Average Lossless AVI Compression Ratio...

[Edited: For typo reasons]

I've developed a method of compressing AVI files (the video portion),
and want to set a goal to work with!  I've currently researched the
internet for current lossless video compression ratios, and I'm a bit
.  I've seen some claiming ranges from 4:1 all the way to 10:1
(which seems a bit too far fetched for most videos).

Does anybody know of a GOOD lossless video compressor that I can compare
to.  I'm going along the assumption that most lossless video compressors
result in about a 2:1 to 3:1 ratio on MOST videos.  Is this too far
fetched a goal to set?

Currently, all I'm comparing with in WinZIP and WinRAR.  So far, my
compression method (which surprisingly uses a variant of RLE encoding
that I came up with) beats both.  WinZip on average gave about a 1.6:1
ratio, WinRAR a 2.8:1 ratio, and mine gives about a 3.4:1.

Thanks...

Respectfully,
Eric D. Brown

"Eric D. Brown" <nospam@nowhere.com> wrote in message
news:10n0nqir9525qc7@corp.supernews.com...
> I've developed a method of compressing AVI files (the video portion), and
> want to set a goal to work with!  I've currently research the internet for
> current lossless video compression ratios, and I'm a bit .  I've
> seen some claiming ranges from 4:1 all the way to 10:1 (which seems a bit
> too far fetched for most videos).
>
depends, one can, eg, losslessly use motion compensation, which can help (if
for final encoding one is using arithmatic or huffman coding, motion
compensation can make a lot more values closer to 0).

> Does anybody know of a GOOD lossless video compressor that I can compare
> to.  I'm going along the assumption that most lossless video compressors
> result in about a 2:1 to 3:1 ratio on MOST videos.  Is this too far
> fetched a goal to set?
>
no, I got a bit better myself, around 1:5 to 1:7 for my test videos (crappy
webcam recordings of me walking around my room). I expect a little better
compression with higher quality video, just I would have to come by/create
some or such...

largely, one of the drawbacks was in places where there was no motion, still
having a lot of noise and similar which generally contributes to a larger
frame.

> Currently, all I'm comparing with in WinZIP and WinRAR.  So far, my
> compression method (which suprisingly uses a varient of RLE encoding that
> I came up with) beats both.  WinZip on average gave about a 1.6:1 ratio,
> WinRAR a 2.8:1 ratio, and mine gives about a 3.4:1.
>
that is likely because at least rle can exploit runs of the same value
pretty well. winzip uses deflate, which is based on lz77, which is not very
good for graphical data in many cases.

I was having some success with:
motion prediction (I split the frame essentially into the motions from the
previous to the current frame, and the difference between the real frame and
the predicted one);
a subtractive filter (for differences between adjacent pixels);
rle+huffman coding.

> Thanks...
>
> Respectfully,
> Eric D. Brown

> Does anybody know of a GOOD lossless video compressor that I can compare
> to.  I'm going along the assumption that most lossless video compressors
> result in about a 2:1 to 3:1 ratio on MOST videos.  Is this too far
> fetched a goal to set?

The yardstick i use for comparing with my own lossless video codec is
the ffv1 video codec which is a part of the ffmpeg project. You might
want to try it.

Eric D. Brown wrote:
> I've developed a method of compressing AVI files (the video portion),
> and want to set a goal to work with!  I've currently research the
> internet for current lossless video compression ratios, and I'm a bit
> .  I've seen some claiming ranges from 4:1 all the way to 10:1
> (which seems a bit too far fetched for most videos).
>
> Does anybody know of a GOOD lossless video compressor that I can compare
> to.  I'm going along the assumption that most lossless video compressors
>  result in about a 2:1 to 3:1 ratio on MOST videos.  Is this too far
> fetched a goal to set?
>
> Currently, all I'm comparing with in WinZIP and WinRAR.  So far, my
> compression method (which suprisingly uses a varient of RLE encoding
> that I came up with) beats both.  WinZip on average gave about a 1.6:1
> ratio, WinRAR a 2.8:1 ratio, and mine gives about a 3.4:1.
>
> Thanks...
>
> Respectfully,
> Eric D. Brown

Thanks moogie & cr88192 for your feedback... much appreciated.  I also
have another question (based upon researching), what exacted is bitwise
encoding?  I've tried searching for the meaning in a few places;
however, most explanations are somewhat vague.

Let's say I wanted to take the first bit of every byte and put them into
an array... continuing to the next bit value of every byte.  I then run
an RLE compression on those bit groups (again, based on the placement
that they appear in the byte).  Would that be called a bitwise encoding
method?

Thanks...

Respectfully,
Eric D. Brown

"Eric D. Brown" <nospam@nowhere.com> wrote in message
news:10n2qag5kugb182@corp.supernews.com...
> Eric D. Brown wrote: 
>
> Thanks moogie & cr88192 for your feedback... much appreciated.  I also
> have another question (based upon researching), what exacted is bitwise
> encoding?  I've tried searching for the meaning in a few places; however,
> most explanations are somewhat vague.
>
> Let's say I wanted to take the first bit of every byte and put them into
> an array... continuing to the next bit value of every byte.  I then run an
> RLE compression on those bit groups (again, based on the placement that
> they appear in the byte).  Would that be called a bitwise encoding method?
>
probably.

I would guess "bitwise" would sensibly refer to things taking decoded
input/output as a bitstream vs a byte or wordstream...

however, I am not very authoritative on matters of definition.