(To The Moderator: I posted this here, but I'm sorry and please move it to the proper forum if I'm wrong)
Hello, and thank you in advance for any help. I've quite a few Blu Ray rips that suffer from moderate to severe video noise. From all I've read I think it's grain (it's as if the video image is covered by a fine layer of white-gray dust that's usually much more noticeable on solid-colored backrounds, especially dark ones). I've attached some images below. I downloaded and used Neat Video's free demo (on my own test piece, not theirs) and was very impressed by the results; however, the pro version (required for 1080p video) costs almost a hundred dollars and is verrrrrrry slooooooow. I was thinking that I could run the videos through VirtualDub, saving them in lossless x264, then recompress them using Simple x264 Launcher, but I'm not sure which plugin(s)-filter(s) would be best. I'm also familiar with MeGUI but very poorly educated regarding AviSynth, although I'm still capable of learning new tricks. Given the attachments' noise, could someone please recommend (a) (preferably free) VirtualDub plugin(s)-filters(s) to help improve my videos? Thanks again for any help, I appreciate it.
![]()
+ Reply to Thread
Results 1 to 15 of 15
-
Last edited by LouieChuckyMerry; 2nd Oct 2013 at 06:24. Reason: Details, grammar, and syntax.
-
neatvideo (amongst others)
http://www.neatvideo.com/examples.html*** DIGITIZING VHS / ANALOG VIDEOS SINCE 2001**** GEAR: JVC HR-S7700MS, TOSHIBA V733EF AND MORE -
Working with HD will require something faster than your current Turion M500, and with more cores. Most would recommend at least a quad-core newer processor at 3GHz or better. My copy of v3 NeatVideo Pro runs standard def cleanup at ~20 fps on my Intel i5, ~12 to 15 for HD. On my pitiful old Athlon II 2.4GHz it runs at a mere 4 fps, and forget using it for for HD on that old PC (which I now use mostly for capture).
You might find it surprising that similar plugins for VDub and Avisynth won't run much faster on a 2.2GHz PC when working with HD. VirtualDub's built-in temporal smoother would probably do the trick at a setting of 3 or 4. I suggest avoiding the temptation to remove 100% of the grain on film sources. Grain is where much of the detail resides in film. Removing it completely softens things somewhat, but the biggest problem will be banding in large expanses of solid color areas that have fine gradations of tone or color (as in big expanses of blue sky, title screens with solid backgrounds, or large areas in the b&w capture you posted). Avisynth has many degrainers, one of them going by the catchy name of RemoveGrain().
A tip about NeatVideo: Never use its default settings, and never use it at full power. Its best used with moderate settings via its advanced interface. If you haven't covered its user manual that discusses the many settings and mods available thru that interface, do that first. Less powerful settings will run faster.Last edited by sanlyn; 21st Mar 2014 at 10:18.
-
themaster1: as I typed in my original post, I'm dreaming of a free AviSynth-VirtualDub filter-plugin that works as well as Neat Video
.
sanlyn: thanks for your reply. I've several short test pieces that I'll run through VirtualDub's temporal smoother and check the results. The computer on which I'm typing this is my old laptop; I've also a newer Lenovo, with an i5-3320M 2.6 GHz dual-core, but it's presently processing video.
When I tested Neat Video I did attack the advanced settings (after viewing these helpful videos: http://www.neatvideo.com/tutorial-Neat-Video-Building-Good-Noise-Profile.html ). How do you think Neat Video rates against any of the available free AviSynth-VirtualDub filters-plugins for results and-or speed? I could probably scrounge up a hundred bucks if it's superior to the freebies. Also, does my plan to run the videos through VirtualDub, saving them in lossless x264, then recompress them using Simple x264 Launcher seem reasonable to you? Thanks again for your time. -
Working with lossless media is defintely the best way to filter and repair video, but why not use more convenient Lagarith lossless compression?
I use NeatVideo specifically as the last step in VHS capture or bad-encodes cleanup. VirtualDub and especially Avisynth can get excellent results, especially with spots, frame repairs, chroma problems, (i.e,, edge stains, rainbows), aliasing, and whatnot that NeatVideo doesn't address. I would say that I use NV on all of my VHS work, and on about 30% of the some of the really rotten problem videos we see in the forum. On that score, I'd say to keep the trial around to test on "problem" material that you've submitted to other filters so that you can compare the results. NV is better at removing a large percentage of grain without crushing too much detail, once you learn to avoid big settings numbers on its high-frequency noise or luma channels. I've found it also to be an excellent sharpener -- and again, get accustomed to the settings. It takes some fiddling at first, but you'll get the idea.Last edited by sanlyn; 21st Mar 2014 at 10:19.
