Fawlty output?

[tgbyhn10]

I've been trawling through these forums for days now and I'm at my wits end. Would someone please help a newbie?

I use DVDShrink a lot and I'm happy with what it produces but some of my larger DVDs (mainly episode DVDs) look awful when shrunk. Having read all the rave reviews about DVD-RB and CCE basic I decided to take the plunge and spend £50 on them but from the encodes I have done so far, I fear I may have wasted my money.

I know it's early days and there are lots of things I can try like filters etc and I shouldn't expect miracles but the primary reason for buying the software was to deal with discs that require heavy compression like Fawlty Towers & Rising damp etc. but these still look awful - better than the best Shrink can produce but still awful.

Knowing how popular Fawlty Towers is, I assume a lot of people have had a go at re-encoding this title with DVD-RB and CCE Basic or HC. With this in mind I was hoping someone could walk me through this encode and suggest which encoder / settings I should use or better still let me know how they did it.

The disc I am trying to encode is a region 2 version of ‘Fawlty Towers Series 2’ (disc 1). Link below for reference.

http://www.amazon.co.uk/exec/obidos/ASIN/B00005NGUQ/ref=pd_bxgy_img_2_cp/026-2467494-3364452

I have done about 6 encodes with hanks HC .17 encoder set to ‘Best’ and CCE Basic with default settings. With CCE; I have also tried lowering the Vbr_bias to 15 which didn’t help much and also changing the selected quantization matrices but I don’t really understand how these work. Should I leave them all at 'encoder default' or should I set 'low bitrate matrix' to 'low bitrate', 'very low bitrate matrix' to 'very low bitrate' etc. There is very little information on this that I can find apart from expert speak that goes way over my head and the manual is no use at all.

I am not a purist and I am not finatical about keeping output as close to the original as possible so using filters is fine. I just want something that I can watch and enjoy so a bit of smoothing / smudging doesn't concern me at all. Halo effects and mosquito noise on the other hand really irritate me, mild blocking irritates me less.

I have been using undot().deen() to try to improve the encodes, is this helping? Should I adopt a more aggressive undot filter?

As things stand I prefer Hanks HC 0.17 'Best' output to watch over CCE Basic with default settings as there is less haloing and mosquito noise but the CCE output looks much cleaner and I think would stand a better chance of being improved by filters. Are filters the way to go? Are there post processing filters I can apply to remove halo effects and mosquito noise? Solids look great with CCE, I can’t see any blocking at all; should I reduce Qual_Prec?

While I am not happy with any of the encodes, they are very close - I’m not really that fussy, honestly.

The average bitrate seems OK at 2900 so why does the output looks so bad? Is it because the original is already compressed?


Before I did any encodes, I reduced the data size by doing the following:

1. Ran the DVD through DVDShrink to shrink the menus to the minimum size permitted (no compression performed on anything else)

2. Ran the output from step one through menushrink to decrease menu size (convert menus to stills).

3. Used vobblanker to remove the BBC and copyright intros.

4. Stripped all redundant audio / subtitle streams.

It's as small as I can get it. 7.05GB (3 hours 14 mins according to dvdshrink) down from 7.75GB.

Should I be able to get a good copy of this DVD or am I wasting my time?

Also, is there any way I can just encode 10 mins of this dvd and preview that instead of having to wait 4-5 hours?

I have downloaded CQME and DVD-RB Matrix editor and lots of Matrices like Bach1 etc but not messed about yet. To be honest I don't really want to, I'd like to use DVD-RB with as many of the presets as possible - after reading these and other forums for the past week or so, my brain's about ready to explode as it is.

Help! I'd hate to think I had wasted £50.


Thanks, and sorry for the long post.

Pete.


P.S. I’m using DVD-RB 1.08.1 and all the default bits it installed.

Here's a log:


