Sighhh... Interlaced, Telecine

[gijski]

I find myself still struggling a bit with making the correct Interlaced & Telecine settings for my conversions with HB.
I've read quite some posts and wikipedia on this topic, but still not sure if I'm doing the right thing...

What I do now:
- load a VOB in Gspot and if it says it's interlaced (I/L) I tick in HB the Deinterlace/fast-setting.
- I still double check with DGIndex. Frame type = Interlace is what I see as a confirmation.
- I am confused by DGIndex's Frame structure = field / frame. I find myself ticking the HB Deinterlace/fast-setting regardless 'to be on the safe side'.

My video's look good this way (in combination with the AppleTV preset), but I'm still doubting if there is something I'm missing... (when do I use the 'telecine'-setting??) and better to ask now, then re-do my DVD collection I suppose, hehe.

[jbrjake]

I updated the documentation earlier today. Check it out.

[gijski]

I did read http://trac.handbrake.fr/wiki/DeinterlacingGuide (incl 'Other resources'), but still it confuses me...

I understand there's probably no easy answer, but on a practical level how do you guys (girls) deal with this when loading a DVD and converting it ? What steps do you take to decide whether you got to to (un)check Detelecine/Deinterlace/etc? (btw I'm using WinXP, so no picture settings/preview pane for me)

[jbrjake]

I meant the page on telecining: http://trac.handbrake.fr/wiki/Telecine

There is no easy solution to this. VFR is a partial solution ( http://trac.handbrake.fr/wiki/VariableFrameRate ) and 0.9.3 will be smarter still, to tell you when a title is probably interlaced, and will even have a "automatic" mode that deinterlaces and detelecines when needed. For now, the best way to go about things is to examining the source frames with your eye.

[gijski]

I now started following this rule:

- open a VOB-file in DGindex & press preview
- when DGIndex says Frame Type = Interlaced & Frame Structure = Field -> I choose in HB to Deinterlace
- in all cases I check the VFR chekbox (Detelecine is then checked by default in AppleTV preset)

Output seems quite ok so far, but would you recommend VFR & Detelecine always on?
(esp. since the wiki says "This feature only works with NTSC content")

[jbrjake]

Output seems quite ok so far, but would you recommend VFR & Detelecine always on?
(esp. since the wiki says "This feature only works with NTSC content")

In eddyg's testing, it does not work reliably on PAL DVDs. One of the first post-0.9.2 changes I'll make to the code is to ignore VFR if the fps is 25. Then it should be safe to use it all the time.

I would not always deinterlace, because it does seriously degrade picture quality. However, if you want something that "always" works you're going to have to accept that quality loss for now.

I'm not familiar with what DGIndex is doing these days; I'm uncertain if it's smart enough to know the difference between telecining and interlacing. It might be seeing the interlaced frame type and field structure on the telecined frames, in which case you'd be enabling deinterlacing in situations that might work fine with VFR alone.

[rhester]

I'm not familiar with what DGIndex is doing these days; I'm uncertain if it's smart enough to know the difference between telecining and interlacing. It might be seeing the interlaced frame type and field structure on the telecined frames, in which case you'd be enabling deinterlacing in situations that might work fine with VFR alone.

It detects telecine only and depends on post-processing to do comb detection and removal.

Rodney

[gijski]

I haven't noticed any problems when using VFR/Detelecine on a PAL DVD. (I've done about 8 of them now).
What kind of problems/artifacts would I be looking for?

And the forbidden question (guess I'm the first one to ask..) any e.t.a. on 0.9.3 (with the auto-detect feature in it)?
Will that be days/weeks/months? In case of weeks/months, I'll better proceed this way, but if you think of days... I'll wait

oh man, I feel so ungreatful already asking for 0.9.3 when 0.9.2 is not even out 24 h.

(Would love to donate you all a beer for the great work, but how?! HB doesn't accept donations )

[s55]

No idea on the ETA. Definatly won't be days anyway

[Maury Markowitz]

load a VOB in Gspot and if it says it's interlaced (I/L) I tick in HB the Deinterlace/fast-setting
[snip]
(when do I use the 'telecine'-setting??) and better to ask now, then re-do my DVD collection I suppose, hehe.

