• Standards
• Tools
• The Process: Capture
• The Process: Post-Production
• Distribution
• Common Questions
• Credits

The end goal of this howto is to produce a final audio/video file that conforms as closely as possible to the following standards:

• Video Codec: MS MPEG4v2 (not DivX)
• Resolution: 480x360 (previously 640x480, but switched due to quality and playback issues)
• Color: 24 bit
• Video Bitrate: ~900 kbps (really though, as high as you can push it and still fit the final file under 700 MB)
• Framerate: 29.97 fps (NTSC) (I know that at least one encoder is working from PAL tapes, in which case it would only make sense to leave it at the PAL standard 25 fps)
• Audio Codec: MPEG1 Audio Layer 3 (MP3[Preferably LAME])
• Audio Bitrate: 128 kbps Audio Channels: 16 Bit Stereo (If your source tape is mono however, it may make more sense to change this)
• Audio Frequency: 44.1 kHz
• Filesize: As close as possible but not exceeding 700 MB (734003200 bytes)
• Length: 92 min maximum (97 for some of the very early Comedy Central episodes [(this includes the entire episode including opening, credits, and stinger, with all commercials edited out, but including the rotating logo bumpers. also, where applicable, this includes the pre-intro reel announcement by TV's Frank])
• Naming: MST3K - Episode Number - Encode Date (YYYYMMDD) - Episode Name.filetype

The encoding of Mystery Science Theater Episodes requires a few different tools.

1. MST Episodes on tape
2. VCR
3.Video Capture Card
4.Virtualdub, AVI_IO – Video Capture Software
5.MJPEG codec/HuffyYUV codec
6.MPEG4 codec (MS or DivX ;-))
7.10-15 gigs per episode (before. they will drop to ~700 megs)
8. relatively fast CPU

MST Episodes on tape:

A good rule of thumb is to have tapes that don't have bad audio (fuzzy, poppy, noisy, dropout) or video (coloration, dropout, bars, static) and can be considered worthwhile for a VCR viewing on a good TV. Anything that is bordering on barely watchable should only be ripped if no one else has it. Bad rips can always be replaced by better rips, so if you have a better quality source than others do, don't hesitate to rip it!

VCR

Good VCR's = better signal. If you can get S-Video out on your VCR, that's wonderful. Stereo is best too. 4 head over 2 head. Buy a 2 RCA ---> stereo minplug cable from Radio Shack (part # 42-2481) and plug that from your audio outs to your sound card's line in. This way, if you have a cheap capture card with mono audio, you'll skip that and get stereo.

Video Capture Card

Hauppage makes the best cards for price/performance with the WinTV line of Video Capture hardware. These cards are based on the Brooktree chipset, normally bt848 or bt878. The cheap ones have mono audio, but the fix described above solves that. ATI makes a few solutions, the All in Wonder which is a TV card and video card in one and the TV Wonder. The first one is not recommended for if you wish to upgrade your video card, there goes your capture card. Pinnacle offers higher priced solutions. If you want low prices, hit eBay, but the WinTV cards are around 50-60 dollars to begin with.

Video Capture Software

Virtualdub and AVI_IO are two programs that can be used to capture the episodes. Virtualdub is an unbelievably useful, albeit complex, program for video capture and linear editing. It is free. AVI_IO is 25 dollars, but is extremely versatile and can schedule recordings. VirtualDub is a must and AVI_IO is a good idea for a serious encoder.

MJPEG codec/HuffyYUV codec

Uncompressed video is laaaarge so a compression codec is pretty much required. HuffyYUV is a high-quality, lossless codec that compresses to about 2.0:1. The data stream is still pretty heavy, so without a fast hard drive or possibly a RAID, the rate may be too much. MJPEG codecs can compress a little further. The PICVIDEO codec, an MJPEG codec, is free to download.Registration is no longer free, and now costs $18 for a single user licence.

MPEG4 codecs

DivX and MSMPEG4 are two MPEG4 codecs. Microsoft created their version of the MPEG4 codec and DivX is the hack of Microsoft's codec. DivX has gained massive support in the pirate community, yet is unsupported and, due to the fact that it is simply a hacked derivation of Microsoft's codec, is mildly illegal. This guide, however, recommends the legal Microsoft codec. In a side by side comparison, the MS codec has been proven to be better quality and is likely to be supported for a longer time down the road than a hack.