[22:20:07] Phase I, PREPARATION started.
- CCE 2.70.1.9 encoder selected.
- AVS Filters are enabled.
- VTS_01: 3,698,940 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 291,322 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 58.9%
- Overall Bitrate : 2,921Kbs
- Space for Video : 4,154,828KB
- HIGH/LOW/TYPICAL Bitrates: 3,208/410/2,921 Kbs
[22:27:03] Phase I, PREPARATION completed in 7 minutes.
[22:34:18] Phase II ENCODING started
- Creating M2V for VTS_01 segment 0
- Creating M2V for VTS_01 segment 1
- Creating M2V for VTS_01 segment 2
- Creating M2V for VTS_01 segment 3
- Creating M2V for VTS_01 segment 4
- Creating M2V for VTS_01 segment 5
- Creating M2V for VTS_01 segment 6
- Creating M2V for VTS_01 segment 7
- Creating M2V for VTS_01 segment 8
- Creating M2V for VTS_01 segment 9
- Creating M2V for VTS_01 segment 10
- Creating M2V for VTS_01 segment 11
- Creating M2V for VTS_01 segment 12
- Creating M2V for VTS_01 segment 13
- Creating M2V for VTS_01 segment 14
- Creating M2V for VTS_01 segment 15
- Creating M2V for VTS_01 segment 16
- Creating M2V for VTS_01 segment 17
- Creating M2V for VTS_01 segment 18
- Creating M2V for VTS_01 segment 19
- Creating M2V for VTS_01 segment 20
- Creating M2V for VTS_01 segment 21
- Creating M2V for VTS_01 segment 22
- Creating M2V for VTS_01 segment 23
- Extracting STILLS for VTS_01 segment 24
- Creating M2V for VTS_01 segment 25
- Creating M2V for VTS_01 segment 26
- Creating M2V for VTS_01 segment 27
- Creating M2V for VTS_01 segment 28
- Creating M2V for VTS_01 segment 29
- Creating M2V for VTS_01 segment 30
- Creating M2V for VTS_01 segment 31
- Creating M2V for VTS_01 segment 32
- Creating M2V for VTS_01 segment 33
- Creating M2V for VTS_01 segment 34
- Creating M2V for VTS_01 segment 35
- Creating M2V for VTS_01 segment 36
- Creating M2V for VTS_01 segment 37
- Creating M2V for VTS_01 segment 38
- Creating M2V for VTS_01 segment 39
- Creating M2V for VTS_01 segment 40
- Creating M2V for VTS_01 segment 41
- Creating M2V for VTS_01 segment 42
- Creating M2V for VTS_01 segment 43
- Creating M2V for VTS_01 segment 44
- Creating M2V for VTS_01 segment 45
- Creating M2V for VTS_01 segment 46
- Creating M2V for VTS_01 segment 47
- Creating M2V for VTS_01 segment 48
- Extracting STILLS for VTS_01 segment 49
[03:14:46] Phase II ENCODING completed in 280 minutes.
[10:43:12] Phase III, REBUILD started.
- Copying IFO, BUP, and unaltered files...
- Processing VTS_01
- Reading/processing TMAP table...
- Rebuilding seg 0 VOBID 1 CELLID 1
- Rebuilding seg 1 VOBID 1 CELLID 2
- Rebuilding seg 2 VOBID 1 CELLID 3
- Rebuilding seg 3 VOBID 1 CELLID 4
- Rebuilding seg 4 VOBID 1 CELLID 5
- Rebuilding seg 5 VOBID 1 CELLID 6
- Rebuilding seg 6 VOBID 1 CELLID 7
- Rebuilding seg 7 VOBID 1 CELLID 8
- Updating NAVPACKS for VOBID_01
- Rebuilding seg 8 VOBID 2 CELLID 1
- Rebuilding seg 9 VOBID 2 CELLID 2
- Rebuilding seg 10 VOBID 2 CELLID 3
- Rebuilding seg 11 VOBID 2 CELLID 4
- Rebuilding seg 12 VOBID 2 CELLID 5
- Rebuilding seg 13 VOBID 2 CELLID 6
- Rebuilding seg 14 VOBID 2 CELLID 7
- Rebuilding seg 15 VOBID 2 CELLID 8
- Updating NAVPACKS for VOBID_02
- Rebuilding seg 16 VOBID 3 CELLID 1
- Rebuilding seg 17 VOBID 3 CELLID 2
- Rebuilding seg 18 VOBID 3 CELLID 3
- Rebuilding seg 19 VOBID 3 CELLID 4
- Rebuilding seg 20 VOBID 3 CELLID 5
- Rebuilding seg 21 VOBID 3 CELLID 6
- Rebuilding seg 22 VOBID 3 CELLID 7
- Rebuilding seg 23 VOBID 3 CELLID 8
- Updating NAVPACKS for VOBID_03
- Rebuilding seg 24 VOBID 4 CELLID 1
- Updating NAVPACKS for VOBID_04
- Rebuilding seg 25 VOBID 5 CELLID 1
- Rebuilding seg 26 VOBID 5 CELLID 2
- Rebuilding seg 27 VOBID 5 CELLID 3
- Rebuilding seg 28 VOBID 5 CELLID 4
- Rebuilding seg 29 VOBID 5 CELLID 5
- Rebuilding seg 30 VOBID 5 CELLID 6
- Rebuilding seg 31 VOBID 5 CELLID 7
- Rebuilding seg 32 VOBID 5 CELLID 8
- Updating NAVPACKS for VOBID_05
- Rebuilding seg 33 VOBID 6 CELLID 1
- Rebuilding seg 34 VOBID 6 CELLID 2
- Rebuilding seg 35 VOBID 6 CELLID 3
- Rebuilding seg 36 VOBID 6 CELLID 4
- Rebuilding seg 37 VOBID 6 CELLID 5
- Rebuilding seg 38 VOBID 6 CELLID 6
- Rebuilding seg 39 VOBID 6 CELLID 7
- Rebuilding seg 40 VOBID 6 CELLID 8
- Updating NAVPACKS for VOBID_06
- Rebuilding seg 41 VOBID 7 CELLID 1
- Rebuilding seg 42 VOBID 7 CELLID 2
- Rebuilding seg 43 VOBID 7 CELLID 3
- Rebuilding seg 44 VOBID 7 CELLID 4
- Rebuilding seg 45 VOBID 7 CELLID 5
- Rebuilding seg 46 VOBID 7 CELLID 6
- Rebuilding seg 47 VOBID 7 CELLID 7
- Rebuilding seg 48 VOBID 7 CELLID 8
- Updating NAVPACKS for VOBID_07
- Rebuilding seg 49 VOBID 8 CELLID 1
- Updating NAVPACKS for VOBID_08
- Updated VTS_C_ADT.
- Updated VTS_VOBU_ADMAP.
- Updated IFO: VTS_01_0.IFO
- Updating TMAP table...
- Correcting VTS Sectors...
- Building ISO Image...
- ISO Image successfully created.
[11:19:50] Phase III, REBUILD completed in 36 minutes.