-
jagabo: thanks for the information, I'll give them a look.
sanlyn: so if money wasn't an issue would you recommend Neat Video over other options? I'll check out Lagarith, why do you find it more convenient? Thanks. -
Rather than recommend NeatVideo over some essential Avisynth plugins, I'd recommend it in addition to them. As stated earlier, NV can't repair many image problems (rainbows, stains, spots, rips, frame hops, aliasing, previous bad encoding, etc). NV works only in RGB, while many vido defects can best be repaired only in the original YUV colorspace. That conversion of YUV->RGB requires care to avoid such effects as crushed darks, posterization, improper HD->SD color gamut changes (some of these effects can be seen in the images you posted earlier, especially the bottom two). NV is a rather specialized motion-compensated filter that does address some problems that I find Avisynth overlooks, such as smoothing out the foggy "tape grunge" and clumpy-grtain fades and dissolves seen on many tape and home video captures. I'd keep the trial version around and experiment with various problem videos to see what it can and can't do.
I use NeatVideo on about 25% of the really horrible videos we often see in the forum, and I use it on 100% of my VHS captures in concert with other filters from Avisynth and Virtualdub. IMO you could work with all of them.Last edited by sanlyn; 3rd Oct 2013 at 07:57.
-
I haven't found any freebie plugin that works as well as NV. It's as good a sharpener as I've found as well.
To speed things up you can use an NVidia video card with a lot of memory - NV can utilize the unused memory to add to processing power. Seems it doesn't work with AMD-based cards. Look in the settings, it will even optimize itself to determine the best configuration.
As previously stated, you'll get best results with a mild application of it. It doesn't take much. Don't try to turn a sow's ear into a silk purse. If you overdo it the video takes on an artificial look. I'm not sure where you'd ever use it at full strength. -
I believe those particular movies were shot on grainy film intentionally. Ie, it was an artistic decision, not a technical problem. Hence it would be "wrong" to eliminate the grain.
-
Tough to say on the shot with Tippi Hendrin. But it's been resized, looks cropped and then enlarged, and I don't know how the pic was made. Looks more grainy than it should, it's kinda thick and clumpy, maybe the result of processing. The color is pretty bad. Having seen that movie, I think it should look more like this:
VirtualDub, using another essential filter: gradation curves (above, original -- below,adjusted):
[Attachment 20311 - Click to enlarge]
[Attachment 20312 - Click to enlarge]
In the original post, grain in the b&w frame looks about right, maybe a bit overdone. The same for the clasaroom scene. The 4th image with the three guys has problems, but the grain looks right (I don;t know that movie at all). Might have something to do with the way the O.P. is processing.Last edited by sanlyn; 21st Mar 2014 at 10:20.
-
The images, in order, are from 3:10 To Yuma{1957}, Marnie{1964}, Fast Times At Ridgemont High{1982}, and Primer{2004}. My semi-thorough
research found that all of these Blu Rays' video qualities suffered from some level(s) of grain, with the explanations running the gamut from "it's in the source" to "it's intentional". In his review of Marnie for High-Def Digest, Aaron Peck wrote:
The biggest problem with the way 'Marnie' looks is that it's so inconsistent, frame to frame, that you never know what to expect on a shot-by-shot basis. There are time where the detail and resolution look immaculate (like the beginning as Hedren walks toward a waiting train and all the textual detail in her jacket is completely visible), and then there are times where the picture is downright murky at best (like more than a few close-up shots of Hedren's face). This all adds up to a very unsatisfying experience.
On the other hand, it seems that actor-writer-director-composer-editor-producer Shane Carruth wanted the heavy grain in Primer, as he felt it lent more authenticity to his films documentary-like feel. Personally, I find video grain distracting no matter where I encounter it and would prefer to eliminate it whenever possible; nevertheless, I can appreciate jagabo's opinion, too. I don't feel it necessary to clean up Primer since the grain was intended by the artist, but I included the snapshot because it's an excellent example of the video noise I was originally attempting to describe. The other three videos, however, I would like to improve, if possible, and I thank you all again for your help.Last edited by LouieChuckyMerry; 4th Oct 2013 at 19:39. Reason: Syntax And Grammar, Again (and there're still probably errors)
-
And I was interested in the opposite filter adding grain
Founded only through a license > http://www.compression.ru/video/grain_degrain/index_en.html -
For avisynth:
AddGrainC http://avisynth.nl/index.php/AddGrainC
Grain Factory3 http://avisynth.nl/index.php/GrainFactory3 (requires AddGrainC)
Works with progressive video. If you're processing film-based video, you should IVTC in the first place.
