Your Tips, Tricks, and Tweeks for retaining A/V quality?

Compressing and encoding an analog source to digital inevitably degrades the original quality. What are your tricks for retaining as much of the original as possible???

Let's assume that the source is VHS and already captured to hard drive as .AVI, and the goal is to burn a DVD. What programs, filters, and settings do you use to get your AVIs to full resolution DVDs with as high quality as possible???

Is it better to pre-filter or post-filter? Which filters? What techniques save time as well as quality? If you deinterlace, do you blend or interpolate? Field Order? Motion Blur? Do you use a frameserver? CBR vs VBR? Does it require a dozen differnent programs or can you do it with just 2 or 3? And can you do it all without it costing an arm and a leg and all the rest of your five limbs?

Let's have it folks, what works...what doesn't.

 

The Best tool for me for Cleaning up video is Useing AVISynth... It has Literally Hundreds of different Filters and Thousands of Functions....

I usually Capture to AVI useing HuffYUV and then Useing AVISynth and a Few Noise Filters and sharpen filter to clean up any noise While Frameserveing to CCE or Procoder, and then Just author the Mpeg2 File to DVD...I allways Use VBR as you get the Same Quality with Smaller File Sizes as when useing CBR especially when useing Multi-Pass VBR which Increases the Quality and Lowers the File size for every Pass...

I sometimes De-Interlace mostly because even Interlaced Video Looks Bad on My TV Set, the De-Interlacer I use is usually a Field Blending, On an Adaptive de-interlacer or my Favorite is the Smart Bob de-interlacer.....


Cheers

 

i heard AVISynth is good for this stuff, but writing the scripts and stuff is just not my bag. are the noise and sharpen filters the only ones you're useing?

Do you pre-filter during capture? seems like a pre-filter would save some time, but i haven't gotten it to work with virtual-vcr or vdub. whenever i set virtual-vcr to use filters it just locks up and i have to restart my computer. in vdub, i can select pre-filters but the effects simply do not show up on the captured file.

the VBR and the deinterlaceing are settings on the mpeg encoder right? i have indead seen improvments in quality using these settings but the drawback is that they increase encoding time...

 

What is good if you don"t like to write Scrips is to use "Gordain Knot" as it will write Scrips for you as it has a Small GUI for AVISynth built in.....

I use AVISynth from anything from Noise reduction and Sharpening ,De-Interlaceing, Inverse Telicline, Frame Rate Conversions ,Resizeing ,Color Correction, Colorspace Conversion, adding Borders, Cropping and Many more things....

I haven"t had Much Luck with Filtering while Captureing either...Virtual-Dub Filters will only work if you capture in RGB Colorspace and Most Capture Cards Natively Capture in one of the YUV colorspaces Like YUY2 or YV12 ot UYUV and when Captureing in RGB and Filtering there is a Chance of Dropping Frames because Filters can use a Lot of CPU...

You can also use Virtual-Dub filters While Frameserveing because Virtual-Dub is also a Frameserver and you can use Virtual-Dub frameserver files with Tmpgenc and CCE and some other encoders but AVISynth files work with many more Programs...

There is also a way to use Tmpgenc filters to Frameserve to other encoders or editing programs so you can use Tmpgenc Filters But use CCE or Procoder to encode or use tmpgenc filters on files you want to edit in other programs Like Premier......

There is an exelent VBR Encodeing mode in CCE called "One Pass VBR" that is as fast as CBR and Produces awesome results with very Small File sizes simular to Tmpgenc"s "Constant Quality" mode But Tmpgenc is far too slow for me....

The only real other Way of Filtering becides trying to Filter while captureing would be to use the Filters that are in Editing programs but that way you either have to render to AVI again or you have to use the Mpeg encoder built into the Edining program what can not allways be the Best quality or very fast....

I still get very fast encodeing speed useing filters while Frameseveing , usually encode at 1.5 times real time when useing CCE SP.....


