Motion Blur

[RB]

Hmm, very interesting. This has always worked for me and a lot of other people (encode BFF with upper field first disabled, then use Restream to clear TFF flag). Anyway, I really don't want to ride a dead horse, it's solved so let's move on :)

[jdobbs]

Originally posted by RB
Hmm, very interesting. This has always worked for me and a lot of other people (encode BFF with upper field first disabled, then use Restream to clear TFF flag). Anyway, I really don't want to ride a dead horse, it's solved so let's move on :) In retrospect, what I wish I had tried first for this fix would be to always set TFF on in the .D2V file (all '2's) rather than do the conversion to TFF. I'm with you. I'd much rather have a pristine duplication as closely matched to the original as possible. In fact I may post a test version to try that. What do you think?


ADDED: I guess, though, in that situation I'd have the lines transposed in the frame sent to CCE and the compression would suffer...

[jdobbs]

BTW. I don't mind beating a dead horse if it is at least an interesting subject. I just don't want to use the word "stutter" in a sentence anymore.

[djackson]

Have just tested 0.38 on Frasier (Season 1, Disc 1&2, PAL R4). I'm now getting a combing effect with motion.

If there is any info I can supply, or any test I can run, I'm more than willing to try.

I appreciate the efforts so far to stomp on this one.

[jdobbs]

Originally posted by djackson
Have just tested 0.38 on Frasier (Season 1, Disc 1&2, PAL R4). I'm now getting a combing effect with motion.

If there is any info I can supply, or any test I can run, I'm more than willing to try.

I appreciate the efforts so far to stomp on this one. That's normal when you play interlaced back on a PC. Have you tried burning it and watching it on a television? If your end-plan is to watch it on a PC you can use the DECOMB function under Options/AVS Options/Advanced (Expert) Options.

[djackson]

Playback was done on a standalone. Playback on my PC has always been OK. I think I'm just jinxed.:confused:

[jdobbs]

You know. I can to make another version that does what I mentioned above just as a test -- are you up for it?

[djackson]

What did P.T Barnum say? "There's a sucker born every minute!" So I must be him... I'm game:D

[RB]

Originally posted by jdobbs
I thought so too. That's exactly what I was doing originally and it was causing the reported problems. I think CCE has issues with BFF material that doesn't have the top_first/offset_line option set. I can't imagine why, though -- especially since it is coming through to CCE as a FRAME anyway. My guess would have to be that AVISYNTH is passing the BFF info through somehow???? Don't know for sure.
Hmm, maybe those reporting problems actually had DVDs with field picture streams, meaning the D2V would be all 0-Flags (BFF) and you misinterpreted this to actually mean BFF so cleared TFF flag during muxing? We discussed the field pictures issue a few posts up on this page, confirmed by FMalibu.

About AVISynth - not sure Mpeg2Dec3Dg passes any BFF/TFF info to AVISynth. OTOH, it should, otherwise SeparateFields/Weave could never be used reliably.

[jdobbs]

Originally posted by RB
Hmm, maybe those reporting problems actually had DVDs with field picture streams, meaning the D2V would be all 0-Flags (BFF) and you misinterpreted this to actually mean BFF so cleared TFF flag during muxing? We discussed the field pictures issue a few posts up on this page, confirmed by FMalibu.

About AVISynth - not sure Mpeg2Dec3Dg passes any BFF/TFF info to AVISynth. OTOH, it should, otherwise SeparateFields/Weave could never be used reliably. I'm pretty sure that isn't the case, because I've found a bug (I'm now working) that makes at least some of those streams fail completely. It is the cause of one of the "Runtime Error '9'" errors in PREPARE. Considering how few complaints I've gotten I think it's safe to say that those are pretty rare.

I think I had a basic misunderstanding as to how those streams work... I got an example in which the TOP_FIELD was tagged as an I frame, and the BOTTOM_FIELD for the same frame was a P frame. I didn't know that was how it worked... I need to go back to the spec and do some homework on those.

[jdobbs]

@djackson and anyone else who wants to try this on a BFF source

Here is a test version of 0.38. Here's what it does:

1. It flags all frames as TFF in the .D2V files so they are served as such to AVISYNTH

2. The top_field/offset_line is set to 0 to it is processed as TFF

3. The flags are reset to their original state (TFF=0) in the REBUILD.

If this works without making an ugly image it would be the best way to handle these.

Thx. jdobbs

[djackson]

Just starting the process off.

This is from the info panel of Bitrate Viewer for the source of Frasier S1, D1, PAL R4 (VTS_01_2.VOB)
Num. of picture read: 26
Stream type: MPEG-2 MP@ML VBR
Resolution: 720*576
Aspect ratio: 4:3 Generic
Framerate: 25.00
Nom. bitrate: 9800000 Bit/Sec
VBV buffer size: 112
Constrained param. flag: No
Chroma format: 4:2:0
DCT precision: 10
Pic. structure: Frame
Field topfirst: Yes
DCT type: Field
Quantscale: Nonlinear
Scan type: Alternate
Frame type: Interlaced
Scene change detection: NOT FOUND
Variable GOP pattern: NOT FOUND
Notes:

[jdobbs]

????? :confused:

That says it's top field first.

