|
07-30-2015, 08:09 AM
|
|
Free Member
|
|
Join Date: Sep 2014
Posts: 45
Thanked 0 Times in 0 Posts
|
|
|
If a AVT is so subtle that an observer can't tell it is on without comparing the exact same footage, why use one?
I'm trying to figure out which combination of svideo cable and AVT produces the best picture and removes the y/c crosstalk. After I have a cable and AVT that removes the y/c crosstalk, I will borrow the JVC HM-DSR100U for another comparison.
I'm not testing audio. I think the syncing issues are in the software settings. Not the cables or player.
"Earlier you mentioned you've been deinterlacing these videos. The videos seen so far are all telecined, not interlaced. Meaning that they are film-based, progressive video with hard coded 3:2 pulldown."
Nothing posted in this thread has been deinterlaced. It has been specifically stated that I have not posted any deinterlaced files. The first post consist of raw captured avi files. The rest are captured with just Lagarith.
|
|
Someday, 12:01 PM
|
|
Ads / Sponsors
|
|
Join Date: ∞
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
|
|
|
|
|
07-30-2015, 09:22 AM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
No one said that you posted deintelaced videos. You simply stated earlier that you had deinterlaced some of your movies. The caution is to be careful about deinterlacing telecined video. None of the samples you've posted so far is standard interlace.
Quote:
Originally Posted by FrostBite
If a AVT is so subtle that an observer can't tell it is on without comparing the exact same footage, why use one?
|
How do you know the source video doesn't have crosstalk artifacts?
s-video cables don't "remove" crossstalk. Luma and chroma are already separated before they enter the cable. Sometimes the output device doesn't always do a clean job of separating y and uv. The cable won't fix that. It's usually handled by y/c comb filters, not wires. Some of your samples have bad combing and aliasing effects. That's not crosstalk. That's just production or playback problems.
Last edited by sanlyn; 07-30-2015 at 09:43 AM.
|
|
07-30-2015, 11:37 AM
|
|
Free Member
|
|
Join Date: Sep 2014
Posts: 45
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by sanlyn
How do you know the source video doesn't have crosstalk artifacts?
|
I doubt anyone can be certain 100%. But I chose a vhs tape from a big studio at the end of the life of VHS specifically to avoid problems in the recording, for the comparison.
Quote:
Originally Posted by sanlyn
s-video cables don't "remove" crossstalk. Luma and chroma are already separated before they enter the cable. Sometimes the output device doesn't always do a clean job of separating y and uv. The cable won't fix that. It's usually handled by y/c comb filters, not wires. Some of your samples have bad combing and aliasing effects. That's not crosstalk. That's just production or playback problems.
|
If the cables won't fix or remove the crosstalk, why have you been suggesting $35+ s video cables? What problem were you hoping the cables would fix?
|
|
07-30-2015, 12:42 PM
|
|
Free Member
|
|
Join Date: Feb 2011
Location: Vancouver, Canada
Posts: 1,323
Thanked 334 Times in 276 Posts
|
|
|
I think a certain amount of rainbowing and dot crawl are inherent to VHS. Some time ago I looked at a diagram that showed the chroma information partly overlaps with the band where the luma is recorded on the tape, regardless of any S-Video cabling. So it's limited to particular frequencies. Assuming I understood the diagram properly.
|
|
07-30-2015, 03:20 PM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
That same luma-chroma overlay can also occur during playback with sloppy y-c separation, which I've seen in a number of players.
Quote:
Originally Posted by FrostBite
If the cables won't fix or remove the crosstalk, why have you been suggesting $35+ s video cables? What problem were you hoping the cables would fix?
|
Because you mentioned in post #13 you were looking for better cables to avoid interference or affects from other components, or some such problem. So I suggested cables with better design and construction, better insulation, better connectors, better insurance from bent pins and cracked solder joints, more accurate color, better contrast, etc. I didn't get that you meant better cables would remove crosstalk, so I mistook your purpose.