Cheers

 

that is some really good info. does gordian knot have a preview screen so you can tell what the changes are going to look like? cuz if it does i'm gettin' it...

but let me see if i understand this... AviSynth works with a special avi format that only some encoders will read, right? so you can't edit the files it makes but can you still edit the original avi from capture? err...i think i already know the answer to that question.

anyway, i will probably keep copy of Vegas 4 around for editing, but i am sick of using it for filtering and cleaning up the video. i tried to run sharpen and crop filters once, it was like 10x real time, rediculous! Do you know if i can frameserve the Mainconcept encoder plugin in vegas or do i need a stand-alone? would it be worth it?

 

Yes, Gordain Knot has a prievew , It uses Media Player Clasic for preview....
The GUI in Gorgain Knot only has a Few Filters ,Namely resizeing and De-Interlace and Noise filter but if you know the Script commands you can add filters and Functions to the Script...

I usually Save My Scripts and then Just edit them for every new File so i don"t have to write a New one for every file...I simply open the Script in Notepad and then Just delete the Old AVI files name and type in the New AVI files name and if I have to I just change a Couple of the Filter strengths in the Scrips and then save it and use it in my encoder, This way I don"t have to start from Scratch every time because for some reason on my system Gordain Know will not open every AVI file I try to open with it...
It also has problems Displaying a Preview with HuffYUV Encoded Files but you just have to add this Line:
ConvertToYV12()
As the First Function in the Script.....

You can also use Virtual-Dub to Test the Script before encodeing....

Gordain Knot is also usefull when Makeing Scripts for files that you download because it has a Good aspect ratio Calculator for Resizeing Wierd resolution Files to DVD Resolution without Stretching or Squishing the Video which Is a problem I have had with many Encoders...

All an AVISynth File is is Notepad with the Command you want to use written on it and saved with an .AVS Extention...
A typical Script would Look a Little Like this:

LoadPlugin("C:\AviSynth Filters\undot.dll")
LoadPlugin("C:\AviSynth Filters\FluxSmooth.dll")
avisource("C:\AVI Files\Girl.Interrupted.avi")
LanczosResize(720,384)
Undot()
FluxSmooth(7,7)
Sharpen(0.85)
AddBorders(00,48,00,48)

These Files can be Played with media Player and can be used as source files in encoders Like "CinemaCraft Encoder SP" and "Canopus Procoder 1.5/2.0" "Tmpgenc" and "Virtual-Dub" but they are not supported in all programs Like the MainConcept Encoder Has Problems with AVISynth files and most Editors like Premier and Vegas do not Support AVISynth frameserver files....

There are a Couple Frameservers for Vegas Video 4/5, Ones that work within Vegas so you can Frameserve a File that is on the Timeline with the effects and Transitions and Filters to any encoder which is a Lot faster than rendering to Mpeg with the editor or rendering to AVI and them encodeing to Mpeg....
The freeware One for Vegas is called the "DebugMode Frameserver" I haven"t had that much luck with it...
There is also another one that is supposed to be really good but it costs something like $30 I think...

As I have been writeing this I have been checking what editing Programs will open AVISynth Frameserver files and So far I have tried vegas and premier Pro and both do Not open the Files but a Program called "Video Edit Magic 3.37" does open the Files..
But most encoders do support them accept Mainconcept encoder sometimes has problems with them But I just tried it with 3 different AVS files and it worked with all 3....

If you want to learn more about it go here :
http://www.avisynth.org/

Cheers

 

you're a good man Minion, lots-o-info. but before i dive into avisynth, i'd like to just get my feet wet.

i'd like to try frameserving with vdub just to see how it all works. i saw your post from the "Burning VHS's to DVD's" thread which gives me a place to start. i was looking at the filters in vdub and they seem kind of limited, are there others i can load to get more functionality out of vdub? is there a mod i should pick up? it doesn't seem to even have a good crop filter...

you were mentioning tmpgenc as a frameserver before, should i look into it for better filters and what not?

should i use the filter called "field bob", they mentioned it on the gordain knot website. they used that along with "null transform" and some others, what do you recommend for vdub?

