Do you have an old VHS or
Video8 or Digital8 or DV tape that you want to convert to a digital
file and maybe upload it on YouTube? First, you need a device
to play your tape. Usually it is the same camcorder
with which you have recorded the tape but if you had sold
it long time ago, you need to get one that is
compatible with your recording. Let me start with this one: it is officially called "DVC"
- Digital Video Cassette - but it is commonly
known as MiniDV. In the early 2000s these
cassettes were re-purposed to record high definition
video in HDV format. To capture DV video you need a
FireWire port on your computer. Many modern computers do not
have one, all they have is USB. There are a few DV camcorders that
are capable of sending video over USB so you may want to
investigate this option. If you have a cassette like this then your recording may be in either
Video8, or Hi8 or Digital8 format. You should not trust
the labeling on the cassette. All of these formats
are standard definition the first two are analog,
while Digital8 is, well, digital. The best option to capture any of these
tapes is having a Digital8 camcorder it will play Digital8
recordings natively and it will transform analog
recordings into digital using built-in
analog-to-digital converter. Then you capture a digital stream
with a FireWire connection, just like DV. In fact, Digital8 uses
the same encoding as DV. There is no difference in quality
between MiniDV and Digital8. Many Digital8 and DV camcorders have
an analog-to-digital pass-through mode, which allows to use them as
standalone analog-to-digital converters. Using passthrough works very similar
to capturing DV video from tape but instead of playing
a tape in a DV camcorder you play it in your analog
camcorder or a VCR. Capture DV stream with
your favorite capture program. For example, WinDV works
great on Windows and is free. The resulting digital file will be encoded
in the industry-standard DV format which is supported by all
major video editing tools. If you have a cassette like this, then your
recording may be in VHS or SVHS format. If you don't remember whether you recorded
your video in regular VHS or in SVHS, the best option would be to get an SVHS
camcorder like this, which covers all the bases. VHS camcorders have no
built-in analog-to-digital converter so the conversion should be
done using a separate device. Regular VHS camcorders have
a composite video output while SVHS camcorders also have S-Video
output, which offers better quality. For analog video capture,
you have several options. If you, like 17 thousand
of other Youtube viewers think that this is an awesome
looking [analog-to-digital] transfer look no further than
a generic EZCap or AV Grabber. It is sold under different
brands for less than $20, hardware and
software included. If all you want is to get
a copy of your tape on a DVD then you need a quality commercial
software that does just that. I cannot recommend
one at the moment. I personally use Total Media package,
but it is not offered for sale anymore. Good software will
afford plenty of bitrate and will preserve interlaced
structure of analog video. Now, if you are more like me and you want to
convert your videos into an edit-friendly format, or you want to create a decent-looking
deliverable for a video-sharing website that does not look like
it was uploaded in 2005 then your chances with off-the-shelf
packages become slim. To be more specific, you want to capture your
home movie in all its antiquated interlaced form and convert it into
a butter-smooth 60p video. You want to crop normally invisible
overscan areas and adjust frame proportions. You want to upscale it so
Youtube afforded it higher bitrate. You want to clean up noise. These demands are hardly met
by consumer-grade packages. But I think that 90% of what
professional hardware and software does can be done with consumer-grade
hardware like Pinnacle Dazzle. This is Pinnacle Dazzle DVC 100, an older
version of the device that Pinnacle sells now. I bought it for $10 on eBay. I did not need software
aside of the Windows driver, which I downloaded from the Pinnacle
website, the link is in the description. I use this converter to
digitize video from a VHS VCR, and from a Super VHS camcorder. To capture from the VCR
I use traditional RCA connectors with yellow composite video and
white and red two-channel audio. My Super VHS camcorder
has S-Video output which provides slightly better
video quality than composite. But it has just a single
monophonic audio output, so I am using a Y-cable
to split it into two channels. Next, I start VirtualDub. In fact, I am using VirtualDub2,
previously known as VirtualDub FilterMod. It is a fork and a further development of the
original VirtualDub, and it has more features. From the device list I choose
my device, Dazzle DVC 100. From Video menu
I select Capture Filter and verify that video standard
corresponds to my analog device. In my case it is NTSC-M, which is
the correct standard for North America. The VCR Input checkbox may improve
synchronization of audio and video, so it does not hurt to check it. Because my camcorder
is connected over S-Video, I select S-Video
from Video Source. In Capture Pin I double-check
the video standard and the frame rate which should be 29.97 for NTSC
and 25 for PAL and SECAM. Color Space shows four-character abbreviations
of supported color subsampling schemes. Navigate to FourCC website for the explanation,
see the link in the video description. In my case, Dazzle offers YUY2, which
is a variant of 4:2:2 color subsampling, and I420, which is a variant
of 4:2:0 color subsampling. For reference, 4:4:4 means that every pixel
has complete brightness and color information only professional video