Done.

[manono]

Hi-

Fawlty Towers, as good as it is, is old, has too much on the DVD with more than 3 hours, and will never look all that good. Why don't you do a regular new movie DVD so DVD-RB and CCE can strut their stuff, and you can feel better about the money you spent. Or you might consider dumping the extras and keeping just the episodes.

You think the R2 Fawlty Towers is giving you fits? You should see what the R1 DVDs look like. Godawful.

[tgbyhn10]

Thanks for your reply manono. You are probably right about it never looking good, makes me think I should have bought 30 dual layer discs with the money for huge backups like these instead of wasting half my life tyring to achieve the impossible. BTW: there were no extras on the Fawlty Towers discs, just the episodes.


I have just checked the backup I did with DVDShrink using 'adaptive error compensation' and that is clearly better than what CCE and HC have been able to muster.

I was going to continue using DVDShrink for anything that needs less than a 20% reduction in size as I usually can't see any difference between these discs and the original.

This would leave me using DVD_RB for everything that needed a greater than 20% reduction in size, however, it seems that DVDShrink is best for lowest bitrate projects too.

So what's the sweet spot for DVD-RB and CCE Basic? When should I use them?

If anyone can still help me out with any aspect of my initial post I would be very grateful. I really want CCE and HC to outperfom DVDShrink at this level of compression.

Thanks

Pete

[dirio49]

1. Ran the DVD through DVDShrink to shrink the menus to the minimum size permitted (no compression performed on anything else)

2. Ran the output from step one through menushrink to decrease menu size (convert menus to stills).


Next time i would say don't even run it through shrink if don't want to keep the movie animated.
just through menushrink. you won't gain any think if you go first dvdshrink and menushrink.
menushrink alone should be fine if you want the menus to be stills.
this my opiniom

[tgbyhn10]

Next time i would say don't even run it through shrink if don't want to keep the movie animated.
just through menushrink. you won't gain any think if you go first dvdshrink and menushrink.
menushrink alone should be fine if you want the menus to be stills.
this my opiniom

Hi dirio49

That's fair comment. I knew it was going to be a tough job so I was trying to squeeze out every last bit of space I could find.

Pete

[cobo]

The best thing to do might be to blank all the segments for second half the episodes with DVD-RB and burn that to one disc, then run it through DVD-RB again blanking all the segments for the first half of the episodes and burn that to another. Using single layer discs and a six disc DVD case for the set would still cost less than using dual-layer discs.

-----------------------------------------------------

For DVD's that run about 3 hours or more, when I want to get everything on a single disc I use QuEnc59ß4 with "Extreme & Slow" settings checked . The results seem slightly better than with just "Trellis Quant" and "High Quality" ticked, which in turn seems to me to produce slightly better results than using 5 passes of CCE. Also, I always use "Scene Detection" without any problem. Also always use 2pass VBR.