[zeus163]

I didn't try the new test version, but did try 0.38 last night on Great Teacher Onizuka. When I woke up the encode was finished, so I rebuilt it, and then burnt the folders (I was even able to include the jacket file!).

The resulting DVD played just like the original, in fact, I bet I could fool people with how good this looks.

Thanks jdobbs what a super package.

[jdobbs]

Originally posted by zeus163
I didn't try the new test version, but did try 0.38 last night on Great Teacher Onizuka. When I woke up the encode was finished, so I rebuilt it, and then burnt the folders (I was even able to include the jacket file!).

The resulting DVD played just like the original, in fact, I bet I could fool people with how good this looks.

Thanks jdobbs what a super package. Ahhh... shucks..:)

[Rombaldi]

@jdobbs

Just wandered in this morning, so the 'test' 38 hasn't been tried, but using the regular 38 on one of my 'problem children' (Land of the Giants, vol.1 from Columbia House) I still have the 'field flicker' problem.

will try the 'test' 38 ASAP, but I still have all the RB files from this attempt, would you care to see the contents of any and what else can I provide. (I haven't tried 38 on one of my homemades yes, wanted to start with this one..)

[jdobbs]

Originally posted by Rombaldi
@jdobbs

Just wandered in this morning, so the 'test' 38 hasn't been tried, but using the regular 38 on one of my 'problem children' (Land of the Giants, vol.1 from Columbia House) I still have the 'field flicker' problem.

will try the 'test' 38 ASAP, but I still have all the RB files from this attempt, would you care to see the contents of any and what else can I provide. (I haven't tried 38 on one of my homemades yes, wanted to start with this one..) You're telling me that the changes in 0.38 had no affect whatsoever on your output?

[djackson]

@jdobbs

Have rerun Fraiser S1-D1 thru the 0.38 BFF2 version just posted above. Perfection! For a while I was beginning to think this would become the never ending thread.

Thanks! Where can I donate to the cause?

[Rombaldi]

Originally posted by jdobbs
You're telling me that the changes in 0.38 had no affect whatsoever on your output?

On the Land of the Giants, none that I can see. I haven't tested my Panasonic recorded discs (yet), those are next. LOTG is processing now (just a 2 pass on CCE) with the BFF2 version, so I should have a report on that in the next couple of hours. I did save the RB created files from the first run with 38, so I will have something for comparision...

whatever the results, I'll do a Panasonic disc next..

[jdobbs]

Originally posted by djackson
@jdobbs

Have rerun Fraiser S1-D1 thru the 0.38 BFF2 version just posted above. Perfection! For a while I was beginning to think this would become the never ending thread.

Thanks! Where can I donate to the cause? Just push the button located in the "About" box.

[jdobbs]

Originally posted by djackson
@jdobbs

Have rerun Fraiser S1-D1 thru the 0.38 BFF2 version just posted above. Perfection! For a while I was beginning to think this would become the never ending thread.

Thanks! Where can I donate to the cause? I just want to make sure I have it straight. You had problems with the output of "Frasier" on the published version of 0.38 -- but the special test version BFF-TEST2 that I loaded this morning?

I want to be sure, because if everything worked okay I'm going to use that new code in V0.39...

@RB -- have you tried this test version yet on some BFF material, I want to be especially sure that it will run alright before I post it? Oh, by the way -- I fixed the Runtime Error 9 you experienced. As I mentioned I had some issues in my field-based processing. Thanks for the file.

[RB]

Problem is, I don't have any DVD with BFF interlaced material... I'll try to make one up from some home made video.

[jdobbs]

@RB

That's ok. Don't go to the trouble -- I thought you already had something.

I'm really getting confused on this one.

1. The first fix I implemented (using the top_field=1 etc.) seemed to fix everything.

2. Then djackson said he was still getting blurring from it... so I made up the BFF2 version as a test to see if it could fix the BFF problem he was having.

3. But then I noticed that his post from Bitrate viewer showed Frasier to be TFF. So the new version I loaded shouldn't have had an effect (it only should affect BFF)

4. Now djackson reports that the problem was fixed with BFF2...

:confused:

I'm trying to decide whether to post v0.39 with the new code or the old code. I was just looking for someone to confirm that the new code doesn't screw up BFF streams because I don't have any.

[Rombaldi]

Originally posted by jdobbs
I was just looking for someone to confirm that the new code doesn't screw up BFF streams because I don't have any.

I'll offer again, PM me with where to send it and I'll send you a Panasonic recording that (appears) to be BFF (still in the test queue, waiting for LOTG to finish) so you have have it in your hands and dissect to pieces...

[jdobbs]

@Rombaldi

Thanks. But I just used my DV camera and QuEnc to make myself a BFF titleset (I examined it with VOBEDIT and it is definitely BFF). The output looks the same either way to me.