Interlaced source does not generally mean you want to turn on deinterlacing. Depending on the sorts of things you're encoding, it might be quite rare.

The general rule in my experience is to look at the frame rate display after opening the input file. Then:

1) if the rate is "movie like" (~24 fps) then leave deinterlace and detelecine off
2) if the rate is "tv like" (~25 or ~30 fps) and the content is "live action" (sports, news, SNL, talk shows, etc.) then use deinterlace
3) if the rate is "tv like" (~25 or ~30 fps) and the content is NOT "live action" then use detelecine

Basically the question you need to ask yourself is "was this content originally shot directly to a videotape?" If the answer is no, you most likely don't want to use deinterlace, you likely want to use detelecine instead. It doesn't work 100% of the time, but when it does work the results are simply fantastic.

The good news here is that if you used detelecine on material that doesn't need it there's no harm done (at least in my experience). Thus its generally better to err on that side. In comparison, deinterlacting will always have noticeable effects on the output, sometimes bad ones. So in the case of deinterlacing, it's good to err on the side of not using it.

Yes, there are some sources that are not live TV and do require deinterlace, but they tend to be rare (and weird). I've only come across one such example, and it was on disk one of a five disk set -- the other four disks worked perfect with detelecine.

Maury

[jasonJgrimme]

Hey guys,

Love HandBrake, have it running nearly 24x7 on my PC to encode all my videos to playback on my HTPC.

I've been fighting the Interlaced stuff (or atleast, stuff dgindex says in interlaced) for quite a while, and this post has been awesome. I mainly have TV series on DVD that I'm looking to encode. I've found using the same parms I use for movies, with the addition of enabling detelecine works pretty well, but I have had a minor issue as of late.

I'm enocoding the episodes of a sitcom, and for the first minute or so, everything appears very *pixelated*, like I'm watching on youtube or something. Then, all of a sudden, things clear up and look great. It has happened on the 3 episodes I have checked without fail.

Any thoughts?

[sdm]

In eddyg's testing, it does not work reliably on PAL DVDs. One of the first post-0.9.2 changes I'll make to the code is to ignore VFR if the fps is 25. Then it should be safe to use it all the time.

Speaking of 25fps/detelecine/vfr, I have something interesting to report. You may or may not be surprised.
I have a set of dvds that are region 1 (and cg animated). I took a quick look to see if they were hard or soft telecined. They appeared to be hard telecined because of the 3 progressive, 2 interlaced frame pattern. When I encoded two of them with HB + vfr + detlecine filters, the resulting average framerates were 25.09fps and 25.25fps reported by QT. QT reports the playing fps mostly at 25fps with some variation (24 - 30 fps)

I think the animation must have been created at 25fps and pulled down to 30fps, but mixed with some natural 30fps content.

The surprising thing is that the playback is really quite smooth (and I do tend to be hyper-aware of jerkiness (judder?)) and I couldn't find a combed frame anywhere. It seems like HB with vfr and delecine can do pullup to 25fps!?!

jbrjake, what do you make of this? Should it have worked, or is my analysis of the source likely wrong?

Any thoughts?
--sdm.

[van]

The pullup code doesn't know anything about the frame rate, only the pattern of fields. And it seems to be pretty robust to variations in that. My wife has a bunch of BBC "Mystery" DVDs that were hard-telecined from 25fps PAL progressive to 30fps NTSC (which results in a 4P:2I pattern rather than film's 3P:2I pattern). VFR initially overdropped low motion & dark scenes on these (lots of juddering) but when I set hard_breaks to -1 (the "drop less" setting) they converted perfectly - no combing & perfectly smooth 25fps output. I've converted about a dozen of these DVDs so far with no visible problems.

But I'm surprised you got smooth 25fps from 3:2 - dropping two fields in 10 should have given you 24fps.

[sdm]

But I'm surprised you got smooth 25fps from 3:2 - dropping two fields in 10 should have given you 24fps.

I may need to recheck the source - maybe the pattern wasn't as I described!

Still, pretty cool you guys!

--sdm.