formats offer this option. 4:2:2 means that color information
is available for every second pixel while 4:2:0 and 4:1:1 describe color
information encoded only for every fourth pixel. I don't see a reason to choose a
lesser format, so I choose YUY2. Next, I select frame size. For NTSC, anything that has
480 lines is good enough. See my video about
frame sizes for details. I am going to choose 720x480 as the
most common and most compatible value although if you don't plan to make a DVD,
then 640x480 may be preferable, and it is easier to work with
because it has square pixels. In Crossbar I verify my choice
of video and audio inputs and how they correspond to each other. Sometimes you may
get video but not audio, so you need to re-jiggle audio
input and source settings to tell the app where
audio comes from. Next, I choose
compression format. There are many options, from
tried and true Huffyuv and Lagarith to UTVideo, Cedocida DV,
GrassValley or Cineform. I prefer Cineform, it is a well-established
visually lossless intermediate codec. It has been purchased by GoPro several
years ago and then open-sourced. VirtualDub2 comes with a native
implementation of Cineform, although playing videos
encoded with this codec in other programs
may be a nuisance, as GoPro does not provide a simple
way to install just a video codec. What is left is choosing
the output file name. Now, I am ready to capture. The preview window shows
video being captured, and there is audio
meter on the bottom confirming that audio
is being captured as well. After capturing what I need, I stop
capture and exit capture mode. I can load the file I just
captured to verify how it plays. So far it is not ready to be uploaded
to YouTube - the video is interlaced, the black bars on the sides
need to be cropped, the noise has
to be removed, the aspect ratio
has to be corrected and the video needs to be upscaled
so Youtube would play it at 60fps. By this point, you should either have
captured DV or Digital8 video over Firewire, or you should have digitized analog video
using something like Pinnacle Dazzle. Let us open the captured file, possibly several
gigabytes in size, and make it beautiful! For this exercise I will use two videos, and
I will be converting them with VirtualDub2, which is a fork and an improved
version of the VirtualDub, a tool that has been
around for two decades. One of the videos is a VHS video
shot on an NTSC camcorder, another is a widescreen video
shot with a PAL DV camcorder. Let me load the first video. I have captured
it with VirtualDub2 and Pinnacle Dazzle. The workspace of VirtualDub contains two
windows: on the left is your source video, on the right is the preview of your
target video with all filters applied. Right now both images look the same,
because I have not applied any filters yet. When you have your video loaded,
try to find a scene with motion. This scene would work, but the car is leaving
the frame, so let me jump to a busier scene. If you watch this video on a small screen, you
can see that this pickup truck looks fuzzy, has a sort of a double
image, it is called ghosting. Let me zoom in,
now you can see individual scan lines in a saw-tooth
fashion, this is called combing. When I advance frame by
frame, the truck remains fuzzy, but the building on
the back looks fine. This is because the building does not move
relative to the camera, while the truck does. If you see combing on moving objects,
it means that the source video is interlaced. Each frame contains
information from two fields; each field stores a picture taken 1/60 (for NTSC)
or 1/50 (for PAL) second apart. This is the situation you will
usually see when working with VHS or Hi8 or DV or even
1080i HDV and AVCHD video. The very first thing I do with interlaced
video, unless I plan to make a DVD, I convert each field into a complete
frame to get rid of combing and to preserve image
rate of the original video. VirtualDub comes with
several pre-configured filters, which become available when you switch
into Full Processing Mode in Video menu. From Filters, select
deinterlacing filter. In the Deinterlacing Mode, choose one
of the first three interpolating filters the top one, Yadif, usually works the best. In the Field Order, choose one of
the Double Frame rate options. Field order - top field
first or bottom field first is important for preserving
continuity of motion. There are only two
choices for field order feel free to choose arbitrarily in a moment we will verify
whether it is a correct setting and we'll choose another one if
the selected option does not work. Now, when I advance
frame by frame, I see movement in the right
window for each keypress, but in the left window, there is
movement for every other keypress. This confirms my assumption that
the original video is interlaced, and it proves that I have successfully
converted each field into a separate frame. This is good. What is not good is that the truck moves
sort of two steps forward one step back. This means that the
field order that I chose does not match actual field
order in the source video. This can be easily fixed by changing
field order to the opposite, in my case I have to change from
Top field first to Bottom field first. Let's verify now the truck moves forward with each
frame advance, so this is the correct setting. What I just have done, I converted
my 30 fps interlaced video into 60 fps progressive-scan video
that I can upload on YouTube and it will not have ugly combing
and will be silky smooth. A quick note about YouTube: until 2014
it supported only videos up to 30 fps but since then higher frame
rates have been added, so you can upload your
home videos at full 50 or 60 fps and they will look the same as they looked
on your TV played directly from a camcorder. I digitized my VHS video
