Setting aspect ratio for 1.33 MKV file

[carlmart]

How about vertical bars? A 1.66:1 ratio movie encoded as 16:9 has pillar bars.

1.66 bars are quite close to 1.77 and can go unperceived.

In fact what I bluntly said is the opposite of what I really think. What I meant to say was that I didn't like letterboxed 4:3 images, and I immediately zoom them in to fill the screen. OTOS I do not like distorting a 4:3 image just to fill a 16:9 screen side to side, which is what many people do here in my country.

A 1.66:1 movie at 16:9 has only half the increased resolution of a 1.66:1 movie at 4:3, when compared to a wider film (1.85:1 or 2.35:1, for example) at 16:9 and 4:3. Therefore, when starting with a good 4:3 source there's every reason to think the upscale can be done successfully with far fewer of the artifacts (aliasing and the like) that plague 4:3 to 16:9 conversions of wider movies.

The following thing I try to keep and respect is the original size the film was shot or screened at. Whatever the ratio. I hate the zooming that was done in TV that cuts away image information. The artist conception has to be respected, even if you don't like it.

As I said, in this case I would like to experiment with the avisynth resize function vs the DVD zoom. But to resize I might have to try other filters to correct minor blemishes due to the upscale.

For instance: quite recently I restored two old John Ford films that I had, using colorshift and undot with good specific results. I also tried BlindDeHalo2 and lsfmod, but I didn't like the results. What I mean is something to compensate the softness from going from a letterboxed 4:3 to a full 16:9 (with small side bars).

[manono]

Funny, I just got done dehaloing a DVD (using DeHalo_Alpha) with a huge amount of edge enhancement, and it softened the heck out of the thing. Using LSF afterwards didn't begin to restore the sharpness lost. That LSFMod is supposed to have a smoother built into it, to help prevent any sharpening artifacts, so I'm not real sure what the problem is. Maybe you didn't set the strength high enough?

1.66 bars are quite close to 1.77 and can go unperceived.

Unless your TV set has a whole lot of overscan, with a 1.66:1 movie encoded as 16:9, you'll see them. Depending on how you do the resize, you might have as much as 32 columns of pixels on each side.

[carlmart]

Funny, I just got done dehaloing a DVD (using DeHalo_Alpha) with a huge amount of edge enhancement, and it softened the heck out of the thing. Using LSF afterwards didn't begin to restore the sharpness lost. That LSFMod is supposed to have a smoother built into it, to help prevent any sharpening artifacts, so I'm not real sure what the problem is. Maybe you didn't set the strength high enough?

I think I did. I was using "lsfmod(smode=5)", which I think is the highest I can get. In any case, until I really master it, probably I should retry it again until I get a repetitive result I can evaluate better.

The film has some grain, so maybe some degraining is also in order.

Unless your TV set has a whole lot of overscan, with a 1.66:1 movie encoded as 16:9, you'll see them. Depending on how you do the resize, you might have as much as 32 columns of pixels on each side.

I'm positive it doesn't underscan anything, except as described below. Of course I see the pillars, but probably because this TV has black borders, the 1.66 side pillars are not as evident as 1.33 ones. The screen is much more rectangular in 1.66, and the brain forgets about the side borders more easily.

[carlmart]

Well, I did research this zoom thing a bit more: the player's and the TV's.

For that I also tried another originally released as letterbox film I have: Cabaret.

What I could check is this:

Cabaret (16:9 letterboxed)

1) The player's zoom overscans it and loses information on all sides. I tried to see if I could set other intermediate zoom levels, but I can't.

2) The TV's zoom plays it fine. Nothing is lost and even a rather thin black bar is apparent on all sides. So the zoom is better here. It allows only some vertical adjustment, but it changes the shape then.

The bride wore black (1.66)

1) The player's zoom plays it fine. No overscan.

2) The TV overscans it a bit on all sides.

So in both cases it might be better to go for a full resizing with avisynth. At least I will have a predictable more precise result then.

[manono]

I think I did. I was using "lsfmod(smode=5)", which I think is the highest I can get.

That's the sharpen mode - what method is used to sharpen it. How sharp you make it is set with the strength parameter. Default is 100. It has soft (softening) built into it, and you can turn on soothe if you like. I don't use LSFMod so haven't bothered to figure out how much the default soft affects the strength. When I use the old LSF together with soothe, with my settings a strength of 100 does almost nothing and I usually use strength values of 4-600. But like I said, I don't use and don't know LSFMod.

http://avisynth.org/mediawiki/LSFmod

[carlmart]

OK. I will start my filtering tests with "The bride wore black".

First of all, and to learn a bit more on using FitCD, how do I set the parameters for loading this film into it, so I can know how to resize it in avisynth? I need precise border figures.

Remember it's a letterboxed 1.66 film.

[manono]

First of all, and to learn a bit more on using FitCD, how do I set the parameters for loading this film into it, so I can know how to resize it in avisynth? I need precise border figures.

Remember it's a letterboxed 1.66 film.

I don't understand what you're asking. When you hit the 'Source' button, it tells you what formats it accepts as input. Or you can just fill in the Input boxes yourself.

When I was checking how much the borders would be for your film when converted to 16:9, I plugged in 1660x1000 and 1:1 Monitor. Then the only remaining decision is whether or not to use ITU resizing. A sample script might then go something like this:

Crop(0,48,0,-48) #if those are the correct crop values
Lanczos4Resize(656,480)
AddBorders(32,0,32,0)

and encode as 16:9.