[jbrjake]

The pullup code doesn't know anything about the frame rate, only the pattern of fields. And it seems to be pretty robust to variations in that.

I just don't want to get people's hopes up -- eddyg's had it fail on one PAL title, and the MPlayer documentation doesn't make any claims about PAL:

Third-generation pulldown reversal (inverse telecine) filter, capable of handling mixed hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive content.

Since it's stateless it's not going to fail outright like a dumb pattern-based deteleciner. However, I googled for info from the filter's author on using it with PAL, and the closest I can find is this statement about using it on NTSC Film->PAL transfers:

You might
find a few (very rare) PAL dvds with 3:2:2:2:2:2:2:2:2:2:2:2 pulldown
(12 frames that look good followed by 11 that look interlaced, and
repeating). I would expect this may be the case in "The Wall" since
speeding up the movie from 24 fps to 25 fps would damage the musical
content. Anyway -vf pullup can repair this sort of pulldown (but every
24th frame will be duplicated)

Oh, wait, hey, I found something else! Shiny! It does work on odd stuff, no matter what the MPlayer docs say!

In any case, pullup
is designed to handle ANY form of pulldown, whether it's 2:2 PAL, 2:3
NTSC, 2:2:2:2:2:2:2:2:2:2:2:3 PAL, or whatever. Of course for pure 2:2
content, no pulldown reversal is required unless the phase is
backwards, since the fields already line up to give you progressive
frames with no rearrangement.

[thundar]

Hello. I'm also having a problem with (what I THINK is) interlaced material. But I'm a little confused. The movie is "Wicker Park". I have used various xbox 360 encoding options (h264). My first encode didn't use any Deinterlacing. I then tried Deinterlacing using both Fast and Slowest. I also tried using JUST VFR. The Framerate is automatically changed from 29.970 to 23.976. I get jagged lines on EVERY encode. So what can the problem be? Would assuming the original Frameratemake any difference? Sorry if this question has already been hammered at a million times.

[thundar]

Well, I tried an encode (one chapter) assuming the original framerate. Still jagged. Hmmm.... I guess this movie is just facked.

[thundar]

Back one more time. I encoded the same chapter using Fairuse Wizard. Same Framerate using Deinterlacing. Now, I'm not trying to compare the two programs. This is for experimental sake only and to possibly help people out in the future. Well, the Fairuse encode looks WAY better. So should I assume that Deinterlacing in Handbrake doesn't work too well?

p.s. In the future, I'll try to keep the posts (in one particular thread) down to a minimum.

[jbrjake]

Thundar, you should know by now to provide activity logs if you want useful answers to questions. I can't even begin to imagine what's going on without getting basic necessities like the filter chain order.

VFR by itself should never produce jagged edges.

There is no chance FairUse Wizard has better deinterlacing -- if it was any better than the slow/slower deinterlacer in HB (yadif) it'd have to be motion compensated which == extremely slow.

[thundar]

Wow, you really do love those logs, don't you?

Well, I don't know what' going on with Handbrake, but there is absolutely no doubt that Fairuse produced much better quality. Here is a bunch of mumble-jumble that the activity window left behind. Can you make any sense of it?

