).

Don't worry, that's how I encode most of the time too :mrgreen:.

I would use 50 instead of 30, at least until we've thrashed out good values for this new GOP.

Very simple solution: buy another TV -- maybe a 50" plasma widescreen? :mrgreen:

Just tell them it's an encoding emergency... ;)

First of all... thank u verry much for the help.
But i still have some questions... :lol:


Sorry but i don't get it yet

btw the quality is really superb with the new GOP

I can't think of a reason why not.

I'm not sure -- best ask kwag about that one :).

Thank you Sansgrip.
I'll just give it a try and post the results tommorrow. it's 3.00 AM in Germany :D

50:48, or better yet: Where Snapshots will vary, depending on the movie time, and 48 is a fixed window width of 48 frame for NTSC or 50 for PAL
The factor i'm using is now .98, and seems to be pretty accurate.

-kwag

Little did I know my wife would decide to order Divine Secrets of the Ya-Ya Sisterhood on PPV. DIVINE SECRETS OF THE YA-YA SISTERHOOD! For this I'm waiting...

How about a trip to your circuit breaker box and 8O :oops: , power fail :lol:

-kwag

Hi there,
I downloaded SansGrip's sample and watched in PC and TV, I now know it is called "Gibbs". This is the thing that I have been trying to describe to you people. What causes "Gibbs" and can we get rid of it??

That's the next target :D
Quantization Matrix optimization :wink:, improvements on the KVCD Q. Matrix.

-kwag

Well, I just finished watching Untouchables on the TV and all I can say is 8O, even though that's not a word ;).

It really does look very close in quality to a DVD (at least on my TV), with the only noticible artifacts being caused by the Gibbs effect -- not so much visible mosquito noise as a "shimmering" of very high-frequency areas. It's barely noticible, though, once you stop looking for flaws and start enjoying the movie: my wife, who is untrained in the art of critical viewing, did not notice it at all :D. I'd really love to try one of these encodes on an HDTV, but no-one I know has one. kwag, could you do me a favour by burning that clip onto CD-RW and testing it on your high-definition system?? I'd be fascinated to hear if the quality stays so good on such a setup...

One problem not related to the new GOP: three times during the very beginning of the movie, separated by perhaps a minute or two, the video and audio froze for about a second, then resumed as normal. Beyond roughly five minutes into the movie it didn't happen again. I know I need to replay and see if this occurs in the same places every time (indicating perhaps a media failure), but if it doesn't, what could be causing it? Since it's only right at the start could that signal a too-low minimum bitrate?

Anyway, apart from those two very minor issues the thing looks incredible considering it's on one disc (heck, it looks better than most of my old encodes that were on 3 discs for a 2-hour movie). Naysayers be damned :mrgreen:.