DV,HDV,AVCHD convert to img seq clarification

[diffid]

Hi

I'm currently using AVISynth scripts to Convert consumer DV, HDV and AVCHD to image sequencess and HuffyUV AVI's.

Although I'm doing this and used this excellent board for info, I'd like to check settings.

The purpose of the conversion is to use video sources like DV, HDV and AVCHD (not all in the same project) in compositing software like Nuke, mixing with 3D renders etc.

For DV it looks like I should use BT601 and for HDV/AVCHD HD BT709? Specifying matrixes PC.601 & pc.709 rather than Rec601 & Rec709?

Is the difference between Pc & Rec about gamut, precision and quality or is it the remapping/scaling 16-235 to 0-255? Should I just pass the video straight through to images with no levels adjustment?

Is it better to do that image processing in AVISynth or in an environment such as Nuke where it is done in a 32bit space/precision? Or does AVISynth work at that precision anyway?


Should coring be set to false in my scripts? Or does this relate to how the luma is scaled or treated and dependant on exact type of source video color space?


With DV I believe there is a way to blur some of the artifacts in certain channels to improve edges? Can I do this in AVISynth and is it beneficial.

Deinterlacing I assume is needed to be done for compositing work. I've used mcbob and yadif. What would you consider a good approach?


Is it possible to use for example imagewriter to export a proxy image for every image I export via VirtualDub

Is it safe to load DV directly in VirtualDub 1.9.6 & VDubMod, I'm using the cedocida codec not panasonic, and export as image sequences. I believe Virtualdub is not always safe to use for this, not sure whether it relates to full processing mode or fast recompress. Should I use AVISynth instead?

As you can see I have some confusion over best settings to steer me through, there is so much information here to read and understand.

[Wilbert]

For DV it looks like I should use BT601 and for HDV/AVCHD HD BT709? Specifying matrixes PC.601 & pc.709 rather than Rec601 & Rec709?
Yes, indeed.

Is the difference between Pc & Rec about gamut, precision and quality or is it the remapping/scaling 16-235 to 0-255?
It's only the scaling.

Should I just pass the video straight through to images with no levels adjustment?
Yes, indeed.

Is it better to do that image processing in AVISynth or in an environment such as Nuke where it is done in a 32bit space/precision? Or does AVISynth work at that precision anyway?

I don't know about Nuke. Why don't you try this out? Some filters in AviSynth work with 16/32 bit processing.

Should coring be set to false in my scripts?
Yes, otherwise luma is clipped to [16,235] (for chroma likewise).

Is it possible to use for example imagewriter to export a proxy image for every image I export via VirtualDub
I'm not sure what you mean here.

Is it safe to load DV directly in VirtualDub 1.9.6 & VDubMod, I'm using the cedocida codec not panasonic, and export as image sequences.
Sure, why not.

[diffid]

Thank you for your help, I was beginning to think I would have to look elsewhere for help. :-)

With regard to proxy images query, I was wondering whether it's possible to create an avs script that would export a lower quality image for each frame I export at full quality for use as a proxy in Nuke or NLE (although they can both generate proxies) where HD playback may not be achievable on my hardware.

So I wondered whether I can simultaneous write two image formats out .png full qual, .jpg proxy to different folders for each image frame.

Thanks again, I'm a little relieved.