Code: Select all

[21:07:01] hb_init: checking cpu count
[21:07:01] hb_init: starting libhb thread
[21:07:01] thread ed2388 started ("libhb")
HandBrake 0.9.2 (2008022300) - http://handbrake.m0k.org/
2 CPUs detected
Opening D:\VIDEO_TS...
[21:07:01] hb_scan: path=D:\VIDEO_TS, title_index=1
[21:07:01] scan: trying to open with libdvdread
[21:07:01] thread ed25e8 started ("scan")
[21:07:02] scan: DVD has 26 title(s)
[21:07:02] scan: scanning title 1
[21:07:02] scan: opening IFO for VTS 1
[21:07:02] pgc_id: 1, pgn: 1: pgc: 0xed5690
[21:07:02] scan: vts=1, ttn=1, cells=0->24, blocks=0->2199280, 2199281 blocks
[21:07:02] scan: duration is 01:54:44 (6884233 ms)
[21:07:02] scan: checking audio 1
[21:07:02] scan: id=80bd, lang=English (AC3), 3cc=eng
[21:07:02] scan: checking audio 2
[21:07:02] scan: id=81bd, lang=Francais (AC3), 3cc=fra
[21:07:02] scan: checking audio 3
[21:07:02] scan: id=82bd, lang=English (AC3), 3cc=eng
[21:07:02] scan: checking subtitle 1
[21:07:02] scan: id=23bd, lang=English, 3cc=eng
[21:07:02] scan: checking subtitle 2
[21:07:02] scan: id=24bd, lang=Francais, 3cc=fra
[21:07:02] scan: checking subtitle 3
[21:07:02] scan: id=25bd, lang=Espanol, 3cc=spa
[21:07:02] scan: checking subtitle 4
[21:07:02] scan: subtitle channel is not active
[21:07:02] scan: checking subtitle 5
[21:07:02] scan: subtitle channel is not active
[21:07:02] scan: checking subtitle 6
[21:07:02] scan: subtitle channel is not active
[21:07:02] scan: title 1 has 24 chapters
[21:07:02] scan: chap 1 c=0->0, b=0->113951 (113952), 342381 ms
[21:07:02] scan: chap 2 c=1->1, b=113952->150392 (36441), 125111 ms
[21:07:02] scan: chap 3 c=2->2, b=150393->232137 (81745), 226202 ms
[21:07:02] scan: chap 4 c=3->3, b=232138->304872 (72735), 243537 ms
[21:07:02] scan: chap 5 c=4->4, b=304873->413995 (109123), 308485 ms
[21:07:02] scan: chap 6 c=5->5, b=413996->516535 (102540), 311278 ms
[21:07:02] scan: chap 7 c=6->6, b=516536->581851 (65316), 195174 ms
[21:07:02] scan: chap 8 c=7->7, b=581852->654398 (72547), 224200 ms
[21:07:02] scan: chap 9 c=8->8, b=654399->755305 (100907), 325290 ms
[21:07:02] scan: chap 10 c=9->9, b=755306->875617 (120312), 393527 ms
[21:07:02] scan: chap 11 c=10->10, b=875618->935229 (59612), 222374 ms
[21:07:02] scan: chap 12 c=11->11, b=935230->1015425 (80196), 229204 ms
[21:07:02] scan: chap 13 c=12->12, b=1015426->1101235 (85810), 283439 ms
[21:07:02] scan: chap 14 c=13->13, b=1101236->1188573 (87338), 282395 ms
[21:07:02] scan: chap 15 c=14->14, b=1188574->1273248 (84675), 281427 ms
[21:07:02] scan: chap 16 c=15->16, b=1273249->1413894 (140646), 402812 ms
[21:07:02] scan: chap 17 c=17->17, b=1413895->1478523 (64629), 188344 ms
[21:07:02] scan: chap 18 c=18->18, b=1478524->1563256 (84733), 276422 ms
[21:07:02] scan: chap 19 c=19->19, b=1563257->1648560 (85304), 282428 ms
[21:07:02] scan: chap 20 c=20->20, b=1648561->1799252 (150692), 500623 ms
[21:07:02] scan: chap 21 c=21->21, b=1799253->1834547 (35295), 124287 ms
[21:07:02] scan: chap 22 c=22->22, b=1834548->1995231 (160684), 518639 ms
[21:07:02] scan: chap 23 c=23->23, b=1995232->2101646 (106415), 318284 ms
[21:07:02] scan: chap 24 c=24->24, b=2101647->2199280 (97634), 278358 ms
[21:07:02] scan: aspect = 16
[21:07:02] scan: decoding previews for title 1
Scanning title 1...
Scanning title 1...
Scanning title 1...
Scanning title 1...
Scanning title 1...
Scanning title 1...