 

The "Null Transform" filter is a Crop filter...So you Just select the "Null Transform Filter" and then the "Cropping" Buttons will become available...
the Field Bob filter is a De-Interlacer which turn your Interlaced footage into a True Progressive Image it is one of the Better de-interlace filters....

There are other V-Dub filters you can download off the Net ,Just do a Google Search....

It is better when Frameserveing with V-Dub when you have a Lot of Filters to extract the audio first and use it as your audio source instead of Frameserveing the audio because when useing a Lot of Filters there can be a Lag between the video and audio serveing which can cause a Sync problem....
I allways Process my audio seperately anyways because I encode it to dolby AC3 with Vegas 5 for DVD authoring.....
I don"t use V-Dub for Frameserveing Very often so I haven"t tried many of the filters available for it....

 

thanks for the tips on filters but now i am a little confused about vdub. let me know if i have the steps right:

1. capture to avi; 2. extract audio; 3. add filters and setup frameserver; 4. encode video to mpeg2; 5. remux the audio somehow??? (does this involve re-encoding the mpeg?) 6. author to DVD.

can you add the audio again during authoring or something? cuz it seems that extracting and then adding the audio is another step similar to if you rendered the avi again before encoding...

 

Most Good DVD authoring Programs will allow you to add seperate audio and Video files to Make your DVD..
The DVD authoring Program I use actually prefers seperate audio and Video streams and will de-Mux any Mpeg files that aren"t in seperate streams...The DVD authoring program will Mux the seperate streams into Vob files when Compileing the DVD....

Extracting the audio Before hand useing Virtual-Dub only takes a Couple Minutes....

You can also Capture to AVI and then Extract the audio and then set up your Filters for Frameserveing the Video and load the Frameserver file into the encoder as the Video source and load the Extracted audio as the audio source in your Encoder and you will end up with a Mpeg File with audio and Video, That is if you select Mpeg1 layer 2 audio as your audio format as Most encoders will not Mux Mpeg video and Wav audio files into Mpeg files....

DVDLab and Tmpgenc DVD author will both accept seperate audio and Video files and so will Most Good DVD authoring Programs ....

Cheers

 

well it can't get any easier then that... "everything is comin' up Milhouse!"

i extracted the audio and added filters with vdub then i setup the frameserver to tmpgenc. works great. i'll remux the audio with DVD architect. so that's all set. but what i'd like to do is use Procoder 2.0 for encodeing but vdub won't serve procoder...

you mentioned using tmpgenc for frameserveing, what file extention do i need to set the tmpgenc output so that procoder will read?

*edit*
hey i picked up the "vfapi converter", it seems like it should convert my tmpgenc projects into pseudo avi files that procoder will read. there aren't any instuctions for the converter so i am not really sure what i'm doing. procoder says it can't read the file i created with the converter...

when i added the audio stream to my tmpgenc project and converted, procoder gave this error: "AVI Reader: An interface has too many methods to fire events from."

ideas?
_
_X_X_X_X_X_[small]i like to eat turkey...[/small]

 

i've been moving ahead on this in leaps and bounds...

i got procoder to read the pseudo avi files by loading the vfapi codec from the vifpset.bat file. the test video i encoded with procoder ran at 0.4 times real time!

also i found that i can run the vdub frameserver into tmpgenc then save the tmpgenc project and convert to vfapi avi. then open and encode with procoder... essentially adding vdub and tmpgenc filters simultaneously. i still have to figure out how that is going to be useful to me, but in either case it is still cool!

 

Well it seems that you have Figured out the Tmpgenc,VFAPI converter thing for your self....

Since you have figured out a couple Frameserveing Methods you should maybe try useing CCE SP to encode as it is so Much faster than any other encoders and produce simular if not better quality and it accepts Virtual-Dub frameserver files which procoder doesn"t...

When Frameserveing from virtual-Dub i can encode in 1.5 to 2 times Real Time so when useing CBR or One Pass VBR I can encode a whole 2 Hour movie in Just over an Hour compared tom many times that when useing Procoder....