With "Extreme & Slow" settings checked my encoding time is usually about 12 hours per hour of video using a P4 2.66B (ranges from 8 to 13 hours per hour of video). With just "Trellis Quant" and "High Quality" ticked it takes half the time.

There's no option in RB for selecting "Extreme & Slow" settings - it needs to be ticked manually in QuEnc's "Advanced Options..." settings before RB starts the encoding phase. All other QuEnc options can be set through RB.

I've tried v.60 but the results seemed no better to me than using CCE, so I've stuck with v.59. I don't know of anywhere to download v.59 now - if you want to try it I could email it to you.

EDIT:
I should have mentioned that sizing tends to vary morewidely for QuEnc than for CCE depending on the length and to some extent the content of DVD's. Achieving the right target size can be a matter of trial and error.

QuEnc tends to undersize with RB default Target Sectors (2236400 - see Hidden Settings in RB Help file).In the 'DVD ReBuilder: Settings' sticky 2260000 has been recomended as a reliable general Target Sector setting for filling out a disk out as close as possible to the end without going oversize.

Sizing is supposed to be more accurate in the latest version of RB. The first re-encoded I did using QuEnc and RB 1.08.1 was a 3:40hr DVD.
Using TargetSectors=2265500 resulted in a 4.29GB disc.
Using TargetSectors=2270500 resulted in a 4.30GB disc.
Using TargetSectors=2301659 resulted in a 4.42GB disc.
Using TargetSectors=2287000 resulted in a 4.36GB disc.
With CCE TargetSectors=2236500 produced a 4.32GB disc.

The next one I tried was a different 3:30hr DVD.
CCE TargetSectors=2257000 produced a 4.37GB(4481MB) disc.
QuEncTargetSectors=2287000 resulted in a 4.40GB disc which I was able to get down to 4.37GB by running the resulting DVDFILES through RB again and only re-encoding the few segments that could take a second reduction. It took half a dozen trials to find the right Target Sectors to get exactly 4.37GB on the second run-through (in this case 2293250), but the encoding time was only around 55 minutes for each trial.

The best way I know to get in the right ballpark for QuEnc Target Sectors is to do a trial run with CCE that results in the correct size (since that takes much less time), then try the prepare phase with QuEnc and various Target Sectors until you find one that gives a Reduction Level that's about half a percent more than you got with CCE.

[tgbyhn10]

The best thing to do might be to blank all the segments for second half the episodes with DVD-RB and burn that to one disc, then run it through DVD-RB again blanking all the segments for the first half of the episodes and burn that to another. Using single layer discs and a six disc DVD case for the set would still cost less than using dual-layer discs.

-----------------------------------------------------

For DVD's that run about 3 hours or more, when I want to get everything on a single disc I use QuEnc59ß4 with "Extreme & Slow" settings checked . The results seem slightly better than with just "Trellis Quant" and "High Quality" ticked, which in turn seems to me to produce slightly better results than using 5 passes of CCE. Also, I always use "Scene Detection" without any problem. Also always use 2pass VBR.

With "Extreme & Slow" settings checked my encoding time is usually about 12 hours per hour of video using a P4 2.66B (ranges from 8 to 13 hours per hour of video). With just "Trellis Quant" and "High Quality" ticked it takes half the time.

There's no option in RB for selecting "Extreme & Slow" settings - it needs to be ticked manually in QuEnc's "Advanced Options..." settings before RB starts the encoding phase. All other QuEnc options can be set through RB.

I've tried v.60 but the results seemed no better to me than using CCE, so I've stuck with v.59. I don't know of anywhere to download v.59 now - if you want to try it I could email it to you.

Thanks for the advice Cobo. I'll PM you.

Pete

[Jeffster]

With CCE; I have also tried lowering the Vbr_bias to 15 which didn’t help much and also changing the selected quantization matrices but I don’t really understand how these work. Should I leave them all at 'encoder default' or should I set 'low bitrate matrix' to 'low bitrate', 'very low bitrate matrix' to 'very low bitrate' etc.

Unfortunately CCE Basic doesn't support adjusting the VBR Bias or custom quant matrices (they're features of CCE SP), so changing those settings will have zero effect on your encode.


I have been using undot().deen() to try to improve the encodes, is this helping? Should I adopt a more aggressive undot filter?