Scanning title 1...
Scanning title 1...
[21:07:05] scan: preview 1
[21:07:05] scan: AC3, rate=48000Hz, bitrate=448000
[21:07:05] scan: AC3, rate=48000Hz, bitrate=192000
[21:07:05] scan: AC3, rate=48000Hz, bitrate=192000
[21:07:05] scan: preview 2
[21:07:05] scan: preview 3
[21:07:05] scan: preview 4
Scanning title 1...
[21:07:05] scan: preview 5
[21:07:05] scan: preview 6
[21:07:05] Title's mostly progressive NTSC, setting fps to 23.976
[21:07:05] scan: preview 7
[21:07:05] scan: preview 8
[21:07:05] scan: preview 9
[21:07:05] scan: preview 10
[21:07:05] scan: 720x480, 23.976 fps, autocrop = 52/60/0/0
[21:07:05] scan: title (0) job->width:720, job->height:304
[21:07:05] thread ed25e8 exited ("scan")
[21:07:05] thread ed25e8 joined ("scan")
[21:07:05] libhb: scan thread found 1 valid title(s)
+ title 1:
  + vts 1, ttn 1, cells 0->24 (2199281 blocks)
  + duration: 01:54:44
  + size: 720x480, aspect: 1.78, 23.976 fps
  + autocrop: 52/60/0/0
  + chapters:
    + 1: cells 0->0, 113952 blocks, duration 00:05:42
    + 2: cells 1->1, 36441 blocks, duration 00:02:05
    + 3: cells 2->2, 81745 blocks, duration 00:03:46
    + 4: cells 3->3, 72735 blocks, duration 00:04:04
    + 5: cells 4->4, 109123 blocks, duration 00:05:08
    + 6: cells 5->5, 102540 blocks, duration 00:05:11
    + 7: cells 6->6, 65316 blocks, duration 00:03:15
    + 8: cells 7->7, 72547 blocks, duration 00:03:44
    + 9: cells 8->8, 100907 blocks, duration 00:05:25
    + 10: cells 9->9, 120312 blocks, duration 00:06:34
    + 11: cells 10->10, 59612 blocks, duration 00:03:42
    + 12: cells 11->11, 80196 blocks, duration 00:03:49
    + 13: cells 12->12, 85810 blocks, duration 00:04:43
    + 14: cells 13->13, 87338 blocks, duration 00:04:42
    + 15: cells 14->14, 84675 blocks, duration 00:04:41
    + 16: cells 15->16, 140646 blocks, duration 00:06:43
    + 17: cells 17->17, 64629 blocks, duration 00:03:08
    + 18: cells 18->18, 84733 blocks, duration 00:04:36
    + 19: cells 19->19, 85304 blocks, duration 00:04:42
    + 20: cells 20->20, 150692 blocks, duration 00:08:21
    + 21: cells 21->21, 35295 blocks, duration 00:02:04
    + 22: cells 22->22, 160684 blocks, duration 00:08:39
    + 23: cells 23->23, 106415 blocks, duration 00:05:18
    + 24: cells 24->24, 97634 blocks, duration 00:04:38
  + audio tracks:
    + 1, English (AC3) (5.1 ch), 48000Hz, 448000bps
    + 2, Francais (AC3) (Dolby Surround), 48000Hz, 192000bps
    + 3, English (AC3) (Dolby Surround), 48000Hz, 192000bps
  + subtitle tracks:
    + 1, English (iso639-2: eng)
    + 2, Francais (iso639-2: fra)
    + 3, Espanol (iso639-2: spa)
[21:07:05] 1 job(s) to process
[21:07:05] starting job
[21:07:05] thread ed4218 started ("work")
[21:07:05]  + device D:\VIDEO_TS
[21:07:05]  + title 1, chapter(s) 19 to 20
[21:07:05]  + 720x480 -> 720x368, crop 52/60/0/0
[21:07:05]  + filters
[21:07:05]    + Deblock (pp7) (default settings)
[21:07:05]  + video frame rate: 29.970 fps
[21:07:05]  + video bitrate 1000 kbps, pass 0
[21:07:05]  + PixelRatio: 1, width:720, height: 368
[21:07:05]  + encoder x264
[21:07:05]  + audio 160 kbps, 48000 Hz
[21:07:05]  + encoder faac
[21:07:05] thread ed82b8 started ("reader")
[21:07:05]  + output: E:\my x264 and xvid movies\converting\29970.mp4
[21:07:05] thread ed8428 started ("muxer")
[21:07:05] thread ed8600 started ("MPEG-2 decoder (libmpeg2)")
[21:07:05] thread ed8928 started ("Renderer")
