DVD to Adobe Premier

[zwalz]

I work for a college athetic recruiting company. We take high school athletes sports videos and compress them for internet viewing for college coaches to evaluate on their computers.

Many of our kids submit DVD disks that have their highlights on them. Originally I was capturing these highlights by way of analogue capture with my video capture card plugged in directly to the DVD player. I would then pull that file into premier for editing, kick that file out using a dv codec and finally put into sorenson for compression for the web.

I found some help on this site referring me to dvd ripping programs that could save me some time from analogue capture. I purchased win avi converter and have been using the dv codec in the list which i assume comes from my computer? The quality however is not that great. My questions are, is the dvd converter the best way to do this and if so, which codec can I purchase / find to convert these sports videos (1hr) in a timely fashion while still maintaining the best quality and still being able to import into premier for editing?

Ive even uploaded one frame of the video that was outputted using the "dv codec" in win avi. As you can see, there appears to be many silohouettes of the original and overall, it looks terrible. Any specific codec recommendations that would output a better looking video?

[edDV]

It's not blurry, it is a poorly deinterlaced display of an interlaced stream. If we are looking at the actual video here, then the damage has been done.

If the incoming DVD's have been deinterlaced, the damage is done.

If the incoming DVD's are interlaced. Keep them interlaced.


PS: "compress them for internet viewing for college coaches"

Interlaced video has 1/60 sec motion resolution for each field. You are seeing parts of two fields here. You are also compressing for computer progressive monitor display. The simple way to do this is to toss all the odd or even fields resulting in 1/30 sec motion resolution and half the vertical resolution. The resulting 720x240 field stream can be compressed to your target size and bitrate.

If you attempt to deinterlace, you will get the kind of artifacts you show above. Sports analysis usually values stop motion. If 1/30 sec motion resolution is adequate, drop the field.

Advise those providing DVD to encode interlaced, not progressive.

[zwalz]

Wow,

Thanks for the quick response! Unfortunately, I have very little knowledge past what Im currently doing so I think this may be way over my head. We have athletes all over the country who submit dvd's from their high schools, independent video companies, home made etc - what we get is what we have to accept. Most kids already have the dvd's before they even sign up with us so suggesting a format would be very difficult.

The image I sent you was the video after it was put through the win avi converter. I dont see any selections on the interface to interlace or deinterlace the video.. To be truthful I dont even know how to tell if the original is or isnt. Maybe the Win Avi converter isnt the right tool for me? Or maybe it is and youre sugesting that the video itself may have been already porly deinterlaced? Im really sorry but confused.

[edDV]

I have never used Win Avi converter. Maybe somebody else can help.

I'll give my "process" suggestions. If nobody responds I'll get back this evening.

Most DV camcorders (unless ABC or ESPN HDTV) output 720x480 interlaced. A DVD can be encoded interlace or progressive. If the DVD is encoded interlace, you can easily drop a field giving the equivalent of a 720x240 entact image. This can be compressed to wmv or whatever and will look good for stop action at 1/30 sec stop motion resolution.

If the DVD was encoded as progressive, it will come to you looking like the picture above. It will also have 1/30 sec stop motion resolution but you are seeing parts of odd and even fields in the same picture. There isn't much you can do to fix it. So you would just compress it as is and explain away the poor stop action.

More sophisticated techniques can use both fields for 1/60 sec stop motion but most coaches aren't complaining about NFL films which mostly uses 24 fps film for 1/24 sec stop motion.

[zwalz]

Ok thanks - that does help quite a bit.. All this time I thought it was the codec selection.

At least I know its beyond my control. Just one last question - why do the two interlaced fields show up after conversion but not on the tv screen when watching the dvd?

[edDV]

Ok thanks - that does help quite a bit.. All this time I thought it was the codec selection.

At least I know its beyond my control. Just one last question - why do the two interlaced fields show up after conversion but not on the tv screen when watching the dvd?

An interlaced TV set displays a sequence of odd and even fields at a 60 field per second rate. A typical 4 head VCR can stop action on any of the fields.

A computer must display a progressive image. When the source is interlace, a progressive image is constructed by combining the two fields into one picture. This can be done in various ways but the result is a blend of two fields that are offset in time by 1/60 sec. When there is motion in the scene, this results in a still like you show above.

If you were just playing the video, the double imagery would be averaged by the eye and playback wouldn't look too bad. If you add compression things would get bad very quickly.

For sports video, people want to freeze and then analyze frame by frame. They want a clean frame, not what you show above. The easy way to get there on a computer display is to drop one field.

The tradeoff is 1/2 vertical resolution and a slightly jerky 1x playback. Most people accept this on a computer display where they would complain if you did it for TV playback. That is why encoding methods differ for TV vs. computer display.

[zwalz]

Awesome! Makes much better sense. Thanks for your help.

[harley2ride]

If you are talking about frame captures from video, after you capture the image, you need to deinterlace it (I use photoshop) to make the captured image look good as a photo. Otherwise, what these other guys said is good..

[edDV]

If you are talking about frame captures from video, after you capture the image, you need to deinterlace it (I use photoshop) to make the captured image look good as a photo. Otherwise, what these other guys said is good..

Photoshop's "deinterlace" filter looks for field split to determine if motion is present. If it finds motion it defaults to dropping a field. It it finds no motion, the fields are weaved. Software deinterlacers used with realtime or post encoded video can't afford the luxury of time to do this on every frame. High end hardware deinterlacers have sophisticated motion analysis and can adaptively deinterlace video in real time. This requires multiple frame buffers to determine motion vectors. The cost of doing this exceeds typical consumer budgets but is beginning to show up in high end HDTV sets.