Hard drive (space and speed)

A good 10-15 gigs per episode is a good idea. A good hard drive, preferably 7200 rpm or SCSI and seperate from your system drive is a good idea. Lower quality can fit on smaller spaces, but it's not recommended for a rip that may be used by hundreds of people worldwide.

Fast CPU

A PII-450 is a decent low end machine for capturing and encoding tasks. In an era of cheap Durons, processors can be updated. Capture can stress the processor a little, but mainly the post-production will be the trouble.

1. To prepare, make sure you have a good amount of hard drive space. It is a good idea to make sure that your drive is relatively unfragmented and also to make sure it is relatively empty, as when you reach the end of a drive, the speed can drop and frames as well. Make sure your codecs are installed and everything is running smoothly. Make sure you don't have too much stuff running in the background, dropped frames can suck.
2. This guide will focus on using VirtualDub, mainly because it is free. However, most, if not all, of these funcions can be duplicated in commercial programs such as AV_IO, as well as professional level non-linear video editing tools such as Adobe Premier and Ulead Media Studio. Click File>>Capture AVI. There are a few things you should set up before you capture. Try playing the video. Chances are you will get a TV picture, but if you get the video's picture, you're golden. Type "S" or mouse to Video>>Source. If you have the signal via Composite (yellow RCA in) S-Video, or Coax, select accordingly. You should now get a picture when you play the tape. Click File>>Set Capture File and set where to save the file.
3. Under Audio>>Compression, select either PCM or Microsoft ADPCM. If you are planning on using he Radium MP3 codec you should use Microsoft ADPCM. It is very good quality and compressed quite a bit from PCM. PCM should be used if you are planning on using LAME to create the mp3 audio and/or have no Radium MP3 codec. The "Attributes" should read 44.1 16 bit stereo. Do NOT compress directly to mp3, it is a waste of CPU overhead. Select Audio>>Volume Meter and adjust the levels. I always have a hard time getting this right, so do a few test captures. If Virtual Dub bombs out because there's no audio, but yet you can hear it, check the control panel 'Multimedia' (Windows 98) or 'Sound/Multimedia' (Windows ME) for your preferred recording device. You may have a different device selected, other than your sound card.
4. Go to Video>>Format. Select YUY2, 640x480. If your CPU is underpowered, select 320x240, but this will degrade quality SEVERELY. Video>>Compression and select the Picvideo MJPEG codec from the list. Click configure and click Advanced. Incidentally, it is in this dialog box where you will register the codec as well. Selecting a quality, on the top slider, of 18 will give adequate results. You can go as high as 20 or closer by individually adjusting the Luminance and Chrominance qualities to points in between the 18, 19, and 20 points. Subsampling should be set to 1/1/1. Click OK.
5. Capture>>Settings, and select capture audio. I like to choose wait for OK to capture, as it provides me an easy way to begin my capture. Frame rate should be 29.97 fps for NTSC (North America) video. Capture>>Preferences, save the settings. Capture>>Stop Conditions will allow you to stop the video at a point. This is useful if you need to go elsewhere during the capture. Select Capture>>Hide On Capture to save on CPU usage. Line up the video and start recording via
   Capture>>Capture Video.
6. Watch the side of the capture window. You will get an idea of where you are in the tape, where you are in hard drive space, and how many dropped frames you've had. 1 every couple of thousand (1/6000 or so) is unnoticable. If you drop to many frames in a row, you will start to lose audio sync. Stop it when you think it's done by using the escape key.

So now you've got a 10-15 gigabyte AVI file sitting on your harddrive. How do you get that down to 700 MB without sacrificing significant audio and video quality? This is where VirtualDub and the magic of video filters become your best friend. Post-production methods will vary depending on the quality of the source tapes and how much clean up work is necessary, as well as the computing resources available to you. There are basically four components to the post-producion process. I'll talk about the subtleties of each one individually, and then say something on the way each is best accomplished, depending on the resources available to you. The four components are:

1. Commercial Editing
   
   This one's a no brainer. Some of the more high quality tape traders have taken to editing out the commercials from each episode; most don't. Actually, its just as well that they don't, as doing this editing digitally in VirtualDub is much easier, and will provide for much cleaner commercial cuts than one could achieve by doing VHS deck-to-deck dubbing. VirtualDub has the ability to delete a specified range of frames, which is the easiest way to achieve this commercial editing. However, there have been some issues with editing in this manner and also attempting to do external audio processing , which resulted in severe audio sync problems in the final product. Because of this, we present two different methods of using VirtualDub to edit out commercials. Which one you should use depends on how much harddrive storage space you have, and whether or not you choose to extract the audio to either run it through noise reduction processing, or to process it down to mp3 format using an external encoder such as LAME.
   