[21:07:05] encx264: keyint-min: 30, keyint-max: 300
[21:07:05] encx264: encoding with stored aspect 32/27
[21:07:05] encx264: opening libx264 (pass 0)
x264 [info]: using SAR=32/27
x264 [info]: using cpu capabilities: MMX MMXEXT SSE SSE2 Cache64 
[21:07:05] thread 119a828 started ("H.264/AVC encoder (libx264)")
[21:07:05] sync: expecting 23498 video frames
[21:07:08] dvd: Beginning of Cell (19) at block 1563257
[21:07:08] sync: first pts is 438032326
[21:13:35] dvd: End of Cell (19) at block 1648368
[21:13:37] dvd: Beginning of Cell (20) at block 1648561
[21:13:37] dvd: Chapter Break Cell Found
[21:13:55] MPEG2: Chapter Break Cell Found, searching for GOP
[21:13:55] MPEG2: Group of pictures found, searching for I-Frame
[21:13:55] MPEG2: I-Frame Found
[21:13:55] MPEG2: Chapter Break Inserted
[21:13:57] work Renderer: Copying Chapter Break @ 25450425
[21:26:18] dvd: End of Cell (20) at block 1799084
[21:26:19] dvd: Beginning of Cell (21) at block 1799253
[21:26:19] dvd: Chapter Break Cell Found
[21:26:19] reader: end of chapter 20 (media 20) reached at media chapter 21
[21:26:19] reader: done
[21:26:19] thread ed82b8 exited ("reader")
[21:26:34] sync: got 23484 frames, 23498 expected
[21:26:36] thread ed8600 exited ("MPEG-2 decoder (libmpeg2)")
[21:26:36] thread ed8600 joined ("MPEG-2 decoder (libmpeg2)")
[21:26:36] thread 119a828 exited ("H.264/AVC encoder (libx264)")
[21:26:36] thread ed8928 exited ("Renderer")
[21:26:36] thread ed8928 joined ("Renderer")
[21:26:36] render: lost time: 0 (0 frames)
[21:26:36] render: gained time: 0 (0 frames) (0 not accounted for)
[21:26:36] fifo_close: trashing 0 buffer(s)
[21:26:36] fifo_close: trashing 0 buffer(s)
[21:26:36] thread 119a828 joined ("H.264/AVC encoder (libx264)")
x264 [info]: slice I:189   Avg QP:17.45  size: 27860  PSNR Mean Y:46.55 U:49.49 V:50.39 Avg:47.38 Global:46.97
x264 [info]: slice P:23295 Avg QP:20.26  size:  3989  PSNR Mean Y:44.73 U:48.00 V:48.47 Avg:45.59 Global:45.20
x264 [info]: mb I  I16..4: 41.1%  0.0% 58.9%
x264 [info]: mb P  I16..4:  8.1%  0.0%  1.9%  P16..4: 35.4% 14.4%  2.3%  0.0%  0.0%    skip:38.0%
x264 [info]: final ratefactor: 19.53
x264 [info]: SSIM Mean Y:0.9791730
x264 [info]: PSNR Mean Y:44.744 U:48.013 V:48.488 Avg:45.609 Global:45.211 kb/s:1002.35
[21:26:36] thread ed82b8 joined ("reader")
[21:26:36] mux: file size, 98427507 bytes
[21:26:36] mux: track 0, 98178583 bytes, 1002.36 kbps
[21:26:36] mux: video bitrate error, +230733 bytes
[21:26:36] mux: overhead, 10.60 bytes per frame
[21:26:36] thread ed8428 exited ("muxer")
[21:26:36] thread ed8428 joined ("muxer")
[21:26:36] fifo_close: trashing 0 buffer(s)
[21:26:36] fifo_close: trashing 0 buffer(s)
[21:26:36] fifo_close: trashing 0 buffer(s)
[21:26:36] fifo_close: trashing 0 buffer(s)
[21:26:36] fifo_close: trashing 0 buffer(s)
[21:26:36] Freed 1 buffers of size 512
[21:26:36] Freed 0 buffers of size 1024
[21:26:36] Freed 2048 buffers of size 2048
[21:26:36] Freed 0 buffers of size 4096
[21:26:36] Freed 0 buffers of size 8192
[21:26:36] Freed 0 buffers of size 16384
[21:26:36] Freed 0 buffers of size 32768
[21:26:36] Freed 56 buffers of size 518400
[21:26:36] Allocated 33225216 bytes of buffers on this pass and Freed 33225216 bytes, 0 bytes leaked
[21:26:36] thread ed4218 exited ("work")
[21:26:36] thread ed4218 joined ("work")
[21:26:36] libhb: work result = 0

