How to make an AVI better quality (keeping output still as AVI)?

[Dogpoundz]

Hello all.
I know there are guides on this website for what I am about to ask (encoding), however, none of them are geared for newbies, So i am hoping someone is willing to help me.

OK, so in DETAIL, here's my Question:



Let's say u come along a movie (in AVI format) that is 1,000 MB. However, the quality isn't that great.
What I am wondering, is if you can better encode the 1,000 MB AVI to a slightly better quality AVI, such as 2,000MB (so pretty much doubling, tripling, or even 1.5X the size)?

I am also looking for a process that won't really take a long time, but you still will be able to notice a difference.

However, I am such a newbie and have no experience whatsoever with encoding. And it seems so TOUGH. So, I really would need a step by step process, Ive tried reading guides all over the internet and this website, but its so tough to do, and the guides are never really geared towards newbies,

THANKS A MILLION


PS. Also, I am guessing it is best to not change formats, since A) it seems you always lose quality when you do so B) may be quicker to keep as AVI, and C) i am watching on PC. (Unless these assumptions are wrong).

Now I know you can use that Tmpgenc program to convert it to DVD. And if I remember correctly, the quality improved. However, it took a real long time to encode. You def lose quality (or remains the same) with all other conversion programs ive used, such as convertxtodvd, winavi, etc

[hech54]

I am guessing it is best to not change formats, since A) it seems you always lose quality when you do so B) may be quicker to keep as AVI, and C) i am watching on PC. (Unless these assumptions are wrong).

Now I know you can use that Tmpgenc program to convert it to DVD. And if I remember correctly, the quality improved. However, it took a real long time to encode. You def lose quality (or remains the same) with all other conversion programs ive used, such as convertxtodvd, winavi, etc

Any kind of "enhancement or filtering" in VirtualDub requires re-encoding....even if you keep the same codec(format in your words). Turning the video file into a DVD does NOT improve the quality. Once the quality has been degraded by compressing it to AVI/Divx/Xvid, the original quality is gone forever. Either buy the DVD or wait for the DVD to be released.....simple as that.

[Cornucopia]

@Dogpoundz,
I hate to let you down, but realistically, NO.

Wherever you start from, you can only maintain or lose quality (irrespective of format). Sometimes, with some effort, you can make what SEEMS to be better quality, but what's really happening is you are trading an improvement in one aspect of quality (likely one that is more important to YOU) for a loss in another aspect.

The only way you can truly get better is to get back to a NEW START. This can be done by using a better quality, more original master, or by CREATING (Capturing an Image, or Drawing/Painting).

Scott

[Dogpoundz]

@Dogpoundz,
I hate to let you down, but realistically, NO.

Once the quality has been degraded by compressing it to AVI/Divx/Xvid, the original quality is gone forever.

Thanks for your quick replies

thats sad,

See i figured, there was kind of a way of unwinding the compression done, and then redoing it to a better quality. for example changing that whole bit rate, video rate, etc etc

The reason, i thought this was because, when u save an AVI in VirtualDub in full processing mode, itll change the size of a 1,000 MB AVI to a HUGE ENORMOUS size, so I thought this was undoing the compression

but i guess the encoding/compression process deletes segments huh, and the segments have no pattern huh?

too bad there wasnt a program that could like re-map the deleted segments based on some kinda like reverse probability forecasting.


For example,

LINE 1
............LINE 2
.......................LINE 3

was compressed to

LINE 1
.......................LINE3

and go back and fill in LINE 2

Doesnt work that way huh? its random?


PS, to HECH, these are actually, homemade movies sent from a friend in the middle east,
and as a general advice on life, it is always wise to keep ur assumptions to urself,

[Ai Haibara]

thats sad,

See i figured, there was kind of a way of unwinding the compression done, and then redoing it to a better quality. for example changing that whole bit rate, video rate, etc etc

The reason, i thought this was because, when u save an AVI in VirtualDub in full processing mode, itll change the size of a 1,000 MB AVI to a HUGE ENORMOUS size, so I thought this was undoing the compression

but i guess the encoding/compression process deletes segments huh, and the segments have no pattern huh?

That happens because if you don't specify a codec for VirtualDub to use (Video menu > Compression...), it defaults to storing the video in raw uncompressed form.

Lossy compression on videos is basically throwing away bits of information here and there. There isn't really an easy way for anyone or a program to guess what's been discarded, and as a result, restore it.

[Dogpoundz]

That happens because if you don't specify a codec for VirtualDub to use (Video menu > Compression...), it defaults to storing the video in raw uncompressed form.

Lossy compression on videos is basically throwing away bits of information here and there. There isn't really an easy way for anyone or a program to guess what's been discarded, and as a result, restore it.

Well thanks for clearing that up for me, goes to show how much of a newbie i actually am, thanks

[jagabo]

Take smooth shiny piece of aluminum file. Wad it up into a little ball (compress it). Open it up again (decompress it). It's now the same size as it originally was but it's full of wrinkles. Wad it up into a little ball again, but slightly larger than last time (recompress it, but not as much as last time). Open it up flat again. Does it have more wrinkles or less?

