1st Xvid Encode

[joshbm]

Hey all,

First to make things clear: I bought Spiderman 2 2.35:1 widecsreen edition and I am making sure I did everything correctly in the transfer from MPEG-2 (DVD) to MPEG-4 (Xvid).

I cropped and resized (don't know should I have resized it? From 1.78 AR to 1:1 AR Lanczos3Resize) the original source from 720x480 to 720x304 and I used 2nd pass with a target filesize of 1224197 kbytes. The quality appears very good. My goal is to be able to store 3 Xvid movies on 1 DVD (4.7GB) while retaining *excellent* quality. I was just wondering would it have been best, or should I say is it a common practice to use AviSynth and an AviSynth filter like Convolution 3d? Also is there any other filters I should apply on the original MPEG-2 source from DVD while encoding to Xvid?

The video is encoded ASP@L5, H.263 matrix. I was wondering if a different matrix would be better as well.

PS- My first encode is about 1.5 GB (2:07:25).

Thanks!
- Joshbm

[Teegedeck]

No, should don't need to denoise a movie when there is no noise in it. You don't even need to denoise if there's a little noise. You've done it right. But you can denoise if you like. I for one prefer a 'swimming effect' as XviD produces it in noisy areas at medium compression over the sight detail removal that almost all denoisers produce (haven't tried Didee's 'Seasaw', yet).

You should always use colormatrix() when transcoding from DVD, it corrects colour rounding that is different from MPEG-2 to MPEG-4.

[joshbm]

Just add the command colormatrix()? I didn't download GK and I am just using VDubMod, DGIndex, and AviSynth. It works well. I've been doing some more reading around these forums and my movie is 1306kbps. I read the thread with different matricies and tried EQM V3ULR rev3. Should I have used this matrix over H.263 for something like film? It has high action sequences by the way. Setting a goal of 1224197 kbytes for the movie (1642769 kybtes total=3rd of DVD) was so I could fit 3 Xvid movies on one DVD. I think I may reconsider and actually do 2-per 4.7GB DVD (2045581 kbytes for movie, and 2464153 kbytes total), because I'd like to have superb quality but at least half the size of a regular movie (so I am able to store more movies on my hard-drive as well as per DVD).

Are there any recommendations on a matrix for both 3-per DVD and 2-per DVD? I have downloaded the entire EQM V3 set and I am looking at them and I have searched the forums.

Thanks for your help so far Teegedeck!
- Joshbm

[Teegedeck]

You can get colormatrix here: http://www.avisynth.org/warpenterprises/

Just put colormatrix() into the second line, directly after mpeg2source("...").

With 3 movies per DVD-R it shouldn't be necessary to resize with EQM v3ULR.

You should try even v3LR for better quality. That depends a bit - as a general rule, I guess v3LR is a good choice if you aim for 25-30% of the size of the original VOB-files (without audio of course). So it should work for 3 movies per DVD-R. For more precise prediction seach the forum for the filesize-prediction tool 'Enc'. I'd say if Enc predicts that v3LR at quant=5 or 6 produces a small-enough file, then use v3LR.

If you put only 2 movies onto a DVD-R, i.e. you have something like ~1700-2000 MB for the video, my recommendation is the SixOfNine matrix. That's again quite a leap in quality (at least if you view the movie on a TFT or plasma screen). Quality should be transparent (indistinguishable from the original) in most cases when you encode movies to XviD at a filesize of ~35-40% of the original VOB-files. The thread 'XviD Presets...' (http://forum.doom9.org/showthread.php?t=107897) could be a worthwhile read for you.

[joshbm]

Doom9 is such a great community, thanks (reading).

[edit]
Looking for the SixOfNine-HVS matrix. I cannot find it.

[edit 2]
Found them... What is the difference between SixOfNine and SixOfNine-HVS?

[edit 3]
http://forum.doom9.org/showthread.php?t=104041&highlight=SixOfNine+matrix .... What was Didee saying (last post):


BTW, we can put all EQMxxx's and 6of9's in the trashbin, anyway. Just today it was revealed that H.263 gives better quality at half the bitrate ... d'oh! Having known this, I could've made all those anamorphic full-size encodings at 600 kbps! <snort>


H.263 gives better quality than SixOfNine or EQMxxx?

- Where do I change the settings: 1st pass @q=3, curve-compr. H10, L3, quantizer-restrictions min. 2, max. 6 ? quantizer-restrictions min 2 max 6 for I, P, and B frames? Does q="Quantizer ratio"?

[shon3i]

H.263 gives better quality than SixOfNine or EQMxxx?SixOfNine is not for 1CD encodings.

1st pass @q=3In zones you have select quantizer and put 3

curve-comprin second pass more option

Try this settings for quantizers

I min 2, Max 4
P min 2, Max 6
B min 2, Max 6

bvops = 1
bratio = 1
boffset = 1

Matrix Sharktooth EULR V3

[Sharktooth]

H.263 gives better quality than SixOfNine or EQMxxx?it was sarcasm...

[Teegedeck]

Doom9 is such a great community, thanks (reading).You're welcome!
Found them... What is the difference between SixOfNine and SixOfNine-HVS?Hm, did I say SixOfNine-HVS? Humm, no, apparently I said 'SixOfNine'. So I guess I was talking about SixOfNine. :)

SixOfNine-HVS needs less bits but is not so outstandingly sharp, so I don't really use it.
H.263 gives better quality than SixOfNine or EQMxxx?Guuuhhhhhrgh.... :scared: Where again is that info? Nowhere? Right, nobody said that. Horribly, horribly wrong and nobody said that. So let's better forget about it right away, okay? :) Puh-lease!
- Where do I change the settings: 1st pass @q=3Zones. But if you do that you'll have to use different settings for the first pass in or it'll be as slow as the second pass. So perhaps you shouldn't do it.curve-compr. H10, L3encoding type: two-pass: more quantizer-restrictions min. 2, max. 6quality preset: user-defined: more-->quantization quantizer-restrictions min 2 max 6 for I, P, and B frames?yepp, all the same quantizer. But that was only meant to produce oversizing or undersizing if people used a preset that doesn't fit their encoding targed. By producing over- or undersizing... Mean idea, isn't it? Does q="Quantizer ratio"?Pardon? Come again? :confused:

Edit: I'm a bit late, ain't I? ;) As for quantizer-restrictions, Didée has enlightened me that b-frames quant-restrictions work like 'max. quant x quantizer-ratio'.

[joshbm]

Doom9 never fails. Thanks a lot guys! I will be doing some encoding :-p tonight! You've guys have created a new MPEG-4 encoder ;-). I just need to find a nice portable MPEG-4 player with a 7" screen that supports these encodes.

Thanks,
Joshbm