Rip done!
[21:26:36] thread ed2388 exited ("libhb")
[21:26:36] thread ed2388 joined ("libhb")
HandBrake has exited.

[jbrjake]

Code: Select all

[21:07:05] + Deblock (pp7) (default settings)

Deblocking can easily lead to results like that. Doing image filtering like deblocking or denoising or deinterlacing *always* degrades picture quality to some extent. They add extra processing cruft and throw away detail, so they should only be used when necessary. Detelecining's somewhat different -- when it alters images, it's to just weave together existing fields in the correct order, so there's not much guessing involved and almost no detail is lost.

Also, I'm not sure how your fps ends up getting to 29.97. Are you not using Same as Source? You should be, pretty much all the time. Or it could be a WinGui thing (you're in Windows, right?).

Also, what bitrate and resoution are the files you're getting out of Fairuse Wizard?

[thundar]

Code: Select all

[21:07:05] + Deblock (pp7) (default settings)

Deblocking can easily lead to results like that. Doing image filtering like deblocking or denoising or deinterlacing *always* degrades picture quality to some extent. They add extra processing cruft and throw away detail, so they should only be used when necessary. Detelecining's somewhat different -- when it alters images, it's to just weave together existing fields in the correct order, so there's not much guessing involved and almost no detail is lost.

Also, I'm not sure how your fps ends up getting to 29.97. Are you not using Same as Source? You should be, pretty much all the time. Or it could be a WinGui thing (you're in Windows, right?).

Also, what bitrate and resoution are the files you're getting out of Fairuse Wizard?

Thanks for such a quick response, jbrjake.

Actually, I tried several different encodes using both deblocking and no deblocking. I've done a few using deinterlacing and no deinterlacing. Same result every time no matter what settings I use. I even did an encode using the original framerate. Same result. I use a bitrae of 1000 in both handbrake and fairuse. resolution in handbrake was (I think) same as source (720x??). In fairuse the resolution is 672x288.

So, maybe you can give me a suggestion as to what options I can try? Then I can come back to you with a log and and such.
I'm using windows.

[jbrjake]

Oh, no wonder Fairuse's image looked better.

720*368 = 264,960
672* 288 = 193,536

And you're dedicating the same number of bits per second to each. Generally you want to preserves as much detail from the DVD as you can, which means not scaling the image down any smaller than it is on the DVD. But 1000kb/s isn't enough to preserve the detail in all that, so you end up with mediocre results. In such cases, it's better to throw out fine detail by scaling down the image, so the bits can be spread around the frame more freely.

Use the same as source FPS. Use VFR. Do not deinterlace. Do not deblock. Do not denoise. Turn off anamorphic, and lower the frame size to 672*288. Compare to your Fairuse results.

Also...consider using at least some basic x264 options, like maybe ref=2:mixed-refs:subme=6:me=umh

[thundar]

Alright, I'll try as you suggest. I'll let you know if the encode looks any better. Thanks.

[thundar]

I tried as you suggested, jbrjak. Same result and with lines running throughout. Lines sort of like streaky-like lines that run through the faces, clothes, etc. Not looking too good. I'm not gonna worry about it. I was just curious as to why Handbrake has a tough time sometimes. By the way, the Fairuse encode still has looks better. I'll check back to see if you responded, but like I said, don't worry about. Peace and love and all that happy good stuff.