As it is, there's what looks like old fashioned dot crawl in the right borders of the samples. Likely a production problem, because it doesn't appear in every shot. IMO opinion it looks like you're using the better player. It's up to you, but I don't think the DSR1000 can compete with the unit you used for the last 4 samples. That border stuff 's a little glitch that's visible, and better fixed by cropping off and replacing the dirty border.
|
|
08-02-2015, 10:08 PM
|
|
Free Member
|
|
Join Date: Sep 2014
Posts: 45
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by msgohan
That Y/C crosstalk (what sanlyn called dot crawl) is atrocious. I think your S-Video cable is faulty. If you used the same cable for the SR-W5U, I guess the HM-DSR100U unit's S-Video output is busted. Try a capture with composite.
|
Do you see any improvement in the last few videos, using the avt 8710?
|
|
08-03-2015, 08:58 AM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
Quote:
Originally Posted by FrostBite
Do you see any improvement in the last few videos, using the avt 8710?
|
The only difference between the AVT and non-AVT captures is that the 8710 is pumping up gamma. But everybody knows the 8710 does that. The two major problems in both sets are chroma ghosting and some easily cleared dot-crawl-like effect on the right borders of all the frames in both sets.
Compare Gold Series AVI frames 569 and 570, where I assume you used the tbc. Remnants of the boy's hands and other colors in frame 569 appear ghosted in frame 570:
Gold Series AVi - frame 569:
Gold Series Avi - frame 570:
Neither a change of tbc nor a change of cable affrects this problem. Compare non-AVT "Gold Series no tbs" frames 441 and 442, where ghost images from the man's clothing and other objects in previous frame 441 appear in frame 442:
Gold Series no tbs - frame 441:
Gold Series no tbs - frame 442:
Also note the black dots along the right borders of all these images, hwich appears in your oldedr posts, the newer posts, and almost all frames with or without a tbc (these are 2X blowups of a proportions of the right-hand border from each of the above images):
The dots would appear to be either placed there in production or is maybe characteristic of the player. IF this happens on all tapes in that player, it's the player. Dot crawl doesn't appear in other parts of the frames. In any case, the tbc has no effect on either these problems. Chroma lag or ghosting is often an effect from aggressive chroma denoising, and can be a mastering problem in the source. Only way to tell is to play other tapes and/or use other players. Offhand I'd say the tape was created with those problems, as we see a lot of tapes that have that have these glitches, even retail jobs.
|
The following users thank sanlyn for this useful post:
FrostBite (08-31-2015)
|
|
08-04-2015, 09:26 AM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
Chroma ghosting is visible on all 3 color clips. You won't see it on the black and white clip unless you have transitions between grayscale and tinted segments (because luma and chroma are stored separately in YUV). The monochrome clip demonstrates that the temporal carryover is in chroma, not luma. The monochome clip does have mild magenta rainbows.
There's no magic in detecting these effects. Mount a video in VirtualDub and use the scroll controls to slow the video or watch frame by frame. Scene transitions are the easiest places to see these effects.
|
|
08-23-2015, 11:58 PM
|
|
Free Member
|
|
Join Date: Sep 2014
Posts: 45
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by sanlyn
Other than the glitches previously noted, the last two samples are a lot more workable than the early ones. Both animations look a little grungy with a ton of slithery low-level noise during camera motion (no surprise, folks, this is VHS!). Maybe the player's sharpening is making noise and rough edges tougher than they ought to be, but I managed to work on giant_3. It was inverse telecined, filtered, color corrected (the darks are on the edge of crushing and have almost no visible details in the original). All the caps seem to have oversaturated reds that exceed RGB 255. Re-telecined with 3:2 soft pulldown, DVD compliant.
|
Quote:
Originally Posted by sanlyn
Ummm, no, not for 3GB. Why couldn't you compress them yourself? What's this with watching them in a media player that adds its own compression? last I heard, media players de-compress video to play them, not the other way around. Maybe you've been misled. I think, too, you might misunderstand the nature of the lossless compressors mentioned so often here and which are S.O.P. for working files.
