should you use filters if compression ratio is high?

[Carpo]

as the title says should you use filters if you want to kep a long movie - 3h 20 mins in this case on one disc - its a 2 disc dvd and dont really wanna split it onto 2 dvdrs if i can help it.

ratio is
Reduction Level for DVD-5: 55.2%
- Overall Bitrate : 2,798/2,238Kbs
- Space for Video : 4,068,436KB
- HIGH/LOW/TYPICAL Bitrates: 8,808/1,972/2,238 Kbs

would it be adviseable to use a filter on compression as high as this and if so which filter (i have searched but not found any definitive answers - so sorry if this has been asnswered and i have missed it :( )

[writersblock29]

@Carpo

Three hours and twenty minutes? I'd probably use Undot and Deen on this one. How clean's the source, though? If it's pristine (without a lot of grain or noise) then you might encode a bit without filters, then stop it to watch an M2V or two in order to decide if you're going to like what you see. The question of whether or not to filter ALWAYS depends on the source; I tend to run more filters on older, grainy, 90 minute movies than I do on longer, more recent (and usually cleaner) titles.

[Carpo]

ye i thought of undot and deen but after reading - there is no grain on this one - rather newish film - and i thought deen was a smother - or is that the point - as i have read on diff forums that a few ppl have said to use that with high compression films.

i think i'll try it with just deen() and see what happens - just going to have to find an encoder that will do more than 2 passes as a few peeps have said 3-4 passes on this film - dont think my cce basic or hc up to this ;P

[writersblock29]

@Carpo

{Beyond 2 passes}

You may be surprised by CCE or HC as far as 2 passes go. I'd be willing to wager that if I sat 10 people down to compare footage -- side by side and on identical equipment -- encoded with 2 passes for one example and 10 with the other, 9 of those viewers wouldn't know which is which. I'd try a small encode at 2X before combing around for another encoder.

I used to do 3 passes for longer movies religiously... but forgot on one project, leaving it at 2 passes (which I usually use if I'm dealing with <3 hour material). Looked great, so now I really don't bother with anything more than 2 passes anymore. If your source is clean, your encoder really won't have much to figit with on each additional pass. It's all subjective, yes... but I'd at least do a trial run with 2 passes to see what you think.

[Carpo]

already doing that - thought b4 i go testing another encoder why not see how it will look with the ones i have - cce 2 passes it does look a bit iffy think poss undot my be required - gonna run hc on it now and see what that comes up like
as i have used hc in the past and had rather good results from low bitrates (gonna give is vbr_bias 0 see if full swing will help it out - wonder if should get the beta of 0.16) - if not i'll give procoder a go :)

who says you cant teach an old dog new tricks :devil:

[writersblock29]

@Carpo

I'm sorry! I forgot to answer that yes, Deen's a smoother. Since encoders work by detecting movement (differences from frame to frame), film grain and pixel dance on the source are interperated as movement by your encoder, since these things change from frame to frame. The encoder tries to replicate the grain/pixelation, raising the demand for bitrate and possibly causing pixelation all its own in the process. By using a smoother, the source has less grain and pixelation (saddly, it also takes a sharpness hit and will make you loose a slight bit of detail), therefore more of your encoder's limited bitrate is used to replicate the main action of the film. There is a slight softness that's added to your project by using a filter such as this... but that softness is usually less obtrusive than any added pixelation caused by a low bitrate. In fact, it's usually only noticeable if you're comparing your copy to the original, I've found.

[Carpo]

hc looks to be doing a better job than cce - with and with out deen - without looks slightly better :/

EDIT: after a few tests i think im going to go with HC :)

[Mr. Monte]

Carpo, using HC..did you use one of the built in lower bitrate matrixs?

How much better?

[dannyv]

Hi guys,

On the topic of filters I was hopeing to get some help. I am doing the babylon 5 dvd series and the original transfer is very grainy and I was thinking of using undot and deen on it to clean it up or smooth it a bit but am confused on how to use deen. I know undot is straight forward

undot()

But I beleive deen needs some paramaters

deen(????)

I've read the text that was packaged with the app and am confused. Can someone explain what paramaters I should use and maybe give me an example. I've done some searching on the subject and could not find a clear answer.

[jdobbs]

Here's what I put when I use it:

undot().deen()

[writersblock29]

Same here. Never had any complaints or issues with the resulting streams.

[Rockas]

You can access to the advanced parameters of Deen on RME... but be carefull... you can double your encoding time :)

[FredThompson]

hc looks to be doing a better job than cce - with and with out deen - without looks slightly better :/

EDIT: after a few tests i think im going to go with HC :)2-pass HC gives a sharper image than 5-pass CCE and is a little quicker. 2-pass CCE is quick but...you get what you "pay" for. 2-pass ProCoder is also very good and very slow.

neither deen nor undot are technically degrainers. they're general smoothers. Fizick's DeGrainMedian is a spatio-temporal smoother specifically for grain removal. Does it matter? Maybe not. Maybe it gives a better result on grain. I'm testing it again Heavy Metal which is a horribly grainy transfer.

Peach is another one to try as it is self-adapting. the undot/deen combination smoothes no matter what.

[Harrysmiith]

Here's what I put when I use it:

undot().deen()


Oh !x$%& just over 5 hours ago I typed

undot()
deen()

and have just reached making ISO stage

will both filters have worked or should I have typed

undot().deen() on the same line with the "." ?

[jdobbs]

Same thing... it's fine.

[Harrysmiith]

Same thing... it's fine.

Thanks - have just watched the result. At 71.1% with undot and deen the result is better than the original. I had expected the output to be less grainy but also less sharp. To my eyes it is actually less grainy and slightly sharper - which is what I wanted.

[dannyv]

Here's what I put when I use it:

undot().deen()

Just viewed one of the mpv files and it seems to have eliminated almost all of the graininess but seems to have overly softened the picture. The closeup scenes look fantastic but the senic shots seem to be softened so much it almost looks blured. I'm wondering if I should use a sharpening filter as the last filter in the string or just use one or the other but not both filters at the same time. Any ideas or sugestions. I finished the entire project this morning but did not have time to really compare it to the original so I'll report back tonight.

[Carpo]

undot will remove grain - deen is a smoother and thats why things look a bit blury

after tests i have been doing i find using undot on most things is ok (except where the film should be grainy - there are some) and use deen when compression is high.

There is a filter that will sharpen think its sharpen() or limitedsharpen()

not too sure