   
   • The straightforward method is to simply select the range of frames you wish to delete and remove each commercial break in sequence. To do this, go to the first frame you wish to delete and press 'Home' to mark the beginning of the selection area. Then, go to the last frame you wish to delete and press 'End' to mark the end of the selection area, and press 'Del' to delete the frames. Repeat this for all portions of the video you wish to remove. This will leave you with a continuous video (between 80 and 95 minutes long), minus the commercials. NOTE: at this point you have not actually modified the source video file; the frames that you have selected for deletion will not actually be deleted until the file is resaved as an AVI. However, these editing marks can be saved by using the File>>Save Processing Setting option.
     
     
   • The second method is a bit more roundabout, but may be necessary to solve audio sync problems on some systems. Under video settings, there's an option called "select range." Find the frame that you wish to start your recording. This may be frame 1000, or frame 1, or frame 60. It doesn't matter. Write that frame down and
     scan through to the first commercial break. Now, under select range, put the second number in the length, then the first number in start offset. It's important to do it this way so it does the subtraction for you. Hit ok and go to save AVI. Tell it to defer processing, then name it a1 or 1 or part1, something like that. Repeat this step, finding where you want the encoding to begin again and where you want it to end, putting the frames in, and saving as a2-3-4-5, or 2-3-4-5, or part2-3-4-5, so that Virtualdub will be able to recognize the parts as a whole. After all of the episode is chopped up and set up as jobs, hit F4 to bring up job control. You'll see all your jobs waiting for encoding. Hit start processing and they'll start. Now, they will take a certain amount of time each, but the beauty of this encoding method is that if you want to play Counterstrike or reboot your machine, you can stop it
     in between parts and you won't really lose that much time. Get it? When they are done, open the first file, hit append video segment, auto append (or whatever that option is) and select the second segment. It will build it all together, just select direct stream copy for video and audio and save it. Voila, finished episode.
     
     
   • Some people have been been having trouble using the second method listed above, getting negative numbers for the number of frames selected; after trying several different combinations, after the the first part of the capture is selected, the subsequent selections would all have negative numbers. A way around this is to find the start of the first segment and press the home key to mark that as the beginning of the selection, then find where the first commercial started, and hit the end key to mark that as the end of the selection. Then add filters, set up the video compression, and save as AVI, deferring processing until later, naming it something like part1.avi Find the next segment and repeat the process, naming them part2.avi, part3.avi, etc. You may want to play the video around the parts you plan to edit to be sure the transition is smooth. The settings for filters and compression should still be set from part1, however I recommend using the 'Save/Load Processing Settings' when you have figured out the best settings for that particular episode. After processing with job control, as mentioned above, you end up with several avis with the video processed (but not the audio at that point) and without the commercials. Load part1.avi and then append the rest of the parts to give a whole episode. I save the audio as a wav file and process it in another program, but you could just set up the mp3 compression in Virtual Dub if you dont want to. Finally, save the avi selecting 'direct stream copy' for the video and selecting 'wav audio' for the audio, if you processed it seperately.
     
     

   
   In addition to editing out commercial breaks, it is also possible to append two separate video files together and merge them into one. Why would you want to do this? Well, many of the MST3K tapes do not have complete opening song sequences for a given episode. However, since we're working in a digital world, it is a relatively trivial task to edit in this sequence from a different source. To do this in VirtualDub, open the first chronological clip, and then go to File>>Append Video Segment and select the next file. Keep in mind, however, that these AVI clips must be in the exact same video and audio format. Also, it takes some serious editing kung-fu to merge these sequences together seemlessly and unnoticably to the end viewer. It is possible, though, and encoders are encouraged to do so in attempt to make each episode released as complete as possible.

2. Audio Processing/Encoding
   
   The bane of VHS tape dubbers everywhere is loud audio hiss that results from aging tapes and nth generation copies. Fortunately, it is possible to remove a significant portion of this hiss with noise gate filters; carefull though, as this filtering has enourmouse potential for overuse and abuse. There are numerous software packages available that will do noise gate filtering, though the basic process is the same: isolate a section of the audio that would otherwise be silent but for the audible hiss, take a filter sample from that section, and apply that filter to the rest of the audio track. Sound simple? It can be; however, do keep in mind that while you are removing the hiss from the audio, you are introducing some amount of distortion into the original audio. There is no hard and fast rule to judge how much filtering is too much; the individual should listen to the audio track on a good pair of headphones and use his or her best judgement about what sounds best.
   
