Color space Q: DV video edited -> Output w/ Frameserver 2.0 and Vegas 5 -> CCE 2.67

[garrickk]

I've got a lot of video edited from my shiny new camcorder and I'm ready to make some DVD's. I have Vegas 5.0b setup and am going to export the video data to CCE 2.67 via the newly updated DebugMode Frameserver 2.0 that works beautifully with Vegas 5.

I have questions about color space before I finish, however.

I have my Vegas 5 project with NTSC DV video (bff). It's my understanding that the video is YUV 4:1:1, 720x480, interlaced 29.97, etc.

Now, when I select Render As -> Frameserver, I can output the video as RGB24, RGB32, or YUY2. I selected YUY2 hoping that I wouldn't have to convert from YUV colorspace to RGB and then back again for DVD mpeg2 encoding. Am I correct?

- Is YUY2 4:2:2 or 4:1:1, or does it matter? If it goes from the 4:1:1 to 4:2:2 is it blending data or simply adding zeros? I have no clue.

My current AviSynth script (to be fed into CCE 2.67), accepts my Frameserved video file as input:

AVISource("E:\temp.avi")
SeparateFields()
#ConvertToYUY2()
#ReInterpolate411()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
#DoubleWeave.SelectOdd()

- Now, ConvertToYUY2 is not currently needed because I have Frameserver outputing the video frames as YUY2. But, should I output the video with Frameserver as RGB32 and then use this filter? Would I have better results (in other words, is it possible Frameserver does a poor job at this). Just curious.

- Is ReInterpolate411 necessary? I can't tell a difference in my few tests in quality, but I am slightly color blind.

- I already have CCE 2.67 setup for bff so the last line DoubleWeave.SelectOdd() is not necessary - I believe.

- My settings in CCE are as follows (altered from default at least):
MPEG-2 ES Multipass VBR 3 pass, rates set per file lengh
Quality: no filters, Quantizer 16, DC precision 10
Video: M=3, N/M=4, add sequence end code checked, offset line 1

Now, I like the quality of the video, but I'm just not sure on color space settings at all. Any suggestions would be appreciated before I begin days of compressing video (to be laid out in DVD Architect 2).

Thanks,
Garrick

[Video Dude]

Originally posted by garrickk
- Now, ConvertToYUY2 is not currently needed because I have Frameserver outputing the video frames as YUY2. But, should I output the video with Frameserver as RGB32 and then use this filter? Would I have better results (in other words, is it possible Frameserver does a poor job at this). Just curious.
You always want to avoid color space conversions if possible. If Vegas does not do any hidden internal RGB conversion (like some other editors) serve YUY2. I never used Vegas so I am unsure how it processes the video. I would think it would be best to serve it YUY2.


Originally posted by garrickk
- Is ReInterpolate411 necessary? I can't tell a difference in my few tests in quality, but I am slightly color blind.If it is NTSC DV, use ReInterpolate411.


Originally posted by garrickk
-I already have CCE 2.67 setup for bff so the last line DoubleWeave.SelectOdd() is not necessary - I believe.Correct.

[bsr]

Which DV decoder are you using? If Panasonic or Sony, then they only decode RGB I do believe.

Not sure but I thought I remember reading once that programs like Vegas and Premiere only work in the RGB colorspace. Maybe someone can shed more light on this.

[hendrix]

Originally posted by bsr
Which DV decoder are you using? If Panasonic or Sony, then they only decode RGB I do believe.

Not sure but I thought I remember reading once that programs like Vegas and Premiere only work in the RGB colorspace. Maybe someone can shed more light on this.
Premiere Pro's colorspace is AYUV - so if youre using FrameServer 2.0 selecting YUY2 will yield better quality. I believe Vegas is RGB so I would choose RGB when frameserving via Vegas

[hendrix]

Originally posted by garrickk
Is YUY2 4:2:2 or 4:1:1, or does it matter? If it goes from the 4:1:1 to 4:2:2 is it blending data or simply adding zeros? I have no clue.
NTSC DV is 4:1:1
PAL DV is 4:2:0
Digital-S, DVCPRO-50, Digital Betacam, D-1 and D-5 uses 4:2:2