If your source is interlaced(?) using spatial denoiser's can have a negative effect unless you filter each field separately (UnDot is a spatial filter and Deen is a temporal/spatial filter). Temporal filters can be used to good effect though on interlaced material (e.g. TemporalSoften).


Are there post processing filters I can apply to remove halo effects and mosquito noise? Solids look great with CCE, I can’t see any blocking at all; should I reduce Qual_Prec?

Some people will no doubt disagree with me but I've found that a Quality Precision setting of 10 in DVDRB (translates to 16 in CCE) gives good results in many cases (of course it's still dependent on the source), so you could try using a lower value than DVDRB's default value I feel.

The disc I am trying to encode is a region 2 version of ‘Fawlty Towers Series 2’ (disc 1).

If I have time I might grab it and have a look. :)

[Trahald]

Thanks for your reply manono. You are probably right about it never looking good, makes me think I should have bought 30 dual layer discs with the money for huge backups like these instead of wasting half my life tyring to achieve the impossible. BTW: there were no extras on the Fawlty Towers discs, just the episodes.


I have just checked the backup I did with DVDShrink using 'adaptive error compensation' and that is clearly better than what CCE and HC have been able to muster.

I was going to continue using DVDShrink for anything that needs less than a 20% reduction in size as I usually can't see any difference between these discs and the original.

This would leave me using DVD_RB for everything that needed a greater than 20% reduction in size, however, it seems that DVDShrink is best for lowest bitrate projects too.

So what's the sweet spot for DVD-RB and CCE Basic? When should I use them?

If anyone can still help me out with any aspect of my initial post I would be very grateful. I really want CCE and HC to outperfom DVDShrink at this level of compression.

Thanks

Pete
There are people who say dvdshrink/transcoders are better than methods that use encoders. its all in the eye of the beholder. ive tried both and have been happier with the output of methods that use encoders (like dvdrb) than transcoders like shrink. everyonce in a while i still try shrink since it has a cool interface and is pretty fast. but still happier with cce. but its just my opinion. and thats what counts for me since ultimately im stuck watching it. it may be possible that in your eyes dvdshrink outperforms cce/hc at any percentage. if so then you may want to stick with that.. or use dual layer on very low % sources or even split to 2xdvd5

[manono]

Hi-

Oh, you've got all 6 episodes on one disc and the extras on another? Impossible job then, I think, as it just doesn't compress well. The R1s are a 3 DVD set of the entire 12 episodes with 4 episodes and some extras on each disc. When I did them, I decided that I had to strip out the extras, as 4 episodes/disc was about all that would fit with decent quality. So, I agree with some of the earlier comments that the best thing for you would be to split it among 2 DVDs. Either that or buy a dual layer disc or 2.

Mine are also heavily edge enhanced. I used Blind Dehalo (I think) to get rid of it. Softened the picture more than I wanted, though.

[Boulder]

With Fawlty Towers, you'll also want to do some rainbow removal.

I also agree that you can't get good quality out of the backup process. Truly interlaced material, haloing etc. plus lots of movement means that the bitrate demand will be quite high.

One thing to try is to resize to 704x576 and use 2 overscan blocks. It'll improve compressibility quite nicely. FitCD can help you with a proper script except that you should use a smart bobber before doing any other processing. Use the search to find help on dealing with interlaced material.

[tgbyhn10]

Wow!!! First off I'd just like to say thanks to everyone for the help. I wasn't expecting so may people to jump to my aid. Many thanks to all of you.

It's going to take me some time to investigate all the advice so if anyone can point me in the best direction that would be great - There's so much to learn, I don't want to get bogged down researching the wrong way of doing things.

Unfortunately CCE Basic doesn't support adjusting the VBR Bias or custom quant matrices (they're features of CCE SP), so changing those settings will have zero effect on your encode.

Bugger. Wasted time, did I. I'm sure DVD RB Pro let me change these even though I had the CCE Basic encoder selected, bit annoyed about that.



If your source is interlaced(?) using spatial denoiser's can have a negative effect unless you filter each field separately (UnDot is a spatial filter and Deen is a temporal/spatial filter). Temporal filters can be used to good effect though on interlaced material (e.g. TemporalSoften).

Sounds like good info, I wish I understood more of it. What is the difference between a temporal filter and a spatial filter?




unless you filter each field separately


You lost me. Does a 'field' relate to the alternate lines when interlaced?

I am about to read up on interlacing here http://nickyguides.digital-digest.com/interlace.htm

Is there anything else I should read on the subject?

Is there a tool I can use to determine if the source is interlaced?