Saw enough to see that the Sr-W5U is the better choice, but you still need a line tbc. Many of the problems should be helped along during capture. That includes a line tbc, or you'll be spending a lot of time with unnecessary work and unsatisfactory results.
The two vids I looked at are telecined. The cause of the audio sync problems is debatable, but a frame tbc will certainly help address that, and a line tbc can help as well.
Had just enough time to try to rescue some of the animation. Still some work needed on sloppy edge problems. The attached video was inverse telecined in Avisynth, color corrected and denoised in VirtualDub, and re-relecined in TMPGenc Mastering Works (3:2 pulldown). Some of the remaining chroma smear and edge problems could be fixed with either more filtering (you will lose some more detail) or, even better, a cleaner signal via tbc. |
I don't think I commented earlier, but thanks for taking the time to look into these files further and try your hand at cleaning them up. You're perspective and aid is appreciated.
Would it be possible to share the script you used in avisynth and the filters you used in virtualdub? Were the filters some of the ones that come in the forum download version of virtualdub?
|
|
08-24-2015, 04:45 PM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
Quote:
Originally Posted by FrostBite
Would it be possible to share the script you used in avisynth and the filters you used in virtualdub? Were the filters some of the ones that come in the forum download version of virtualdub? |
I made more than one version, but they used similar scripts. I just now went over them and made somne change, which I'll post here.
I didn't post the scripts and VirtualDub settings earlier because (a) it took some real effort to clean up and color correct the clips. The noise and color aren't your fault, the tape is just a nightmare (welcome to VHS LOL!). And (b) I didn't think you'd go through all the trouble, but I'm used to it with some of my horrible tapes. The attached mpg is a new one.
So here goes....
I used two scripts, an "A" script for the first 27 frames of the clip, and a "B" script for the rest of it. Like a lot of VHS issues, the first scene was so out of whack it looked like something from a different tape. No surprise, as I say. The "AB" script joins the two segments.
Script "A" (first scene):
Code:
Import("Drive:\Avisynth 2.5\plugins\HQdering.avs")
Import("Drive:\Avisynth 2.5\plugins\FastLineDarken 1.3.avs")
LoadPlugin("Drive:\AviSynth 2.5\plugins\old plugins\masktools.dll") # <-- Old masktools v1.5
### --- NOTE:
### Older MVTools, above, is used for FastLineDarken. The new version sucks. ####
AviSource("Drive:\path\to\IronGB.avi")
AssumeTFF().TFM().TDecimate()
Crop(16,0,-8,-4).AddBorders(4,2,4,2)
ColorYUV(off_y=-20)
Trim(0,48) # <-- trim off start of shot + some frames for audio delay.
ConvertToYV12(interlaced=false)
HQdering()
QTGMC(InputType=1,preset="medium",sharpness=0.7)
DeHalo_Alpha(rx=2.0)
MergeChroma(awarpsharp2(depth=30))
FastLineDarken()
Santiag()
ConvertToRGB32(matrix="601",interlaced=false)
return last
The above script was run in Virtualdub with some VDUB filters loaded on output. The VDub filters were Camcorder ColorDenoise and gradation curves. The settings I used are in the attached " IronGB_A.vcf". The output was saved as "IronGB_A" as YV12 Lagarith.
Script "B" for the second segment:
Code:
Import("Drive:\Avisynth 2.5\plugins\HQdering.avs")
Import("Drive:\Avisynth 2.5\plugins\FastLineDarken 1.3.avs")
LoadPlugin("D:\AviSynth 2.5\plugins\old plugins\masktools.dll") # <-- Old masktools v1.5
AviSource("Drive:\path\to\IronGB.avi"")
AssumeTFF().TFM().TDecimate()
Crop(16,0,-8,-4).AddBorders(4,2,4,2)
ConvertToYV12(interlaced=false)
HQdering()
QTGMC(InputType=1,preset="medium",sharpness=0.7)
DeHalo_Alpha(rx=2.0)
MergeChroma(awarpsharp2(depth=30))
FastLineDarken()
Santiag()
ConvertToRGB32(matrix="601",interlaced=false)
return last
The above script was run in Virtualdub with some VDUB filters loaded on output. The VDub filters were Camcorder ColorDenoise, gradation curves, and ColorMill. The settings I used are in the attached " IronGB_B.vcf". The output was saved as "IronGB_B" as YV12 Lagarith.
