Hello,
First, i need to know how to do all MeGUI configuration. Second, i record from TV which means i do SD and HD..., the main thing i want is how do i make SD quality better to watch? Which filters do i have to use.
Look at these two screens, the first has been recorded from HD channel 1920 * 1080 and it turned to SD 656 X 368
And the second is originaly SD...
How can i make mine better?
Greetings
Try StreamFab Downloader and download from Netflix, Amazon, Youtube! Or Try DVDFab and copy Blu-rays! or rip iTunes movies!
+ Reply to Thread
Results 1 to 30 of 52
Thread
-
-
No, he wants someone to suggest some Avisynth filtering he might try to improve the output.
Nobody tolls any more. It's not the 90's. The only trolls left seem to be the ones accusing others of trolling.
yaston,
the SD video is never going to look like the HD one, because it was probably compressed a bit too much when transmitted and it doesn't have the same amount of detail to begin with. It also looks like it may have been sharpened already. In what way are you wanting to improve it? -
No no.. Of course it won't be like HD but at least make it better the i recorded it...
Do you know the way ?? -
It really depends on what you want. I use QTGMC a lot for cleaning up video, but it's pretty slow. It's actually a de-interlacer but it has a progressive mode for progressive video and has built-in noise removal if you need it. The noise removal on a low setting also seems to clean up mild "blocking" without losing a whole lot of picture detail (as deblocking filters tend to do). With low noise removal settings (or no noise removal) it also tends to sharpen a little. The simplest way to use QTGMC in a script is like this (progressive mode).
QTGMC(InputType=1)
Or you can add some noise filtering. Low values are generally all that's required. If I use noise filtering, I generally add dither to the end of the script to help reduce colour banding issues.
QTGMC(InputType=1, EzDenoise=1)
gradfun3()
Adding QTGMC and Dither to a script is pretty easy. Getting them to work initially can be a different story as they require several other plugins in order to work. If you need help with that, post back. The required plugins are listed. Follow the instructions nice and carefully, then when it doesn't work, you need to find out what you did wrong.
http://avisynth.nl/index.php/QTGMC
http://forum.doom9.org/showthread.php?t=156028
http://forum.doom9.org/showthread.php?p=1386559#post1386559
Don't expect QTGMC to perform miracles. Crappy video tends to be crappy video and trying to clean it up with filtering has the potential to give you filtered crap. There's no doubt filtering can improve the quality, but it's very subjective and experimenting with filtering can be fairly time consuming as what works for video "A" mightn't work for video "B". Hence the question regarding what you're trying to achieve. -
I tried but i failed !!! Would you send me an avs script file contains these filters you proposed?!!
-
@yaston, others have asked for more information about the videos you work with. hello_hello asked for a copy of the script you are trying to use. No one here understands what "I failed !!!" means in any technical way. What error messages did you see? What script did you use? We have no video sample, no script from you. There is little we can do with no information or with cropped, resized RGB images.
Like hello-hello, I see nothing wrong with the color or levels in the top image you posted of the old woman. We don't know how it looks in the original video because the image has been processed. Except for too much sharpening it looks "normal" to me.
Your second image above of the young woman is too bright. Contrast Levels and color values exceed video standards. It is too red and too saturated. Did it look that way in your original video? We have no way of knowing. Below, I made corrections entirely in VirtualDub with levels, saturation, and gradation curve filters. But I am working with a processed and cropped still image. You cannot use the same filters in the original video.
Without more help from you, there is little anyone can do. If you don't know how to cut unprocessed video samples and post them here (I doubt that you know how to do it), post more information about your source videos. We can't work with processed images. Also, many corrections are not appropriate for Avisynth alone or with MeGUI.Last edited by LMotlow; 5th Jan 2015 at 15:40.
- My sister Ann's brother -
These two pictures are not mine, i just found it and i liked the filters that the encoder used in. i need to know which filters he used ?
If i show you other picture you'll see, take a look:
All of these pics are smooth and face color is brown, i need just to which filters i have too use to get a ggod qyualityLast edited by yaston; 12th Jan 2015 at 14:28.
-
Sorry, but IMO those images look like typical processing errors. They are over filtered and over sharpened, and red and green are too saturated. They look plastic, or like 3D computer animation. The skin tones aren't brown -- they're orange. You can make videos appear any way you want, and perhaps other readers will be willing to teach you how to make eccentric color effects with video. But they won't help you if the only samples you post are still images.
Last edited by LMotlow; 5th Jan 2015 at 20:21.
- My sister Ann's brother -
Ok, i'll give you an original sample with an image after encode and see the deffirence.
-
Hi, look at here again...
Here's two videos before encode (original copy) and after encode (MKV x264 copy)
So, the original copy has been recorded from TV (TS format and 1080p) and it has been converted to 720p
Original copy (before encode):
MKV copy (after encode):
and right now could you explain to me how it done PLZ? -
YouTube re-encodes everything you upload (apparently, I never upload myself) so comparing videos after they've been uploaded to YouTube and re-encoded again isn't exactly ideal. You can attach samples of video to your posts here, or uploading them elsewhere for others to download would be better, as long as they won't be re-encoded again in the process as YouTube does.
Having said that, I'm at a loss as to what it is about the encoded version you're trying to duplicate. The original looks better to me. More natural. The main difference between the two encodes seems to be the colour has been boosted for the second encode, and/or the contrast boosted, but I don't like it. It looks over-saturated to me. The original looks better. -
It's not just the contrast... there's no grain and it's clean and the skin and the hair is glassy little bit. which filters are used?
So, for you what do you propose on me to make the first video (original) better? -
With ffmpeg the filter to look for is "hqdn3d" - dunno if MeGUI supportes this.
https://www.ffmpeg.org/ffmpeg-filters.html#hqdn3d-1I'm the developer behind FFQueue. My posts might reflect this! ;-) -
The first video doesn't have any grain, either. The lower video is too filtered, too sharpened, too saturated. You can make your own videos into whatever you want, but the lower video is ugly. It looks like bad anime. The audio is over modulated and sounds distorted.
We don't know. The top video is not your original video and is not 1080p.
Look at the bottom of any page in any thread. Below the "Reply" window and to its left is an icon that says "Upload files/Manage attachments". Use that control to upload unprocessed original samples to this forum. A few seconds of recorded 1080p will probably not be more than ~150 MB. Give it time to load.
A few days ago you PM'd me with links to a download site that spent 1 hour downloading a 140MB video. The second, smaller link was taking more than 1/2 hour to download 68MB, so I cancelled. Members are willing to help, but post your samples here or find a faster file service.
- My sister Ann's brother -
- My sister Ann's brother
-
I have VideoReDo and use it often for HD edits, but never had this problem. Perhaps others can chime in on decoding the sample. TVMW5 worked for me. But it's not free.
I had same problem with the earlier samples you linked in PM. Thanks for using the forum to post links and samples. We cannot get into detailed discussion in PM, so thanks for understanding that PM is not suitable for this purpose. You will get more help by using forum posts.
Personally I wouldn't do anything with the color in the aa.ts sample, and it has very little noise. You can do as you wish, of course, but it looks clean to me. Some scenes are darker, some are lighter, some have more contrast, but that is the nature of TV broadcast and stage lighting. I downsampled aa.ts and re-encoded to 720p 50fps progressive for mkv in Avisynth. Link: yaston_aa_720p_6k.mkv (33MB), or see the attachment, below. I didn't use MEGUI to encode, as I find it too inflexible for my workflow. I do use other free h264 encoders. The mkv was encoded with TVMW5.
Linked here is the original "Intro - A.ts" that you linked in PM. The original download site showed this file as 140MB (it is 6.6MB). I took the liberty of temporarily loading a copy of it at MediaFire (https://www.mediafire.com/?xcq260zbwzz8ds7) and will remove it shortly. It has few problems, but is a little bright. The first two camera shots are too green, the last shot is too red.
You also linked in PM to a re-encode that you made from Intro - A.ts. The re-encode is too saturated and over filtered, IMO, and the color balance is not quite correct. Link: Intro - B.mkv. I will remove the "B" version shortly.
I worked with the original "Intro - A.ts" yesterday and made a color corrected 1280x720 mkv. Link: yaston_IntroA_720p.mkv . I used Avisynth for the resize and mild cleanup. The color work was done entirely with VirtualDub. I didn't use MEGUI for any of it. I don't think the color work can be done entirely in Avisynth.
I can post Avisynth scripts later, but some Avisynth procedures won't work in MEGUI -- as I understand it. But MEGUI users can show you how to load that file and insert a script with MEGUI. I can post the scripts I used when we pass that point. I'm certain that others have script ideas, too. MEGUI isn't the only free encoder that can re-encode your work.
[EDIT] Link for yaston_IntroA_720p.mkv corrected above. OOOPS!Last edited by LMotlow; 24th Jan 2015 at 13:26.
- My sister Ann's brother -
aa.ts is interlaced - this will cause some sort of blur / linear distortion if not decoded properly. Removing the interlacing will make the image look sharper. Also; when you downscale an image it usually gets sharper. I've attached a quick encode made with the following preset (x264 vbr/crf 20-35) for FFQueue:
///yAQAAeNpNT21PgzAQ/k7Cf+AHdOZg7K2xS5hLzBKjH5x+WZ
al0AKNHWApKok/3qMTZ9retXfPy/XQGNlKe/Q9Jdj2fraZQr1J
kruHfZIkq6rFyJjvVfwsWQhLUMFkHSwiaHzvQwlZn7JayIxplX
5F89j33mR/yg2iW0YIARLDCHzvuFa2Z0AWMxKS+RJDHCNkBQ6H
IUQw74QaVbO66ceK4RYniAYWIWPxqrlEqi27c9oyp0SaqnB5Bi
QackTQFVtA3b4BQEautJWGhYfv41aqCu+aZ5I10gTuE0HDDepT
3tn6Umkp1xrht6+77e5x3XOhcgZ0ElIYSk8v+zV2gUxCAnih9M
9kis/njGvJPpWwJcW/QFBKVZT2os91UaNbeaap0qqS3AQCn9K4
9tWzdSIDfbDNNS9aNjL+zeAAFGiE53e5gX4AIa2X8w==I'm the developer behind FFQueue. My posts might reflect this! ;-) -
- My sister Ann's brother
-
Code:
ffVideoSource("aa.ts") SelectEven() AssumeTFF() QTGMC(preset="fast") BicubicResize(1280,720)
Code:x264.exe --preset=slow --crf=21 --sar=1:1 --colormatrix=bt709
-
I'm the developer behind FFQueue. My posts might reflect this! ;-)
-
Depends on the video. I'm not saying Bicubic() is best for this video. It's just what I picked. The OP will use whatever he wants.
And the OP probably wants to do something ridiculous like this:
Code:ffVideoSource("aa.ts", fpsnum=25000, fpsden=1000) AssumeTFF() QTGMC(preset="fast") Tweak(sat=1.3, cont=1.3, bright=0.6) LanczosResize(1280,720) Sharpen(0.5)
Last edited by jagabo; 12th Jan 2015 at 14:01.