Cheers

 

CCE SP price tag is a bit steep... don't have the $2000 just lying around at the moment. but what is your pick for second best mpeg encoder? can mainconcept beat procoder in raw speed if quality is the same? is the CCE basic worth a look?

like i was saying before thou, i can use tmpgenc as a work around to get vdub to frameserve procoder. the work around happens to also work with Vegas. still, i am certainly not producing encoding times like 1.5 times real time - let alone 2 times. both procoder and vegas (mainconcept plugin) are running around 0.3 to 0.4.

 

I'm glad I found this post as I'm too looking for ways to clean edit my HuffYuv captured AVI from VHS sources that are often not of such great quality.
I would like to clarify a few points. Minion if you would?
1. You wrote: "There is an exelent VBR Encodeing mode in CCE called "One Pass VBR" that is as fast as CBR and Produces awesome results with very Small File sizes...". I'm not sure I get your meaning here.
2. You mentioned Undot and FluxSmooth (btw, why 7,7?)
I'm trying them both now.
Also, you added Sharpen(0.85). I didn't see a DLL associated with this command in the AVS example. Is there one? why 0.85?
3. Can you recommend any other cleaning filters that no self respecting, captured VHS source, should do without?
4. Can you put the record straight concerning
ConvertToYUY2()? I read this: http://www.avisynth.org/Convert
but didn't fully understand: do I need to use ConvertToYUY2(interlaced=true) when working with AVI from VHS which is interlaced or can I leave it as ConvertToYUY2()?

Thanks in advance and my appreciation for all the great help you give here in abundance.

 

Update: DVD2SVCD stuck in mid process so I checked and apparently Smoothflux has two options as found in here: http://www.indeus.com/sansgrip/avisynth/FluxSmooth-readme.html
Do I understand correctly that FluxSmoothST is preferable? Anyway, that's what I'm using, in hope it produces better looking video.

 

Hi, I"ll try to answer some of your Questions....

#1)Well "CinemaCraft Encoder SP 2.67" has a encodeing method Called "One Pass VBR" ...In this encodeing mode you select the Minimum Bitrate and the Maximum Bitrate and another setting called the "Q" setting ...("Q" Means Quantization)

When useing this encodeing Method for Mpeg2/DVD encodeing I set the Max Bitrate to 7000KBS and the Minimum BItrate to 2000kbs and the "Q" level to "30"
And the Files produced useing these settings(And some other settings) I can encode a 90 Minute full resolution File to Mpeg2 with a File Size of about 2.0gb with Great Quality and in under an Hour....

#2) The FluXSmooth.dll Filter is a Noise Filter and with a setting of (7,7) you get a Medium amount of Noise Reduction...If you set it to (10,15) then that is a Lot of Noise reduction but it will also make the Image Blurry....
The UNDOT.dll is also a Noise Filter that Helps eliminate Static Dots on old Video, It is actually Quite effective....
The Sharpen Function is one of the Filters that is Built into the AVISynth DLL..
Why did I choose (0.85) for Sharpening well because That is a Good setting to Sharpening the Image without Over Sharening the Image which look Bad..
I think the Sharpen Filter goes all the way from (0.01) to (1.13 ...You should allways use some Sharpening when useing Noise Filters because they can Blurr the Image a Bit....

#3)You can see a whole list of AVISynth filters here:
http://www.avisynth.org/warpenterprises/
Check out the "Convolution3D.Dll" and the "TemporalSmoother.dll"....

#) Most of the Video I work with is Progressive so when I actually have to convert the ColorSpace I can simply write "ConvertToYV12()" But for Interlaced Footage you cave to use :
ConvertToYV12(interlaced=true)....

You only convert Colorspaces if your Source Video is one Colorspace but you use a Filter that doesn"t work with your Colorspace and then you would add the Line to Convert the Colorspace....
You should Only do it if you have too because it can really slow down the encodeing...


I hope it helped...?

 

Actually CCE has 3 VBR Modes and 2 Multipass Modes and 2 CBR Modes..

It has the "One Pass VBR" setting
It has the "Multipas VBR" Setting
It has the "First Pass of Multi-Pass" setting....

It also has a Multi-Pass CBR setting....