The "AB" script joins the two segments and doe final cleanup. No VDub filters were used. Saved as "IronGB_AB" with Lagarith YV12.
Code:
vidA=AviSource("Drive:\path\to\IronGB_00A.avi").Trim(0,27)
vidB=AviSource("Drive:\path\to\IronGB_00B.avi").Trim(28,0)
vidAB = vidA + vidB
vidAB
# ---- smooth chroma and luma gaps in the corrected histogram ----
SmoothLevels(16,1.0,255,16,255,chroma=200,limiter=0,tvrange=true,dither=100,protect=6)
dfttest(sigma=8) # <- smooth noise in flat gradient areas.
GradFun2DBmod(thr=1.8,mask=false) # <- blend any hard edges in flat gradient areas
AddGrainC(1.5,1.5) # <- add controlled noise to avoid plastic look
return last
The attached mpg is encoded for DVD as 23.976fps with 3:2 pulldown for 29.97 playback.
|
The following users thank sanlyn for this useful post:
FrostBite (08-31-2015)
|
|
08-24-2015, 05:26 PM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
The SVR-10U and SR-W5U Hogarth clips use an unusual method to get to 29.97fps. They appear to be 25fps PAL film transfers with duplicate frames, then 2:2 blends applied for NTSC. They're a combo of dupes and blended frames. Consequently, you could use regular TIVTC to restore the film speed, or you could deinterlace first and use SRestore(frate=23.976) to get the same thing. It works here, but it might not work on other videos. I'd stick with TIVTC.
I'm not sure what your scripts were designed for, but it's been a long time since I saw anyone use Telecide, which has been replaced by TIVTC. I don't know what TDecimate(mode=1) was to accomplish; it seemed to work here (I haven't looked at every frame), but it's the long way around and is usually used for 15fps anime.
All of the animation clips you posted were telecined or dupe-blended. The scripts won't work on the "power" videos. They're not telecined, they're interlaced. And not interlaced very well, either, with a lot of line twitter and buzzy edges. They look like either DV originals recorded to tape, or just sloppy production. I haven't had time to get into them yet.
I thought the SRV-10U samples looked cleaner. The W5U oversharpens, oversaturates (especially red), and seems to have more chroma smear than the SR-V10U. Unfortunately the 10U can't be used for long-play tapes. The BJC cable seems to have cleaner color and good contrast without trying to blow out highlights. The difference over cheapo or OEM cables is subtle, but after a while you'll see what I mean. The two "Iron" clips and one of the power clips played with bad audio sync. I corrected sync in the attached "Iron" videos.
The script for: "Iron1_SRV10U":
Code:
AviSource("Drive"\oath\to\irontest1.avi")
AssumeTFF().TFM().TDecimate()
Crop(10,0,-10,-4).AddBorders(2,2,2,2)
ConvertToYV12(interlaced=false)
MergeChroma(awarpsharp2(depth=30))
QTGMC(InputType=1,preset="fast",sharpness=0.7)
dfttest(sigma=8)
GradFun2Dbmod(thr=1.8,mask=false)
LSFmod(strength=50,edgemode=2)
DeHalo_Alpha(rx=2.5)
ConvertToRGB32(matrix="Rec601",interlaced=false)
return last
The VirtualDub filters used were Camcorder ColorDenosie and ColorMill. See attached " Iron1_SRV10U.vcf".
