TMPGEnc - source file - film or video - deinterlace

[midwestarts]

I am capturing The Sopranos off of TV. I am capturing with an AIW 7500 in MPEG 2 format at the highest constant bitrate possible with my AIW software. I only have enough hardrive space for 30 min if I capture to an avi so I have to stick with MPEG 2.

I then take the video and put it through TMPGEnc to lower the file size. I use an average bitrate of 4700 kbit/s and do a 2 pass VBR with the slow motion search precision (not the slowest).

List of questions:

TMPGEnc asks if your converting video, film or both. What should I use for the Sopranos?

Is it better to deinterlace the video when capturing or do it in TMPGEnc?

Should I get a larger second hardrive and capture the video in an AVI format instead of MPEG2 and then put it through TMPGEnc? Will it produce better results?

Thanks for the help!




Couple of tool things:
I recently picked up a copy of Dazzle DVD Complete - So far it's pretty cool and has a ton of features for a $99 program

Should TMPGEnc be marked as shareware? It is currently listed as freeware but costs $49.00 now.

[tonyp12]

Not enough HD space for AVI, what about Huffy AVI?
It cuts down needed space by 60%

Huffy codec is in the tool section on this website.


TV-captures are interlaced.

Keep it interlaced in tmpgenc.

select: converting video.

[midwestarts]

so I should leave the video interlaced?

[tonyp12]

Leave it interlaced, you will loose details by de-interlace.

PowerDVD have a interlace filter when you watch in on the PC.

Other media players may show the interlace lines,
but will look OK on a TV.

[midwestarts]

thanks! I will check and see if my ATI AIW software can use the Huffy codec

[huykin]

Leave it interlaced, you will loose details by de-interlace.

PowerDVD have a interlace filter when you watch in on the PC.

Other media players may show the interlace lines,
but will look OK on a TV.

The deinterlace of PowerDVD is very basic (probably blend or discard one field), produces ghosting on interlaced video. If your target is TV only, then keeping it interlaced probably gives the best results. But if the target is both TV & computer, a good adaptive deinterlace (say with Smart Deinterlacer or Deinterlace - area based VirtualDub filter) may be better. You may lose a bit of detail on TV, but get a much better quality on computer.

[DJRumpy]

I don't agree. If a source is properly telecined, then reversing the affect should result in a better looking video. Your only returning your video back to the original format. It's a very specific proceedure, with very specific steps to reverse it. I do this all the time on my captures now, and I'm much happier with the result. Look here if your not sure how to determine if your source is telecined:

www.lukesvideo.com

I usually step through a few frames of a movie in VirtualDub to see if it's telecined. It's very obvious once you know what to look for.

Definately drop a some cash for a bigger hard drive. The increase in quality will be noticable. Your loosing quality each time you re-encode your MPEG. Capture with a lossless codec like Huffman. You'll get about a 2 to 1 compression ratio with no loss in quality.

Use VirtualDub, or AVISynth to perform IVTC (this assumes your source material is telecined, and not true interlaced..most current tv shows are telecined).

[bbb]

You can actually create a progressive video stream from ANY source with great quality results. The key is to deinterlace only the parts where the interlaced artifacts show up.

The often misused generalized notion that all deinterlacing of TV material causes loss of detail is an overstatement and simply not always true. Using quick and dirty blend (double) deinterlacing (which I previously used for many months) may make the video look blurier if a subsequent sharpening isn't applied, however, this can be avoided with VirtualDub's Smart Deinterlace filter.

In fact, encoding video as interlaced actually creates problems. First, playback will eventually reveal interlaced artifacts and small areas of color flickering due to imperfect deinterlacing by the dvd player or the dvd software. Second, interlaced artifacts waste bitrate. Why use bitrate to encode the interlace lines when you can eliminate the lines altogether and better use the bitrate to encode actual video detail. Also, I've downloaded interlaced video clips that have macro blocks that show interlaced lines permanently encoded into some of the blocks--obviously a poor encode job which could have been avoided with proper motion based deinterlacing of the source. This can be done using Smart Deinterlace which analyzes motion (from interlaced data) and creates complete, clear picture frames from that data.

One great example of the power of Smart Deinterlace was when I encoded an animated sequence. The problem was that the image (actually several such images switching to completely different images at a blindingly fast rate) was present in between just 2 frames (half in one frame, half in the other frame). At VCD resolution, you can see the image clearly as if it wasn't interlaced. But at SVCD resolution or higher, the entire frame image became a complete frame of interlaced lines--image was totally undisdinguishable from anything. So, I inverse telecined the clip but to my dismay, it completely did away with the image. I tried blend deinterlacing but it blend two images together--making it still hard to distinguish individual images and it became noticeably blurry. Then I used Smart Deinterlace and finally created a separate clear frame of the image (and of subsequent different images--creating 1 frame per image--what I wanted).

Just as a comparison to make sure that Smart Deinterlace (as it was my first time using it) wasn't degrading the image quality or motion smoothness, I encoded 2 clips. One encoded as interlaced with no prior deinterlacing. The other encoded as progressive with Smart Deinterlacing. On a dvd player connect to an analog TV, both looked equally clear and smooth. The difference was on the computer (using WinDVD), the interlaced clip revealed some interlaced artifacts while the progressive clip did not. Also, the progressive clip was still very clear and did not appear to have lost any detail.

You will never go back to interlaced encoding once you use Smart Deinterlacing. And you will be glad you've done so once HDTVs become the household TV standard in the coming years.

[MrKGB]

You can also consider using PicVideo MJPEG instead of huffyv.
It gives very good quality at 19 quality setting with compression ratio about 7. It is also very fast. But it costs about $20.
For capturing I would suggest VirtualVCR (free). It is much better than VDub's capture module.
But the best thing you can do is the following: try encoding using your current method and try to encode the same material using uncompressed AVI->MPEG2 method and then compare the results.
I doubt there will be a lot of difference.
I read an article there one guy tried reencoding a mpeg2 video few times, and he claims that he couldn't see a lot of difference.

[bpjenn]

Just capture at 320x240 and you won't need to deinterlace. The quality from a VHS/8MM/TV source is no better if captured at higher resolutions (640x480/720x480). I have done numerous tests and can find no reason to capture any higher than 320x240. Save yourself the effort and disk space/file size.

[medievil]

Just capture at 320x240 and you won't need to deinterlace. The quality from a VHS/8MM/TV source is no better if captured at higher resolutions (640x480/720x480). I have done numerous tests and can find no reason to capture any higher than 320x240. Save yourself the effort and disk space/file size.

I beg to differ, there is a huge difference in sharpness. I could, however, depend on your capture card. I have a 1996 made Wincast (Wintv) BT848 based card and I can see huge differences in 320*240 VS 720*480 from ANY source.

[bpjenn]

I have used a ATI All-in-Wonder Radeon 32MB, AIW 7500, and the Canopus ADVC 100. I see no difference in quaility from a 320x240 vs. 720x480 encode to MPEG1/2. It could be the source material that I am capturing from-VHS and 8MM.