[Des]

Take smooth shiny piece of aluminum file. Wad it up into a little ball (compress it). Open it up again (decompress it). It's now the same size as it originally was but it's full of wrinkles. Wad it up into a little ball again, but slightly larger than last time (recompress it, but not as much as last time). Open it up flat again. Does it have more wrinkles or less?

Great analogy Jagabo

--dES

[hech54]

[PS, to HECH, these are actually, homemade movies sent from a friend in the middle east,
and as a general advice on life, it is always wise to keep ur assumptions to urself,

Of course they are....we ALL believe you ....
And no....I will not be keeping my assumptions to myself. Never have.....never will.

[Dogpoundz]

Of course they are....we ALL believe you ....
And no....I will not be keeping my assumptions to myself. Never have.....never will.

so now u speak for other people, what are u a spokesman?

[Dogpoundz]

Take smooth shiny piece of aluminum file. Wad it up into a little ball (compress it). Open it up again (decompress it). It's now the same size as it originally was but it's full of wrinkles. Wad it up into a little ball again, but slightly larger than last time (recompress it, but not as much as last time). Open it up flat again. Does it have more wrinkles or less?

i agree with des, that pretty much explained, however, too bad there was no way of creating a program that would figure out the exact pattern that a particular codec "wrinkled" the movie, and undoing the "wrinkles", but i guess this is way over my head, which is why i cannot see why it cant be done

For example. lets say CODEC A was used to compress a movie,

CODEC A cuts and deletes the movies at every 0.000005 seconds,

ok, and then u run a program that goes in, examines the data at every 0.000004 (POINT A) seconds and 0.000006 (POINT C) seconds, and uses predictive analysis of what was cut at every 0.000005 (POINT B) seconds and fills in the missing data

BUT from what it seems everyone is telling me , it doenst work that way, it seems POINT B is cut, and POINTS A and POINTS C are squeezed,compressed together, therefore, losing the original position of POINTS A and C, therefore being unable to predict what POINT B originally was.

is this the way to think of it?

u see what i am trying to ask, pretty much, there is no way to predict the pattern of how a particular codec compresses a movie, and not just undo the compression but also fill it the deleted segments based on a predictive nature

[jagabo]

too bad there was no way of creating a program that would figure out the exact pattern that a particular codec "wrinkled" the movie, and undoing the "wrinkles"

I have a list of 5 numbers. They range from 1 to 100. I'm going to compress that down to one number by averaging them together. The result is 37.5. What are the 5 numbers?

Of course, there's no way for you to know. Ever. It's the same with compressed video. Information has been removed. There's no way to restore it.

[AlanHK]

too bad there was no way of creating a program that would figure out the exact pattern that a particular codec "wrinkled" the movie, and undoing the "wrinkles"

Because all normal video codecs are lossy.
The video data is simplified and the patterns of the SIMPLIFIED version are what are stored.

Detail is lost. Lost, not compressed.

Codecs that CAN be undone perfectly exist, these are called "lossless", such as Lagarith.
These are the only ones that you can get back exactly the original image.
The files they make are often hundreds of times larger than say DivX.
If you could exactly reverse something like DivX, you could make a great deal of money.
Then you could invent perpetual motion.


Similar processes are used in storing images.

Use any image editor that can save as JPEG, save an image, then zoom both the original and the saved version and compare them. You will see they are not identical. Edges have been blurred, colours slightly changed, etc.

For example. lets say CODEC A was used to compress a movie,

CODEC A cuts and deletes the movies at every 0.000005 seconds,

ok, and then u run a program that goes in, examines the data at every 0.000004 (POINT A) seconds and 0.000006 (POINT C) seconds, and uses predictive analysis of what was cut at every 0.000005 (POINT B) seconds and fills in the missing data

BUT from what it seems everyone is telling me , it doenst work that way, it seems POINT B is cut, and POINTS A and POINTS C are squeezed,compressed together, therefore, losing the original position of POINTS A and C, therefore being unable to predict what POINT B originally was.

is this the way to think of it?

u see what i am trying to ask, pretty much, there is no way to predict the pattern of how a particular codec compresses a movie, and not just undo the compression but also fill it the deleted segments based on a predictive nature

Video compression doesn't work like that at all. Frames are not deleted.

Each frame is analysed and the patterns (say, areas of the same colour) and mapped out. And usually they also compare the frames with the previous frames, and store only the differences, since most video frames change only slightly from frame to frame (until you come to a "cut").

The method you describe is actually used to create intermediate frames when you want to change the fps -- eg from 15 fps animation to 30 fps. But its just a GUESS as to what an intermediate frame might have been, it isn't exactly what you would have got if you'd shot at the higher frame rate to begin with.

[Baldrick]

I removed some off topic crap about the source. Please stay on topic and JUST discuss why it's not possible to improve the quality from compressed video.

[RogerTango]

Basically, GIGO... the old computer programmer's term for "Garbage in, Garbage out"... The resulting video cannot be of better "quality" than the source. Some filtering can be done to do some smoothing and deblocking, however I have always found that the effort was not justified. Expectations may be high, but results are disappointing.

The only real solution is to obtain a higher quality original or be satisfied with the original you have now.

Cheers,
Andrew

[Dogpoundz]

Thanks everyone for replying to my thread, i appreciate all of the comments, ive learned allot