Deen is a temporal/spatial filter

Is this automatic or do I need to pass parameters?




TemporalSoften


I will investigate this filter. Are you suggesting this as an alternative to undot in this instance?



If the source is interlaced what should I do differently - deinterlace with decomb via option in DVD-RB? Will this always improve things? Should I not do this sometimes? Should DVD-RB do this automatically? Should I definatly do this with Fawlty Towers?

If I have time I might grab it and have a look.

I do hope u get time.




it may be possible that in your eyes dvdshrink outperforms cce/hc at any percentage. if so then you may want to stick with that.

I’m sure I will be using all these tools in the future, the difficult part for me is finding my sweet spot for each of them.



When I did them, I decided that I had to strip out the extras, as 4 episodes/disc was about all that would fit with decent quality. So, I agree with some of the earlier comments that the best thing for you would be to split it among 2 DVDs. Either that or buy a dual layer disc or 2.


I wish I could settle for a 2 disc split, unfortunately some part of my psyche wont let me do that, I wish it would.



Mine are also heavily edge enhanced. I used Blind Dehalo (I think) to get rid of it. Softened the picture more than I wanted, though.

I’ll check it out. Thanks for the tip.



One thing to try is to resize to 704x576 and use 2 overscan blocks. It'll improve compressibility quite nicely. FitCD can help you with a proper script except that you should use a smart bobber before doing any other processing. Use the search to find help on dealing with interlaced material.

Thanks, I'll search for FitCD and Smart Bobber.

Would the 704x576 resize lose any of the picture?


Sorry to post so many questions, I am trying to research all the advice you guys have given me but a quick yes or no from someone who knows could save me hours of hit and miss trawling.

Once again, thanks for all the help.


Pete.

[Boulder]

Thanks, I'll search for FitCD and Smart Bobber.

What I mean about smart bobber is using for example AssumeTFF().TDeint(mode=1) (you need tdeint.dll) to create a progressive stream. Then you can place all the filters and as the last item, use AssumeTFF().SeparateFields().SelectEvery(4,0,3).Weave() in the script to reinterlace the stream.

Would the 704x576 resize lose any of the picture?

704x576 is enough to display all the details that are there if that's what you mean. Using two overscan blocks means that you resize to 672x544 and add 16-pixel borders on each side. If you have a regular TV, the borders shouldn't be visible when watching the video. FitCD can give you the correct Avisynth stuff you can add to your script.

I use this approach if it looks like the final result won't look good enough.

[Jeffster]

@tgbyhn10
Sorry for the delay replying, I didn't know the best way to answer your questions, someone more knowledgable than myself would be more suited but I'll try anyway...

I can't give you a technical explanation of the difference between spatial and temporal filters, only that if your source is progressive you have the freedom to use either but interlaced source material requires special attention and certain types of filters can mess things up. It was a point I thought important you should be aware of because I've often found TV shows to be interlaced, especially BBC ones (and since no-one has suggested otherwise I guess it's true of Fawlty Towers).

I'm not aware of any tool to determine if the source is truly progressive/interlaced other than your eyes, however it's not as simple as it sounds.

Well, yes I was suggesting you may want to use something like Temporalsoften as an alternative to UnDot in this instance, however, I get the impression this DVD would benefit from more aggressive treatment (perhaps Boulder's suggestions). I'm not very familiar with Deen but I don't think you can disable spatial filtering.

No I wouldn't recommend deinterlacing when your output is for DVD. I believe JDobbs added the option on request but does not advise it either, if I'm not mistaken.


Another suggestion which comes to mind that may help quality a little bit, setting the GOP size to 15 instead of the DVD-RB default of 12.



Also, is there any way I can just encode 10 mins of this dvd and preview that instead of having to wait 4-5 hours?
I didn't notice this Q in your first post. If you want to interrupt the encode after one segment, or several segments, simply press the "Stop/Pause" button in DVD-RB. You can then view the encoded M2V files (video only) which will be located in your Working Folder.

[Boulder]

Fawlty is a hybrid. The indoor shots are shot on video, the outdoor stuff is shot on film. Treating as interlaced is the way to go.

[tgbyhn10]

@tgbyhn10
Sorry for the delay replying, I didn't know the best way to answer your questions, someone more knowledgable than myself would be more suited but I'll try anyway...

I can't give you a technical explanation of the difference between spatial and temporal filters, only that if your source is progressive you have the freedom to use either but interlaced source material requires special attention and certain types of filters can mess things up. It was a point I thought important you should be aware of because I've often found TV shows to be interlaced, especially BBC ones (and since no-one has suggested otherwise I guess it's true of Fawlty Towers).