The script for "Iron2_W5U.mpg":
Code:
AviSource("Drive:\path\to\irontest2.avi")
AssumeTFF().TFM().TDecimate()
Crop(6,0,-12,-6).AddBorders(2,2,0,4)
ColorYUV(cont_y=10,off_v=-5,cont_v=-25)
ConvertToYV12(interlaced=true)
dfttest(sigma=8)
DeHalo_Alpha(rx=2.5)
MergeChroma(awarpsharp2(depth=30))
# --- convert to VirtualDub filters ----
ConvertToRGB32(matrix="Rec601",interlaced=false)
return last
VirtualDub filters were CamcorderColorDenoise and ColorMill. See attached " Iron2_W5U.vcf".
-- merged --
Oops!!! There's a typo in the second script posted above for "Iron2_w5U". Sorry. Stayed up too late.
Quote:
Originally Posted by sanlyn
The script for "Iron2_W5U.mpg":
Code:
AviSource("Drive:\path\to\irontest2.avi")
AssumeTFF().TFM().TDecimate()
Crop(6,0,-12,-6).AddBorders(2,2,0,4)
ColorYUV(cont_y=10,off_v=-5,cont_v=-25)
ConvertToYV12(interlaced=false)
dfttest(sigma=8)
DeHalo_Alpha(rx=2.5)
MergeChroma(awarpsharp2(depth=30))
# --- convert to VirtualDub filters ----
ConvertToRGB32(matrix="Rec601",interlaced=false)
return last
|
|
|
08-25-2015, 11:32 AM
|
|
Remembered
|
|
Join Date: Aug 2013
Location: Switzerland
Posts: 453
Thanked 84 Times in 74 Posts
|
|
|
Hi sanlyn
LSFmod(strength=50,edgemode=2)
LSF = LimitedSharpenFaster of Didée ?
RemoveGrain V1.0 or V0.9 PR?
either ... or...
see screen
|
|
08-25-2015, 11:46 AM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
LSFmod is an update of LimitedSharpenmFaster. I use both versions. Found here: http://www.digitalfaq.com/forum/vide...dsr100u-2.html. LSFMod has settings that emulate the old version.
I use Avisynth 2.6 and have both RemoveGrain v1.0b and the newer RGTools together in my plugins. Have had no problems. LSFMod requires some support files, listed on the web page link.
GradFun2Dbmod can use either version of RemoveGrain, but as I say I have both versions installed.
|
|
08-26-2015, 05:24 AM
|
|
Remembered
|
|
Join Date: Aug 2013
Location: Switzerland
Posts: 453
Thanked 84 Times in 74 Posts
|
|
Quote:
|
LSFmod is an update of LimitedSharpenmFaster.
|
Thanks, that was new to me.
'M Just a little dated
Quote:
|
I use Avisynth 2.6 and have both RemoveGrain v1.0b and the newer RGTools together in my plugins.
|
RemoveGrain I've known since the first version..war but many years ago.
Avisynth 2.6 I have, actually I RGtools and RemoveGrain v1.0 [2007]
With a test film I've applied your script execution course in VDub.
Approximately 5fps ... is too slow for me.
I now try with the YUV curve in Edius to get an equally good result.
Of the 3 versions
DeHalo_alpha_MT.avsi
DeHalo_alpha_MT2.avsi
DeHalo_alpha_cs.avsi
I use the "MT" version.
DeHalo_alpha_MT (rx = 2.5)
is a compromise, it makes it more one sees strong Blur
Here I actually only interlaced material.
TGMC filters only when the film is in progressive, then go back again to interlaced.
|
|
08-26-2015, 07:21 AM
|
|
Premium Member
|
|
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,308 Times in 982 Posts
|
|
Quote:
Originally Posted by Goldwingfahrer
TGMC filters only when the film is in progressive, then go back again to interlaced.
|
I'm not certain what you mean. After TFM() and TDecimate(), the video is progressive. In the script I used, QTGMC(InputType=1) works with progressive video, not with interlaced, and outputs progressive video with the same number of frames and frame rate as the input. There is no need to interlace.