[Rombaldi]

@jdobbs

NP, I'm still about 45 minutes away from seeing how LOTG comes out. If that works, then I'll feed it a Panasonic recording and see what
happens....

don't wait up on .39 on me tho... I'll hammer it as I can and find the cockroaches as they turn up.

[jdobbs]

Thanks. I put the new code in 0.39 -- but I have the old ready in case it needs to be reinserted.

[djackson]

Originally posted by jdobbs
I just want to make sure I have it straight. You had problems with the output of "Frasier" on the published version of 0.38 -- but the special test version BFF-TEST2 that I loaded this morning?

I want to be sure, because if everything worked okay I'm going to use that new code in V0.39...

Yes... The published 0.38 had the combing/blur, but the special BFF2 version was OK on the Frasier discs. This is even though Bitrate viewer says the source stream is TFF.

The thing is I didn't have this prob with "Pirates of the Carribean" with an earlier version of DVD-RB. I'll run this disc tonight (at work now) with the special version to see what happens. Maybe thre may be a need have both sets of code in a later release, and have a setting in advanced options.

[Rombaldi]

Originally posted by jdobbs
Thanks. I put the new code in 0.39 -- but I have the old ready in case it needs to be reinserted.

@jdobbs

well, no joy in Mudville, the LOTG still has the field flicker. I have maintained all the generated files from 38 and 39BFF2, so what
can I post for you to persue???

[Rombaldi]

@jdobbs

well, I'm losing what little mind I have....

I WIPED everything off my system,
cleaned the registry,
reinstalled DVDRB, the DLL's, everything
installed QENC (just to be different) and re-ran the LOTG
thru using QENC as the encoder..

NO FIELD FLICKER!

However, now the audio is about 2 seconds out of sync and the picture stalled every few seconds...

ARGGGGGGGGGGGGGGGGGGGGGGGGGGHHHHHHHHHHHHHHHHHHHHHHHH!!!!!

I'm going to re-run this with CCE 2.50, check back in about 4 hours (note to self, get a faster CPU)

[redfive19]

@jdobbs

I am still a bit new to the field order stuff. Does this mean we wont have to deinterlace with decomb? Does Decomb make the material look crappier?

-redfive

[babaz]

it's all about keeping the interlaced...
by not deinterlacing it, we consider to mainly play back this content onto TV screens, and (as most do) i tend to prefer the more smoothness an interlaced source can allow...

of course, if you want to, you can "decomb" it to progressive
there should be no issue at all, then

my 2 cents

[redfive19]

right but before jdobbs added the decomb feature, i would get jaggy ghosting lines everytime i encoded TFF. I did the same DVD with DoItFast4U and DIF4U had a call to decomb in the AVS files. when I finished the DVD with the Big3, I did not have these lines. BTW, those lines were only visible on my tv.

-redfive

[jdobbs]

You still have the option to use decomb with DVD-RB -- it hasn't been removed. Different strokes for different folks.

[Rombaldi]

@jdobbs

I am OFFICIALLY losing my mind now...

as the previous report said, when using

RB 0.39 & QEnc 0.45

starting from an 'clean wipe' (uninstall AVISynth, RB, CCE, QEnc, ReJig) and a registry clean/compact.

reinstall AVISynth, DLL's, RB and QEnc

running LOTG thru produced NO FIELD FLICKER but the audio was out of sync from the get go and the disc FROZE every few seconds like it was trying to catch up...

CHANGING NOTHING ELSE except installing CCE 2.50 and ECLCCE 1.8a
and configuring same, switching to CCE Encode, 2 pass.

Burned the disc

With fear I press play.

Audio in proper sync - check.
No 'freezing' of the picture - check.
NO FLIPPING FIELD FLICKER, NONE - check!

in other words.. PERFECT.

Rule 1 of beta testing, clean the environment every so often.

I'll run thru a Panasonic recorded disc later tonight on 0.39, and
report back..

I can't explain it any other way than a 'contaminated environment'.

[Paced]

Originally posted by Rombaldi
@jdobbs

I am OFFICIALLY losing my mind now...

as the previous report said, when using

RB 0.39 & QEnc 0.45

starting from an 'clean wipe' (uninstall AVISynth, RB, CCE, QEnc, ReJig) and a registry clean/compact.

reinstall AVISynth, DLL's, RB and QEnc

running LOTG thru produced NO FIELD FLICKER but the audio was out of sync from the get go and the disc FROZE every few seconds like it was trying to catch up...

CHANGING NOTHING ELSE except installing CCE 2.50 and ECLCCE 1.8a
and configuring same, switching to CCE Encode, 2 pass.

Burned the disc

With fear I press play.

Audio in proper sync - check.
No 'freezing' of the picture - check.
NO FLIPPING FIELD FLICKER, NONE - check!

in other words.. PERFECT.

Rule 1 of beta testing, clean the environment every so often.

I'll run thru a Panasonic recorded disc later tonight on 0.39, and
report back..

I can't explain it any other way than a 'contaminated environment'.

Woah, first lab-one, and now you :) It's good to hear that these problems may not be DVD-RB's fault after all :cool: