Hey, first off this procedure for all you peeps is for the best quality
of music videos I know how to do..using tmpgenc and vdub...BTW this
guide works with all capture cards or should..I just use and ATI so
that's what I can base my status on..
First, get VirtualDub http://www.virtualdub.org , then go to the
VCDHelp tools section and download the MJPEG codec for compression
capturing... (NOTICE: MJPEG has a watermark which wont disapear until
you purchase it..so do what you gotta..)
Install MJPEG and then open your standard TV program...for ATI, MMC is
the one..then close that after you see a picture and open VDub.
In VirtualDub go to file -> Capture AVI...
Your TV should now show up on your computer..then, go to Video ->
Compression, and use the settings below for MJPEG..
NOTE: after further study if you change the compression/quality to 19 the file size will drop 3 times! 1.80 gig file will be 600 mb. Play with the settings a bit to check filesizes.
Then, go to Video -> Format and make it 720 x 480 (720 x 576 for PAL),
go to Capture -> Settings and make sure capture audio is checked, and
on the bottom left check off Lock Video Stream To Audio...
Second last look at the bottom right of VDub and you should see 3
little button things saying 11k/8/m 15.00 fps and whatever the last
says..change the first which is the audio to 44k/16/s and the second
which is video to 29.97 fps (25.00 fps for PAL)... and finally go to
Capture -> Preferences and you should see 4 unchecked boxes or 3 that
say SAVE current...check off all 4 or 3..so it will save your
Now, turn on BET or whatever and capture a video or whatever..
When done, look to the right and check the frames dropped, if its over
25 frames for a music video...then you need to clear some space and fix
your settings, but I'm not going to get into that now..
Now on your HDD you have an AVI file about 1.5 - 2 gigs with an ATI.
(400+ mbs or so with a different card).
**SKIP THIS STEP IF YOU DONT WANT TO EDIT YOUR VIDEO BUT DIRECTLY
ENCODE IT AFTER CAPTURING**
Now, if you want to edit the mp3 in or the graphics into the video, get
Adobe Premiere 6.0 http://www.adobe.com
Okay now go to http://www.videotools.net and download Premiere Video
Server Plugin v0.93(with installer). Read the directions on how to
install and install it. (NOTE: It might be hard to work on your first
Go back into Premiere and hit Ctrl + M to output the video and click on
settings on the bottom right. Where it says file type click the
drop-down menu until you get to "video server".
Use the settings as shown below..
Then back at the export movie settings windows click where it says
general and select video.
Make the frame size 480 x 480 and uncheck 4:3 aspect and the frame rate
29.97 (25 for PAL), and over to the right make the Pixel Aspect Ratio:
Square Pixels 1.0. Click Ok.
Now at the export movie screen and call it whatever as it wont affect
anything. Hit Save.
**END OF PREMIERE EDITING**
Now if you are the using the method directly above involving Premiere,
you now have to go into TMPGEnc. http://www.tmpgenc.com
Where it says video source click browse and load c:\ipcserver.avi or
whereever you made ipcserver located.
Now the video source as well as audio should be occupied by
For regular users who skipped the Premiere step just load the avi into
Now, click on settings, bottom right..This is where you make the video
look good. You already captured it and can't do anything about that but
now is the time to make it look good.
In the settings menu use the following settings:
Stream Type: MPEG-2 ...for SVCD/DVD
Size: 480 x 480 ...SVCD size
AR: 4:3 ...So it can be displayed correctly on a TV
Frame Rate: 29.97 (25 for PAL) ...The Standards
Rate Control Mode: Constant Quality ; Setting: Highest 100, Max
2520, Min 2520, Enable Padding, P Picture 0, B Picture 20 ...I use CQ
b/c quality will be constant without any loss but if you use CBR
quality loss can happen, but don't use VBR except if you want to wait
over double the time but if your willing go head.
Profile & Level: MP @ ML ...This is the DVD standard and the
Video Format: NTSC for USA and Pal for Europe
Encode Mode: Interlace ...This makes scanlines appear during
viewing which is the correct way and makes the picture looks sharper
YUV Format: 4:2:0 ...I only have one choice here
DC Component Precision: 10bits ...10bits is the clearest
Motion Search Precision: Whatever you want ...I use Highest
Quality Very Slow because during motion scenes there will be a lot less
pixelation, but sometimes I use normal b/c it saves 10 mins but you
should be able to choose yourself.
This section is very important to the way you view the video..colors,
Video Source Type: Interlace ...The source which was the TV was
interlaced so this is the correct choice.
Field Order: Feild B ...I don't care much for this since I dont
know really what it does but I use the default B.
Source AR: 4:3 525 line NTSC 704x480 (4:3 625 line for PAL)
...This is the AR of the source which is the TV unless you rip it off a
DVD then it would be 4:3, but anyways I usually choose 4:3 Display but
when I load the file it automatically chooses 4:3 525 line so I don't
want to lead you in the wrong direction so choose 525 line..
Video Arrange Method: Fullscreen (Keep AR) ...Yep.
Now we are at the filters menu. This is inportant as I stated above so
adjust your settings accordingly.
First off, double click Noise Reduction and choose 0,1,100 for your
settings. Note: Noise Reduction adds A LOT of time to your encode but
if you don't enable it the picture will look like you know..So the
0,1,100 settings are set to work good without adding a lot of encoding
Next, double click sharpen edge, this is up to you, but don't make it
to high so the picture will again look like.. 84,84 is good 50,50
whatever you like but slide the bar at the bottom around to view the
effects of the filter during different parts of the video.
Now, double click simple color correction, this part will greatly
improve or hurt your picture. First off IF THERE are letterboxes (bars)
make sure you get them to be solid black so tone the brightness down in
the negatives. For darker videos tone the contrast up to 37 or so.
Don't put the gamma in the negatives if you don't want it to look like
its midnight, and I always keep the red, blue settings at 0. NOTE: for
every video the colors are different, so for a dark video don't enable
a high brightness but then again for a lighter video don't enable a
high darkness or it won't look to great. Check the image below.
Look at the letterboxes or bars between to 2 above, if you don't notice anything thats because of the small filesize I had to use and the lack of colors but the one on the left is clearly darker.
And lastely double click de-interlace and select double.
Color corrections thanx to DJ-Rom
Check out my color settings..
Now just click the Audio tab and choose 224 for the bitrate, stream
type mpeg-1 audio layer II, then the system tab choose MPEG-2 program.
If your doing an animated video click on Quantize Matrix and where the
drop-down says default choose cg/animation. Note: I've never tried this
but looking through the settings it seems so be the setting to use for
Choose an output filename and hit Start and wait for it to finish and
your done. If the encoding freezes with an AMD CPU buy a small fan and
open your computer and place the fan next to the computer and by the
way you might want to go to http://www.download.com and search for
HMonitor and change your settings from C to F and if it raises to 150
degrees F your computer will freeze so watch out.
Note: If you wanna create a DVD, just change the size from 480x480 to 720x480!
Much Luv to #urban-videoz on efnet for helping me somewhat with some settings..
Pz, hXizzle 8)
+ Reply to Thread
Results 1 to 24 of 24
In my tests, I haven't noticed much of a difference in file size between MJPEG at a compression level of 20 and huffyuv. So why not just use the free huffyuv codec and get lossless captures.
I was just telling the way I do it becuase my first MJPEG capture was better than all my previous Huffyuv caps
Originally Posted by x2pacalypsehalx
yeah, im talking about the final result..after encoding, but thats just my testing.
Excellent guide! Screen caps showing your best settings, good move. Steps very detailed too, I like that.
I'm saving up to buy the Radeon 8500DV (looks like $300 on pricewatch.com), so I'll see how it goes 8)
Originally Posted by broadband
I am also into music video capture. My tastes are more for the Modern Rock on MTV2's 120 minutes than BET, but video is video.
My method is a little different in the pre-MPEG conversion area and allows you to capture a full show if your HD has the space. No premier is needed.
I capture the full 2 hours using HuffYUV to a 40G drive using AVI/IO (http://www.nct.ch/multimedia/avi_io/) at 480x480 . I like AVI/IO because it remembers your setting and you can open the program and start capturing in less than 2 clicks usually. It is not a free program, but there is limited demo if you want to try before you buy. It has a timer feature which is nice for unsupervised captures. You can also do segmented captures in VirtualDub if you don't want to shell out the money for AVI/IO. I just like AVI/IO's quick setup and interface better.
After I do my capture in AVI/IO, I open the first AVI file of the segments in Virtual Dub (it auto-appends the rest of the segments) and then start the frame server (be sure to manually type in .vdr on the frameserver file name - Vdub forgets to add it). I also sometimes apply a logoaway filter in Virtual dub before starting the frame server, more about that later.
Now open TMPGenc. I use the AVIsynth plug-in (see tools on left) to read the virtual dub frameserve file into TMPGenc. Just open the .VDR file and TMPGEnc treats your full 2 hour capture as one file.
Now, you can set up TMPGEnc conversion settings as x2pacalypsehalx described. I generally leave my brightness/color settings alone. I experimented with the brightness/contrast/color settings of my capture card so that the correction is done in the capture step. This saves time on encoding. Just experiment to find what results in an SVCD with the same brightness as the original source. If you have a DVD or some other source with color bars, use that to find your settings.
To edit out specific videos quickly, open up "source range" in TMPGEnc's advanced settings and set the start and end frames for the video you want. Be sure to pay attention to the audio track at the bottom. Sometimes the video will fade out, but there will still be some audio in there.
Click OK, OK and then CTRL-ALT-M to add that segment to the batch encoder. Give the file a name.
If there are more videos you want, re-open source-range. If the videos follow each other, just click "move to end frame" and then click "set start frame". That will start the next segment where you stopped the previous one. Set the new end frame and then ok,ok, CTRL-ALT-M and save the next segment to the batch list.
Now that you have all the videos lined up in your batch list, just hit RUN and go do something else( or continue using your computer if you need to).
I've been using CBR at full bit rate instead of CQ. I'll have to give CQ at the best setting a try.
I use TOOLAME for audio encoding in TMPGENC. I think it does a better job.
Something I've been using lately is the logoaway filter for VirtualDub. You can find it here: http://republika.pl/vander74/virtualdub/index.html
I started using it because one weekend MTV2 applied a large "Hip-hop month" logo to their already bright and solid MTV2 logo. The filter can blend the surrounding video over the logo to make it less distracting.
Thanks x2pacalypsehalx for your guide. I'll have to try out your TMPGEnc settings. I'm up to over 360 music videos on SVCD so far.
One thing I'm not sure about is why you are using the de-interlace filter. Seems you would not want to do that for SVCD if you plan on watching them on a TV (or a computer with good de-interlacing in the playback program).
hey, how big are each of your 4 minute avi's if you can capture 2 hours on a 40 gig drive!! and logoaway sounds awesome I have to give it a try! thanx for your input, and de-interlace..I don't understand what your saying but, a tv shows horizontal lines and when caputring from a tv you get them lines on playback of the avi..so to get rid of them use the de-interlace feature..
and btw, the advanced users of the "music video scene" on mirc all use premiere b/c we edit the cd audio into the stream with the sound efx, which is sometime really hard but we do it to get better audio and for uncensored audio, and how do you apply that filter logoaway?
A 4 minute 480x480 AVI using HuffYUV = 1.12 GB
About the de-interlace:
You see interlace lines on your computer because it is a progressive display - all the lines are shown at the same time. A TV shows only half the lines at a time but twice as fast and slightly offset from each other. That is why things that are moving show up with lines on the computer - because you are only supposed to see half of those lines at a time
Since your TV is an interlaced display, you don't want to de-interlace the video. Even though you will see the lines on the computer, when you play it back on your DVD player, the TV will show it as it was originally captured - interlaced. The motion will appear smoother that way. You may want to check the "field base" box on your sharpen filter since it is interlaced. I'm testing on that.
I see the reasoning behind using premier for mixing in the original CD audio. I did that once on one video that had really bad sound. I did it by using the WAV file as my audio source in TMPGEnc and adjusting the "audio gap correct" setting by trial and error. I usually do not want to spend that much time working on my videos and just keep the audio as it is.
The LOGOAWAY filter works to reduce a logo by blurring the surrounding video over the logo. There are other settings in there as well. You have to give it the X-Y location and size of the logo and then adjust the settings.
It works very well to remove the logo when there are large areas of the same color around the logo. Areas with a lot of detail and motion show up as a blurred box, which is still better than the logo.
Running some video through tonight with your TMPGEnc setting (QC instead of CBR at max with noise and sharpen filters added). I have video with my old setting still on my HD to compare it to on the DVD player. I'll give you my results later.
I de-interlace b/c im in a mvi group..and viewers usally view the vids on their computers..then delete..sometime they might burn..
HOW DO YOU INSTALL "LOGOAWAY"?
i installed it but what x, y axis is a bottom right logo and the mtv2 one.
Yeah, I can see de-interlacing for computer viewing, although I think most computer MPEG2 players will do their own de-interlacing.
I don't know the position for the lower right logos, you'll have to figure that one out on your own.
For the MTV2 logo, I use position 364, 38 and size 78,47.
These numbers may need to be adjusted for your particular system and if you capture at 720 x 480 then you will need to increase your x values by around 1.5.
The easiest way to find it is to use the solid color filter and then adjust the values until it covers the logo.
for some reason this filter doesnt seem to work do i have to enable it or something
You have to go to video/filters/add and then double-click it to configure it.
Video must be set to "full processing mode" for it to work.
Been Ripping Music Videos With AIW Radeon Since It Cam out ANd Used a Channel Surfer Before it.
HUFFY UV CODEC IS "LOSSLESS" so it retains all RGB information in the Picture. PICVIDEO MJPEG (WHICH I HAVE REGISTED IS LOSSY or Not LOSSLESS) Th MJPEG is Faster And Smaller File Size.
But Better Quality Depending on You Machine. also Most Digital System actually broadcast at 704x480 NTSC and not 720x480 doing so will take CPU power to stretch image the extra 16 pixels to 720i
About the LOGO away Program. I dont really see how software can depict the image under a slodi colored logo. I have tried it and if only workz in light watermarks.
De-Interlacing DOUBLE!??!?!?! That will cause Unwanted Ghosting in high motion scenes and it actually Blurs he Image in low motion scenes. I personally recemmend EVEN (adaptive) for NTSC over BLEND or DOUBLE
the Color adjustments look good.
Edge Enhance ment has its ups and downs . Up is obviously clearer picture.
Downs are It also Sharpens any noise on the picture and it also causes anti-aliasing
if yuou use good Cables and have a good source that wont be needed.
Dont Use Standard Cable Box With Coaxial Cable to try and Capture at 704x480 because analog calbe's max resolutionis 640x480 (4:3) and wont go higher that that, niether will coaxial cable.
Encode Mode Should be NON-INterlace. Because not all DVD- Players support Interlaced SVCD playback.
As For Noise Reduction. It causes RGB data loss and makes image look too plain if over used.
hope This Adds to the grat Explanation above.
Source AR: 4:3 525 line NTSC 704x480 (4:3 625 line for PAL)
is good. But if your Exporting to a frame server at 480x480 (or anything other than 720x480 or 704x480) you need to set "1:1 VGA" or you get the video pushed in at the sides. resulting in "tall Screen Video" where video is tall and way off aspect ratio.
For a while i capped using Virtual Dub and PICVIdeo MJPEG edited in aadobe premiere 6.02 exported audio to wav file. exported video to AVISynth frame server opened both in Tsunami using my custom template which would be better than the settings above in some ways.
and Encode using Constant Bit Rate 2520. That Way the Bit rate is always maxed out as for Constant Quality Will Drop Bitrates to lower onesduring low motion scenes causing block for sudden high motion scenes.
(2 - Pass Might Fix this I never tried though)
Originally Posted by DigitalOv
Originally Posted by broadband
Go with the Canopus ADVC and the result will be far better. You will need a firewire card but those are easily found for around $30 or so.
Why do this instead of buying a $300 ATI card? Quality and overhead.
I've been using All-in-Wonders since they first came out. They're not bad BUT they have to fight all the electronic noise inside the case and don't handle signal extremes. Light colors all blend together so sand, tans, reflective metal, and some people look all washed out. Same with light reflected on water or glass. (see that white cap on the guy in the screen grab? Bet the details int he cap are washed out when compared to the original signal.)
Also, consider that a major portion of that $300 is for a graphics accelerator and that gives you NOTHING with regard to capturing video. Every dollar that goes towards that part of the card is a dollar that DOESN'T go to video capture.
Also, HuffYUV is lossless but at what cost? Lossless compression of a muddy, low-quality input signal is a complete waste. The Canopus DV support far exceeds anything that is possible with a tuner card and the files are smaller, roughly 13G per hour. The inside of any computer is full of noise. Do the conversion outside with quality equipment and the result will be far better than anything you can do inside the box (unless you have pro equipment.)