The video was input to the encoder as progressive 23.976 fps. The encoder applied 2:3 pulldown flags for proper frame rate 29.97 fps during playback.
Original script posted:
Code:
AviSource("Drive"\oath\to\irontest1.avi")
AssumeTFF().TFM().TDecimate()
Crop(10,0,-10,-4).AddBorders(2,2,2,2)
ConvertToYV12(interlaced=false)
MergeChroma(awarpsharp2(depth=30))
QTGMC(InputType=1,preset="fast",sharpness=0.7)
dfttest(sigma=8)
GradFun2Dbmod(thr=1.8,mask=false)
LSFmod(strength=50,edgemode=2)
DeHalo_Alpha(rx=2.5)
ConvertToRGB32(matrix="Rec601",interlaced=false)
return last
MediaInfo of video for the mpg posted:
Code:
Video
ID : 224 (0xE0)
Format : MPEG Video
Format version : Version 2
Format profile : Main@Main
Format settings, BVOP : Yes
Format settings, Matrix : Custom
Format settings, GOP : Variable
Duration : 18s 285ms
Bit rate mode : Variable
Bit rate : 5 533 Kbps
Maximum bit rate : 8 000 Kbps
Width : 704 pixels
Height : 480 pixels
Display aspect ratio : 4:3
Frame rate : 23.976 fps
Standard : NTSC
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Scan order : 2:3 Pulldown
Compression mode : Lossy
Bits/(Pixel*Frame) : 0.683
Time code of first frame : 00:00:00:00
Time code source : Group of pictures header
Stream size : 12.0 MiB (93%)
Writing library : TMPGEnc Video Mastering Works 5 Version. 5.5.2.107
Color primaries : BT.601 NTSC
Transfer characteristics : BT.601
Matrix coefficients : BT.601
Last edited by sanlyn; 08-26-2015 at 07:37 AM.
|
The following users thank sanlyn for this useful post:
FrostBite (08-31-2015)
|
|
08-26-2015, 07:48 AM
|
|
Remembered
|
|
Join Date: Aug 2013
Location: Switzerland
Posts: 453
Thanked 84 Times in 74 Posts
|
|
Quote:
|
'm not certain what you mean.
|
I meant that I often have films in interlaced here.
Since I have to use first TGMC for filtering in Avisynth.
..deutsch to English is a horror.
Quote:
|
After TFM() and TDecimate(), the video is progressive. In the script I used, QTGMC(InputType=1) works with progressive video, not with interlaced, and outputs progressive video with the same number of frames and frame rate as the input.
|
I know that a few years
Quote:
|
There is no need to interlace.
|
yes, but in my films there is the image as TFF, I have to proceed differently.
For example, a cutout
Code:
ConvertToYV12(interlaced=true)
TempGaussMC_beta3(2,1,3,0,0,0,Edimode="NNEDI3",nthreads=8,truemotion=true,sharpness=1.75,Sbb=2,SLrad=1,SVthin=0.75,Sovs=2)
LSFmod(strength=50,edgemode=2)
DeHalo_alpha_MT(rx=2.5)
AssumeTFF().SeparateFields().SelectEvery(4,0,3).Weave()
With VDub after Lagarith YUY2
Then, the film is processed in Edius more ... YUV curve .... etc
|
|
08-26-2015, 07:55 AM
|
|
Site Staff | Video
|
|
Join Date: Dec 2002
Posts: 13,633
Thanked 2,458 Times in 2,090 Posts
|
|
Quote:
Originally Posted by Goldwingfahrer
..deutsch to English is a horror.
|
Nah. It's good probably 99% of the time. 
|
|
08-26-2015, 08:14 AM
|
|
Remembered
|
|
Join Date: Aug 2013
Location: Switzerland
Posts: 453
Thanked 84 Times in 74 Posts
|
|
Optimist
I have here Promptly professional 9
Luckily have your one big serving of humor 
|
All times are GMT -5. The time now is 12:25 AM
|