into a 720x480 frame but according to several
international standards only 704x480 part of the
frame represents 4:3 image so I am going to chop off
16 pixels on the sides. In this case I have black overscan borders
that are roughly the same width so I can crop 8 pixels
on each side. If my video was a little skewed, which
often happens with analog video, I could instead crop, say, 9 pixels
on one side and 7 pixels on another. The total has to
come to 16 pixels. The remaining 704x480 frame
represents 4:3 image. If you have originally digitized
into 640x480 frame instead of 720x480, then black bars on the sides
would have been removed automatically and the frame would have been
sampled with nice square pixels. Whichever frame size you have at
this point, 704x480 or 640x480, it represents the complete 4:3 frame. If your video looks slightly
stretched horizontally, right-click on the right sub-window
and select "4:3" aspect ratio. Now it looks right. The next step is optional. On this step I want to remove
several lines on the bottom that show a defect
caused by VCR tracking. But I cannot simply remove them I need to also remove some pixels on the
sides to keep the proportions of the frame. Dividing 704 by 480 gives 1.47,
which is almost 1.5, which is 3 over 2. This means that for every 2 rows that I remove,
I must remove 3 "columns" so to speak. I want to remove 8 rows on the bottom, so I need to reduce the frame width by
12 pixels to keep 4:3 frame proportions. I am going to lose some information,
so it is a judgment call whether you want to preserve
as much as possible, or whether you agree to lose some image
on the sides to obtain a clean frame. Another optional step is reducing image
noise in the form of "analog shimmering" when pixels sort of
sparkle sporadically. Depending on your source and on your capture card,
this shimmering may be more or less noticeable. There are several filters that help reducing
noise. One of them is Temporal Smoother. In my practice, strength between 5 and 7
reduces shimmering completely. But this does not come free: you may lose fine
details, and your image may start looking "plastic-y". It is up to you whether to use temporal
smoother, and how strong you want it to be. Next step is upscaling the video. This step is necessary if you want
to upload your video to YouTube at 60 fps. YouTube enables 60 fps mode only
for videos that are 720p or larger. So, you should resize to at least
960x720 or even to 1440x1080. To do this, choose Resize filter,
turn off aspect ratio control, and set frame size to 1440x1080. Now you
have a nice 4:3 frame in HD with square pixels. According to current broadcast and Blu-ray
standards HD must always have 16:9 proportions, but Youtube is more flexible and
allows HD to have any aspect ratio. The last step is verifying and correcting,
if needed, black and white levels. 8-bit computer imagery uses
0 for black and 255 for white, but 8-bit digital video uses
16 for black and 235 for white. This discrepancy can cause your video
either to look milky with low contrast, or, which is more likely, to lose detail
in near-black and near-white areas. To verify the levels, temporarily add
Histogram filter at the end of the filter chain, then skip to a portion of the video
with dark or black areas and check whether black starts
from the very beginning [of the graph]. In my case, it does. So, I need to sort
of compress the luminance levels. To do this, I am going to add Levels filter
and remap 0 level to 16, and 255 to 235. Put it above Histogram
and verify Histogram again. you should see some unused space on
the left and on the right of the graph. Turn Histogram off or just remove it
from filter list, it is not needed anymore. Now let's choose the encoding
format for the final render. AVC, also known as H.264 is a popular option.
Choose 8-bit variant for better compatibility. Specify either desired
quality level or bitrate. Do not forget a very important
setting: sample aspect ratio also known as pixel aspect ratio. This ratio defines the shape of pixels: whether the pixels are square or rectangular,
and if they are rectangular, what are their proportions. In our case, we have switched from
whatever pixel aspect ratio the original video has to square pixels, so specify 1 in
both SAR width and SAR height. The final step is choosing
audio compression. I choose a ubiquitous MP3 format. Considering the source is linear VHS,
I think that 160 Kbit/s is enough. This is it, now I need
to render the resulting video. I prefer MP4 container, it has
more robust metadata support than AVI and is more compatible with my
editing tools than MKV or MOV. This is how it looks like, ready
to be uploaded on Youtube! The second file that I am
going to encode for Youtube has been shot with a DV camcorder
in Digital Cinema mode. This means, that the camera section
was shooting at 25 full frames per second, but these frames have
been sliced into 50 fields so that they could be recorded and processed
in common at that time interlaced format. When I advance frame by frame,
I don't see any combing or ghosting. This means that I do not need
to convert fields into separate frames
and I can treat source video
as progressive scan video. I still need to crop 16 pixels on the sides if I want
to obtain a frame with 16:9 proportions. I am going to upscale this video to FullHD
so that Youtube afforded higher bitrate for it. Histogram shows that
the black level is correct. Similarly to the VHS video, I choose the
same H.264 encoder with square pixels. Do not forget to set Sample aspect
ratio width and height to 1. I will encode audio
using MP3 at 256 Kbit/s. I will save the resulting
video in MP4 container. That's it! I hope that this was helpful. Please, like and subscribe! Good bye!