I'm not aware of any tool to determine if the source is truly progressive/interlaced other than your eyes, however it's not as simple as it sounds.

Well, yes I was suggesting you may want to use something like Temporalsoften as an alternative to UnDot in this instance, however, I get the impression this DVD would benefit from more aggressive treatment (perhaps Boulder's suggestions). I'm not very familiar with Deen but I don't think you can disable spatial filtering.

No I wouldn't recommend deinterlacing when your output is for DVD. I believe JDobbs added the option on request but does not advise it either, if I'm not mistaken.


Another suggestion which comes to mind that may help quality a little bit, setting the GOP size to 15 instead of the DVD-RB default of 12.



I didn't notice this Q in your first post. If you want to interrupt the encode after one segment, or several segments, simply press the Abort button in DVD-RB. You can then view the encoded M2V files (video only) which will be located in your Working Folder.


Thanks for all your help Jeffster it is very much appreciated.

Looks like I'll have to lock myself away for a few months while I test all these filters / setting changes.

Pete

[tgbyhn10]

What I mean about smart bobber is using for example AssumeTFF().TDeint(mode=1) (you need tdeint.dll) to create a progressive stream. Then you can place all the filters and as the last item, use AssumeTFF().SeparateFields().SelectEvery(4,0,3).Weave() in the script to reinterlace the stream.

Sounds good, deinterlace, filter, then re-interlace. I will look into this. Can I still put this in the script entry area in dvdrebuilder or would it be best to do this directly in AVISynth?

704x576 is enough to display all the details that are there if that's what you mean. Using two overscan blocks means that you resize to 672x544 and add 16-pixel borders on each side. If you have a regular TV, the borders shouldn't be visible when watching the video. FitCD can give you the correct Avisynth stuff you can add to your script.

I use this approach if it looks like the final result won't look good enough.

Thanks Boulder, lots of good info.

I loaded up FitCD and started to think that maybe all this is getting out of hand - there just seems so much to take in and evrything I read up on leads me to several other things I need to read up on. Anyway, here goes.

OK so is this utter garbage or does it resemble something I could try?

# -= AviSynth v2.5.6.0 script by FitCD v1.2.8 =-
AssumeTFF().TDeint(mode=1)
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\DGDecode.dll")
Mpeg2Source("C:\DVDs\DVDR_work\D2VAVS\V02.D2V")
BicubicResize(688,544,0,0.6,0,3,720,570)
AddBorders(16,16,16,16)
#Trim(0,264740).FadeOut(150)
undot().deen()
AssumeTFF().SeparateFields().SelectEvery(4,0,3).Weave()


I assume the # denotes is a comment line.

Should I run this from within DVD-RB or Avisynth?

Would it be best to run a prepare in DVD-RB and then pick the .D2V file as above for the encode?

Should I undot/apply other filters before the resize?

Thanks


Pete

[Boulder]

Try this (add it to the filter editor):

AssumeTFF()
TDeint(mode=1)
undot().deen()
LanczosResize(688,544,0,3,720,570)
ConverttoYUY2()
AssumeTFF().SeparateFields().SelectEvery(4,0,3).Weave()
AddBorders(16,16,16,16)

Some notes:

1) For better compression, choose BilinearResize instead of LanczosResize. The image will be softer though.

2) Choose 704x576 as the destination resolution in FitCD. If you do that and use CCE, you must edit the rebuilder.ecl file after running the prepare phase in DVD-RB. Replace all width=720 lines with width=704. With HC, you do not need to do this.

3) Use HC, it usually produces better quality than CCE with interlaced material. Choose the KVCD notch or QLB as the quant matrix. If you use HC, remove the ConverttoYUY2 line from the filter editor.

4) For rainbowing (which Fawlty has a lot), I would try FFT3DFilter/FFT3DGPU. The latter is faster if you have a decent graphics card. Try FFT3DGPU(sigma=3,bt=3,plane=1,degrid=1,mode=1). For FFT3DFilter, you need to specify a different plane parameter, see the docs which setting treats all chroma planes.

[tgbyhn10]

Try this (add it to the filter editor):

AssumeTFF()
TDeint(mode=1)
undot().deen()
LanczosResize(688,544,0,3,720,570)
ConverttoYUY2()
AssumeTFF().SeparateFields().SelectEvery(4,0,3).Weave()
AddBorders(16,16,16,16)

Some notes:

1) For better compression, choose BilinearResize instead of LanczosResize. The image will be softer though.

