Log in

View Full Version : help with avi filters..make my encodes look nice :P


kimi_mo
9th September 2004, 18:39
hi,
can someone please help me
i recently started encoding in ogm
and i usually encode 1st pass ogm no sound or subs
and 2nd pass sound and subs added
straight in to .ogm
my friend does it to avi first
is there any quality difference between them?

and i am going to list my avs filters i use,
maybe you can reccomend me some more that will help me greatly
my filsize for my ogm files
is 250 MB with very bad quality
and i want it to be around 200 - 220 MB With Good Quality

--------Filters-------
_2DCleanYUY2_for_25.dll
aWarpSharp.dll
Dup.dll
MaskTools149.dll
MipSmooth.dll
RemoveDirt.dll
tre.dll
VagueDenoiser.dll
WarpSharp.dll
atc.dll
Decomb511.dll
libcurl.dll
MPEG2Dec3.dll
UnDot.dll
VSFilter.dll
zlibwapi.dll
----------------------
please tell me if theres anymore filters i should be using
or any of these filters i Shouldnt be using

thanks very much
;)

Mug Funky
9th September 2004, 19:28
if you use all those filters there'll be nothing left :O

subtlety is the key here - you want to remove noise, but not detail - the codec itself will decide what detail it wants to remove, and if it's a good enough codec it wont need help from smoothing filters.

lately i've been using removegrain then removedirt. possibly use a script to prevent removedirt b0rking on earthquake motion, but it's not really that bothersome an issue (you can find a dodgy one i scraped together in the removedirt thread. search is good).

source -> avi -> ogm will be the same as source -> ogm, provided there's no actual re-encoding taking place. otherwise it'll be a lot worse.

it's like putting a painting in a different frame - it's still the same painting, but one frame may be slightly bulkier than another, but on the other hand the less bulky frame might not work in every gallery :)

i'm still using avi at the moment, just for simplicity. subtitles are in different files with the same name, yada yada. the overhead isn't a big deal to me, as i'm not a distributor, but just a "fair-user".

[edit]

btw, if you want to get smaller files you'll have to either:

- resize downward
- use a more efficient codec for low bitrates

ogm doesn't determine quality - it'll only give you 1 or 2 megs more headroom per CD.

for medium bitrates Xvid is good

for low bitrates RV10 is good (some say it's also good for medium, and are probably right, but i just don't use it for no particular reason)

for high bitrates anything goes... maybe Xvid with mpeg matrix, or mpeg-2 for really high bitrates.

don't touch WMV. it's naff. i don't go near it simply for political reasons, but as it turns out i'm not missing much. go to RV10 before you consider WMV.

i hear good things about h-264 and vp6, but haven't tried either seriously (and good codecs for those standards are hard to find and even harder to pay for).

kimi_mo
9th September 2004, 21:01
cool, thankyou
so any other filters anyone reccomends
thanks alot
:D

analfilter
10th September 2004, 04:19
Also depends on the source...if you're encoding from a somewhat clean source, like a dvd, then like Mugfunky said, subtlety is key. You pretty much don't need any filtering at all. But if you're doing analog tv episodes with lots of noise, then sometimes you need to hit it over the head with some heavy filtering. I've been doing all kinds of experimenting with this, but it's mostly up to personal preference; i.e, i like a much smoother image with some loss of detail rather than noise distracting me from any details I otherwise would have obliterated. Recently, I've started using DePan w/FluxSmoothT to "supposedly" eliminate some noise, then hit it over the head with some SpaceDust or GoldDust to smooth it out, depending on how much noise is left over. Dust gives a smoother image on default than other filters, which is why alot of people like it on analog captures. The only thing is it requires YUY2 input, which may make for some messy colorspace conversions in your script.

One other thing, decombing/deinterlacing properly is also key, if you do it at all. I dunno the benefits of encoding with an interlaced source, maybe one of its proponents can explain this. But when I was even noober than I am now, I was FieldDeinterlacing all of my captures, even though I eventually found it was all IVTC 3:2 pulldown. So now I just Decomb/Decimate before I filter. I hope this helps, just my $.02.