As far as that goes, use a clean input to the converter. You could use a digital cable tuner or a good VCR. I use a JVC HR-S7600U S-Video deck with integrated TBC and some other goodies. (Sometimes I also use a miniDV camcorder as a recording deck.) These provide far cleaner, more stable signals than a standard VCR. Don't use composite inputs, use S-video. The output from one of these is truly impressive compared to a standard VCR. Black is black, not a swimming mess of red and green. Also, it has some digital filtering for stabilization. Imagine the difference truly horizontal lines make. The whole image isn't one of swirling colors and that means a LOT when you want a quality result.
Oh, that reminds me, just from reading the earlier part of the thread I'm guessing the original source was from a standard video source or integrated tuner which is why MJPEG would look better than HuffYUV. MJPEG throws out some of the signal so it would actually smooth out a low-quality source whereas HuffYUV would keep everything.
For de-interlacing and resizing, use Donald Graft's filters. They're far better than the default ones. http://sauron.mordor.net/dgraft Use frame-only, blend, and motion map for regular video and you'll get very good results. Use this filter with cropping to cut off the ragged edges then his resizer to letterbox with a black background to dimensions in multiples of 16.
Um..ok...yes, broadcast signals aren't exactly 720x480 but grab at that because you're probably digitizing an analog source and the overscan can vary. If your intent is to reduce the image to 1/4 its original size (half width, half height) the math works out far better this way. Half of 702 is 351 which means you'll introduce noise from rounding numbers. Be sure to letterbox to a multiple of 16 in both directions.
Don't misunderstand, the ATI cards are nice and have some great features but you won't get high-quality video from them no matter how hard you try. The system I'm typing this on has an ATI card and the channel surfing and pleasure-watching is a little easier than using a VCR but the capture quality pales in comparison to DV.
Shouldn't try to do this when tired.
Yes, it's not a cap the guy is wearing but the concept is still valid.
Yes, 704 ws the suggested number, not 702. Dumb mistake on my part.
Have a Hercules version of the ATI All In Wonder 8500DV running on ATI install/software CD, (MMC V7.5) on W98SE.
Liked your guide a lot so downloaded VirtualDub V1.3c only problem is when I run it and select capture shows 'error418 - capture device was not detected'. Checked the Control Panel - Multimedia - Properties - Video capture devices and the board doesn't show! Tried re-installing drivers/MMC, but no change. Question - does a capture codec have to be installed for the card to show, or should it show anyway?
If anybody can throw any light on this problem would be most grateful as VirtualDub won't open the .mp2 files that the ATI digital VCR creates!
VIrtualDub does not have MPEG-2 support. You'll need a codec which can work with ATI's files.
Check sites like doom9.org, digitaldigest.com and others.
I have a few links for these on Donald Graft's site:
You could also try using ATV2000 from this site and use a good codec capture codec:
Ssome people like to use MJPEG to make video captures. The most commonly used are from Main Concept, PicVideo, and Morgan (see my links page for demos)
You might want to use HuffYUV as it is lossless but it also needs lots of space (see AVISynth section on my links page) Some people rave about this codec but don't seem to have really tried good DV codecs. HuffYUV takes up a LOT of drive space.
I use the MainConcept DV codec because it is very high quality and fast. It's NOT lossless but there's virtually NO change I can see. It's about $30 to register. Files are significantly smaller than HuffYUV uses. Needs a better CPU but the drain on the rest of the system is far less using DV.
You can probably find some if not all of this software here on VCDHelp as well.
Can you use these settings to capture regular tv programs ? (i.e Old Martin reruns
You could, but why? Grab in one step, compress in another. MPEG compression is lossy and good quality MPEG-2 or 4 needs information about past, present, and FUTURE frames. Can't have future info on the fly unless you've got a large buffer in which to store images.
Then again, if you've got pro studio equipment for in-line encoding you can have good results but you're looking at a huge investment in equipment.
For what you describe, old TV broadcasts, you probably don't need to grab at a high resolution but try playback with the new ffdshow codec. Pick sharpen and add a little video noise. Sharpening gets rid of the muddiness, noise helps hide the edges of the pixels. It will look a lot more like an analog transmission.
Get the codec at http://www.doom9.org
Well, liquid did help me in the vcdhelp channel on IRC but I would like to ask here anyway...he only helped me find a way to frameserv the capped video while adding a text filter...but what I would like to do is filter in the text while capturing the live streaming video...is this possible? I only want to add one simple line of static text that will be in every frame of the entire cap!