2) Choose 704x576 as the destination resolution in FitCD. If you do that and use CCE, you must edit the rebuilder.ecl file after running the prepare phase in DVD-RB. Replace all width=720 lines with width=704. With HC, you do not need to do this.

3) Use HC, it usually produces better quality than CCE with interlaced material. Choose the KVCD notch or QLB as the quant matrix. If you use HC, remove the ConverttoYUY2 line from the filter editor.

4) For rainbowing (which Fawlty has a lot), I would try FFT3DFilter/FFT3DGPU. The latter is faster if you have a decent graphics card. Try FFT3DGPU(sigma=3,bt=3,plane=1,degrid=1,mode=1). For FFT3DFilter, you need to specify a different plane parameter, see the docs which setting treats all chroma planes.

Thanks Boulder you are a star. I'll make the changes you suggest.

I don't have a KVCD Quat Matrix but I do have the default DVD-RB 'QuEnc Lower Bitrate (QLB)' matrix, should I set the 'main feature matrix' to this and leave all the others set to 'same as main feature?' or would I need to download the KVCD / QLB matrix from somewhere? I have the ones that come with DVD-RB Matrix editor.

Should I choose QLB/KVCD regardless of endocder when used with this type of material (low bitrate/interlaced)

I'll give HC a try again although I was disapointed by the results it churned out for 'Rising damp.' CCE produced clearly better results - I assume this would have been interlaced too.

How do you know if something is TFF or BFF? Does it matter that much?

Still got to try QuEnc as per Cobo's suggestion but that sounds like a long one so I'll try it at the weekend as I'm away for a couple of days.

Wish I could leap forward 20 years and come back with a faster computer.

Does anyone know if it is OK to reduce an encoders process priority via Windows task manager?

Will the DVDs produced by all this work on stand alone DVD players?

Well I can't think of anymore questions.

Thanks again.

Pete.

[Boulder]

I don't have a KVCD Quat Matrix but I do have the default DVD-RB 'QuEnc Lower Bitrate (QLB)' matrix, should I set the 'main feature matrix' to this and leave all the others set to 'same as main feature?' or would I need to download the KVCD / QLB matrix from somewhere? I have the ones that come with DVD-RB Matrix editor.

Yes, leave it as the main feature matrix. The notch matrix can be found in robot1's RBOpt package for example.

Should I choose QLB/KVCD regardless of endocder when used with this type of material (low bitrate/interlaced)

Yes. I would try QLB first as it'll keep more details.

How do you know if something is TFF or BFF? Does it matter that much?

There's a good tutorial included in the Decomb package. I'd say 99% (or more) of DVDs are TFF. Knowing the correct field order is essential! There'll be Very Bad Things happening if you set it incorrectly at any phase.

Wish I could leap forward 20 years and come back with a faster computer.

FYI, the script you are about to use is slow..if you want to make it a fair bit faster, you can use LeakKernelBob(order=1,sharp=true,threshold=4) instead of TDeint. You'll need the LeakKernelDeint plugin. You might not notice any difference in terms of video quality. The good thing about DVD-RB is that you can stop the encoding and continue it anytime you want to.

Does anyone know if it is OK to reduce an encoders process priority via Windows task manager?

It is okay.

Will the DVDs produced by all this work on stand alone DVD players?

Yes, 704x576 is a valid DVD resolution.

[feedback]

Since Boulder suggested the KVCD Notch Matrix, I have made it readily available with the link below. Plus, there are many other matrices that are within Matrix Zip Folder 1 and 2.

http://forum.doom9.org/showpost.php?p=687112&postcount=3

You can import them with the Rebuilder Matrix Editor by Rockas so they will be available for use with Rebuilder.

Have fun with your filters and matrices:)

[tgbyhn10]

Thanks to everyone that tried to help me with this.

I tried everything suggested by everyone and to be honest I wasn't happy with any of the results. I guess there are some DVDs that aren't meant to be DVD-5.

Did a half-D1 on everything in the end and seems watchable but not happy with it.

Thanks again.

Pete

[Harrysmiith]

I tend to use DVD Rebuilder and CCE Basic and nothing else - having gotten very confused with undot deen and the rest. If there a six episodes per dvd and 2 episodes would fit on a dvd would there be any need for filters etc if the six episodes were simplt buned to 3 dvds without any compression ?

[JohnGalt]

no, definitely don't filter if you're not compressing, unless you're using a corrective filter to clean up the source or something (adjust contrast, remove noise, etc.). anyway, the short answer is no, don't filter.