DV source video encode

[jopereira]

I have used DVD2SVCD only 3 weeks now, but I must say the image quality is very different from what I used to do: DV video to SVCD.

I would like to post a question to all of you who have encode 4:3 and 16:9 programs.

I found the SVCD quality very good when encoding a 16:9 (DVD) program into 4:3 format (adding black borders).
OTOH, I found quality "not very good" when using DV (4:3) source encoded into 4:3 SVCD, especially when I have very fast movements.

Is this because DV source will use 480x576 vs DVD 16:9 which will use less vertical resolution (due to black border)?

Is there a better setting when using DV source? (I did found DVD2SVCD standard settings very good for DVD source)

[mrbass]

I'm not exactly sure. I've play with miniDV 4:3 footage and did encoding CCE 4 pass vbr and wasn't too impressed. Funny thing was then I actually hooked up my minidv camcorder to the TV and was like woooaa..this quality is great and all but not nearly as good as DVD so I think my expectations were too high. You mentioned fast movement. Be sure to set the bottom field first in CCE.

I plan on getting into avi2svcd hopefully this week or next week. Just recently made room on my home computer's hd so I can hopefully play with it.

[jopereira]

Well I realy have Digital8 (sony camera), which is 'something less' than miniDV (less resolution). But we must consider that our cameras are not as good as cameras used by movie makers... and even with very good resolution, that's not all!

I would like to know about your tests. For now I'll try to make some tests using deinterlacing and bottom field first (as you said), since there are some comments in this forum about interlaced movies been more difficult to encode...

Other thing that may cause DVD image been 'better' than miniDV is that we mostly see 16:9 movies on 4:3 TV, which cause us to see 'much better resolution' from DVD.

[gerti67]

Hi there,

last week I had the chance to play around the miniDV camcorder of my parents-in-law and did a small movie (about 8 minutes) with D2S.

I first used the standard settings and as you already described this gave me pixelation during fast motion scenes.

I then tried to optimize the results by playing with the parameters of D2S. First you should try to decide if you want to keep your material interlaced or if you want to deinterlace it because DV material almost always is interlaced with bottom field first.

So if you want to keep it interlaced you should uncheck "Zigzag scanning order" and "Progressive frames" in CCE Encoder tab advanced settings because these settings are for progressive material and they should only be used if you want to deinterlace your material so you convert it to progressive material then.

And also I made sure to select "Bottom field first" with CCE.

Then I edited the AviSynth script to crop the TV overscan area so there's more bitrate left for the visible part of the footage:

...
Crop(24,17,672,542)
BicubicResize(448,542,0.00,0.60)
AddBorders(16,17,16,17)
...

This helped a little bit improving the visual results.

But I didn't had enough time to do more testing - perhaps next week. I plan to use Temporal Smoother, the "mb1 interlaced DV" matrix patched into CCE, the BilinearResize and a higher noise filter setting which is said to improve the results also - and I will report back on that.

Bottom line is, interlaced material needs more bitrate to give comparable results to progressive material like with a DVD source. So you have to make sure that no bitrate is unnecessarily wasted or you deinterlace and turn your material into progressive material. This will help the encoder but deinterlacing also effects the visual quality. I guess the best way would be to use Smartdeinterlace then.

Hope that helps a little, but be aware I just started to play around with DV source and I might be wrong with what I wrote in here. ;)

Greetings,
Gerti

[jopereira]

Originally posted by gerti67

But I didn't had enough time to do more testing - perhaps next week. I plan to use Temporal Smoother, the "mb1 interlaced DV" matrix patched into CCE, the BilinearResize and a higher noise filter setting which is said to improve the results also - and I will report back on that.


Gerti,

Can you point me to CCE patchs for matrix use?


So if you want to keep it interlaced you should uncheck "Zigzag scanning order" and "Progressive frames" in CCE Encoder tab advanced settings because these settings are for progressive material and they should only be used if you want to deinterlace your material so you convert it to progressive material then.

And also I made sure to select "Bottom field first" with CCE.


Your tests are exactly what I was about to do...
I have used TMPEnc until now, but until I found this forum I had no idea about changing parameters to get better results...
Other messages on this forum provide the settings you just describe.

[gerti67]

Hi again,

hmm - you know patching a different matrix into CCE is possible with the use of a little tool called "tsunami cinema craft encoder patcher" but the use of this tool has to be considered as illegal because it patches the EXE file - so I can't give you a link to it because it's against the forum rules. But you might want to try a search with "Google" for example.

Greetings,
Gerti