Both free.- My sister Ann's brother -
Wow, two years; time flies when you're encoding video
. Since Gravitator has resurrected this thread--it's alive!--I thought I'd offer an update in the event someone noobier than I actually finds it helpful. After many starts and stops and turns and restarts, and much patient help from many kind people here and on Doom9, I've settled on a basic "go-to" script which, using Komisar's (thanks!) modded 10-bit x264 and SEt's MT AviSynth 2.6, has output consistently good (to my eye, at least), artifact-free results when using uncorrupted progressive SD and HD sources:
Code:###### Quad-Core i7 3840QM With 16GB RAM ###### SetMemoryMax(2048) SetMTMode(3,7) LoadPlugin("SourcePath\LSMASHSource.dll") LWLibavVideoSource("SourcePath.mkv") SetMTMode(2) SMDegrain(TR=1-3,ThSAD=100-1000,RefineMotion=True,Plane=0,Chroma=False,Lsb=True,Lsb_Out=True) ### "TR" and "ThSAD" Depend On Source Quality; The Cleaner The Source, The Lower The Numbers ### F=DitherPost(Mode=-1) S=F.FastLineDarkenMod() ### "FastLineDarkenMod(Strength=24)" For Animated Sources ### D=MT_MakeDiff(S,F).Dither_Convert_8_To_16() Dither_Add16(Last,D,Dif=True,U=2,V=2) GradFun3(Radius=16,Lsb_In=True,Lsb=True) ### Preview Source OR Send 16-bit Output To x264 10-bit ### # Trim() # DitherPost() Dither_Out()
or:
Code:###### Dual-Core i5 3320M With 8GB RAM ###### SetMemoryMax(1024) SetMTMode(3,3) LoadPlugin("SourcePath\LSMASHSource.dll") LWLibavVideoSource("SourcePath.mkv") SetMTMode(2) SMDegrain(TR=1-3,ThSAD=100-1000,RefineMotion=True,Plane=0,Chroma=False,Lsb=True,Lsb_Out=True) ### "TR" and "ThSAD" Depend On Source Quality; The Cleaner The Source, The Lower The Numbers ### F=DitherPost(Mode=-1) S=F.FastLineDarkenMod() ### "FastLineDarkenMod(Strength=24)" For Animated Sources ### D=MT_MakeDiff(S,F).Dither_Convert_8_To_16() Dither_Add16(Last,D,Dif=True,U=2,V=2) GradFun3(Radius=16,Lsb_In=True,Lsb=True) ### Preview Source OR Send 16-bit Output To x264 10-bit ### # Trim() # DitherPost() Dither_Out()
For interlaced MPEG sources, I drop the MT mode and index the demuxed .m1v or .m2v video with DGIndex, then run simultaneous encodes equal to the number of cores after using "Trim()" to split the video into two or four parts.
Of course, as sanlyn mentioned above, no single tool works for every source. As I (quite happily, if very slowly) slog through my video collection, in reverse chronological order for the most part, I'm discovering that as the sources get older they're less willing to cooperate with the above basic script, as the individual sources are seemingly less consistent in quality (that is, unlike a modern film shot on digital video, older sources used different film stock for different scenes so the overall quality of the source varies much more). I therefore will need to improve and adapt, ha ha, and most likely post more silly questions.
Last edited by LouieChuckyMerry; 6th Oct 2015 at 23:26. Reason: Bad Tags!
Similar Threads
-
Recommended VirtualDub (Or Avisynth?) Plugin(s)-Filter(s) To Remove Grain
By LouieChuckyMerry in forum RestorationReplies: 2Last Post: 27th Oct 2013, 22:57 -
How to remove horizontal white lines in VirtualDub/AVISynth
By FrankDrebin in forum RestorationReplies: 11Last Post: 18th Aug 2013, 15:32 -
The best way to remove grain?
By FatalX in forum EditingReplies: 6Last Post: 11th Nov 2010, 11:37 -
avisynth scripts as avidemux2 video filter in Linux (external plugin )
By fahr in forum LinuxReplies: 0Last Post: 12th Mar 2009, 08:37 -
using virtualdub filter to remove text on screen
By snafubaby in forum Capturing and VCRReplies: 2Last Post: 11th Jan 2009, 00:09