   Regardless of whether or not noise reduction is necessary or not, the audio must be compressed down to mp3 format as part of the final encoded file. Depending on whether you used PCM or ADPCM format for audio during your initial capture, the audio portion of your capture AVI file will be between 500 - 1000 MB. Depending on whether or not it is necessary to extract your audio out to a wav file first to process it with noise reduction, there are two different ways to accomplish the audio compression
   • If you do not need to do any other post-production work on the audio, you can do the mp3 compression at the same time as the video compression using the LAME ACM for VirtualDub. To do this, select Audio>>Compression, and choose 'MPEG Layer-3' (you may have more than one option with this title, particularly if you have the hacked Radium codec installed. Make sure that you choose the one labled as LAME, however, as this is generally considered superior to any other mp3 codec out there). The format to choose is '128 kbps, 44,1000 Hz, Stereo (Lame). When you start your final dub to save as AVI, VirtualDub will compress the audio using these settings.
     
     
   • If you've extracted your audio to a wav file to run noise reduction filters on it, you should then use one of the multitudes of LAME mp3 codec frontends to process this wav file to mp3. You will then need to use a program such as 'WaveMP3' to give your file wav headers so that VirtualDub can read it back in. Once you've done this, you can recombine your audio with your video by opening the video file in VirtualDub and selecting Audio>>Wav Audio, and selecting your now compressed wav file. If you use this method to recombine your audio with your video, it is imperative that you check for sync. Check spots in the video at both the beginning, middle, and end to ensure that lips move in sync with the audio and everything is generally kosher. Don't be fooled by the multitudes of badly dubbed Gamera movies that Joel and the 'Bots were forced to watch over the years. Check against host segments to make sure everything is ok.

    

3. Video Filters
   
   This is the part that will make or break the quality of your finished product. VirtualDub has some wonderful filters available that can significantly clean up video noise and other wierdness that is associated wil older VHS tapes. To apply filters in VirtualDub, go to Video>>Filters and add each filter in sequence. Keep in mind that each filter is applied to a given frame in the order that it is listed, and that this order does matter. Here's a list of what most of the encoders use, though your milage may vary, and as always, you need to make a personal judgement on how things look with different combinations of filters:
   • deinterlace (default settings)
     
     

   • levels - chop off the video's pure black and pure white, then adjust the middle fader to get an even range of black to white. make sure you use the intro/host segments as "the true" calibration setting with only minor tweaks for the specific movie. this is because it seems BBI has adjusted their movies' blacks so as to keep the theater silhouettes visible in dark scenes. so unfortunately no movie will ever be calibrated properly, the blacks are always lighter in the movie. so all you can do is calibrate to the host segments with minor adjustments if necessary.

   • hue/saturation/intensity as needed....

   • smart smoother (diameter 7 is the best), strength 30-40 - to figure out the proper strength setting for your tape, calibrate in close up, mid range, and long distance shots. it helps to start with the slider all the way to the left, look at an area of flesh on someone's face. note the video noise. start bringing the slider right until the noise disappears, and no further. the ideal value is actually different depending on how far the camera is from the subject (closeups have different thresholds of what looks good), so you will have to make a tradeoff and I would recommend favoring mid-range shots as this is the predominant shot in a movie (2 or more characters, waist-up field of view). this is a powerful filter and it's easy to overuse it. my rule of thumb is, if my strength is above 40, it's too high. however, it should be noted that you should keep your diameter at 7. it just looks the best in the final video. this should go after your level and color adjustment but before your noise reduction.

   • dynamic noise reduction - you shouldn't have to use this much after the smart smooth. i keep it around 5 or less... this goes AFTER smart smooth and levels.
     

   • fxVHS - this works quite well as a general replacement for the internal dynamic noise reduction filter, and contains some advanced features that are usefull for cleaning up jitter from very poor VHS sources. For the DNR function, again put this after smart smooth and keep it at ~5 strenght. Use the other functions as necessary
     
     

   • crop and resize - apply a resize filter and set it to 480x360. then click on crop with the resize filter selected. trim any distorted edges, as little as possible, and hit ok.

   • logo i put this last. here are the settings. r:24 g:24 b:24 tolerance:7 alpha: 250
     note: the MST3K DAP logo can be found in .BMP format (needed for the logo plugin) here
     
     
4. Video Encoding
   
   Though we all sit here waiting anxiously for the day where bandwidth is plentiful and DVD-Rs are as common as floppy drives, the fact is that the biggest bottleneck that we face in producing high quality video is filesize. In a perfect world, we'd all be trading uncompressed video/audio files around on our terabit internet connections and burning them to FMD discs for archiving. Fortunately, modern MPEG4 video compression combined with mp3 for audio affords compression of ~90 minutes of audio/video to under 700 MB, with minimal loss of quality.
   
   For the MST3K-DAP, we use the official Microsoft MPEG4v2 codec for video compression, as opposed to the hacked 'DivX' binaries that have become quite popular for video distribution. The reasons for this decision were many, but mostly revolved around demonstratably superior video quality and future development/support.
   
   If you don't already have the codec installed, you can download it from Microsoft. In VirtualDub, go to Video>>Compression, and choose the 'Microsoft MPEG-4 Video Codec V2. Click on 'Configure to set the compression settings:
   
   
   • Keyframe - The greater number of keyframes in the video, the smoother seeking to specific points in the video will be, though the qualitity of a given frame will decrease slightly. It is suggested to use a lower number of keyframes; leaving this at the default 'Keyframe every 8 seconds' should be adequate
     
     
   • Compression Control - This will determine the tradeoffs made by the codec when rendering a given frame/motion sequence. It is suggested to set this between 75 and 100, but like all video adjustments/filters, you should test the video yourself to determine what looks best.
     
     
   • Data Rate - This is the most important setting. It determines how much bandwidth is used for the video stream; more kbps = greater video quality. For a 92 minute video (the length of the majority of MST3K episodes from intro to stinger) with a 128 kbps audio stream, the video bitrate should be set to ~ 878 kbps in order to ensure a filesize < 700 MB. It is possible to push this limit a bit, up to as high as 900 kbps, especially if you use LAME to encode mp3 efficiently to a smaller size, leaving a bit more room for the video. If you want to calculate these bitrates manually, you may be able to squeeze a bit more quality out of your encoding. One thing to note, however, is that some of the earlier Comedy Central episodes of MST3K were actually 97 minutes long, in which case you may have to bring this data rate down to as low as 850 kbps

   
   When you save your AVI, combined with either the pre-compressed wav audio or compressing the audio to mp3 at the same time, VirtualDub will compress using these settings to give a single < 700 MB file containing both the audio and video. Encoding time can take anywhere from 8 to 35 hours, depending on the speed of your processor and the amount of video filtering that you apply.
   

As I said, the actual method you use for the post-production process is going to depend on what sort of hardware resources you have available to you, particularly harddrive space and CPU speed.

• If you've got lots of harddrive space, you can afford to process and compress video and audio completely separately, and then multiplex them back together into a single AVI file as the very last step. If you do this, you would edit out the commercials from the original capture, and then output seperate audio and video files by choosing Audio>>Direct Stream Copy and Video>>Direct Stream Copy respectively. Output the audio portion first with File>>Save Wave, and then exlude the audio with Audio>>No Audio, after which you can save the video only with File>>Save AVI. Using this method, after you have processed and compressed both the audio and the video, you can recombine them by opening the final video file, and then choosing Audio>>Wav Audio and pointing it to the final video file. Test to make sure that everything is synced properly, and set both audio and video to "Direct Stream Copy," and save the AVI to the finished product.
• If you don't have enough harddrive space to store another large video file, but still want to process audio and video seperately, you can output the audio as described above, but instead of outputting the video using Direct Stream Copy, do all of your filter/compression work on the original, but edited, file. Be aware, however, that you may experience synchronization problems using this method.
• If you have absolutely no harddrive space to spare, you can do all of your audio/video processing in one step. After you've edited for commercials, make sure both audio and video are set to "Full Processing Mode." Configure the video filters/compression, as well as audio compression. Make sure your audio is set to Audio>>AVI Audio, and then use File>>Save AVI to save your finished product.

Once you've got your finished product as a single AVI file < 700 MB, its a good idea to check three of four spots at different points in the movie to ensure that you haven't lost audio sync anywhere along the way. If everything is kosher, generate an MD5 sum for identification and you're ready for release.

Once you have your completed episode, make arrangements to get it out to people, and then submit a story to the front page to let people know. Besides whatever you decide to write in the story submission, include the following information: the filename, the file size, the running time (hh:mm:ss) and the MD5 sum.

The majority of the information in this Howto was originally posted by members of the MST3K Digital Archive Project in the forums, including:

Austin(stanfinger), drted, fade, paranoid, & Zathras[ars].