Cineform HDLink Preference Settings

[David Newman]

Marc,

The pulldown is working fine. The very first few frames the IVT routine has to hunt for cadence and it will deinterlace during that time (upto 7-8 frames only.) After that you have perfect 24p. If your clip did start with so much motion, you may not even have notice. This IVT is designed for live data, that is way it works so well for HDMI 24P sources, so it doesn't search ahead like AE does (slow.) Try a ranges of clips (and longer ones) and I think you will like it.

[Marc Ramirez]

I'll try some longer ones, but I really think that these types of artifacts were persistent for a while. They weren't like this for only a few seconds, more like minutes. I'll try it on some other clips. I think this one is just a little odd for some reason.

[David Newman]

The image you posted is correct on the 8th frame onwards. Happy to look at other clips.

[Marc Ramirez]

I'll try to post another shortly. Continuing on this one.

Thanks for your help.

[Marc Ramirez]

Will you take a look at this one? It is almost 100mb.

It is an extended (funny) version of the short clips I sent you. It was captured in Vegas and then converted in HDLink with pulldown removal enabled.

http://www.yousendit.com/download/WU...TmEzS28wTVE9PQ

I don't know if it the media player classic/ffd show I am using to watch it, but I distinctly see artifacts in the wild girl's face. I can see lines where it was in a previous frame. Can anyone else comment?

[David Newman]

The pulldown has worked correctly. I single stepped the whole sequence in VirtualDub, there are no blended frame artifacts after frame 7. There are some MPEG compression artifacts that are typical for such a heavy motion sequence. The only slightly frame cross talk is in chroma due to 4:2:0 nature of 60i HDV MPEG. In the Canon XL H1 using 24F mode, the frames are encoded progressively, so the x:2:0 chroma signature, which averages chroma line pair will not introduce any cross talk. In the HV20 the 24p signal is encoded as 60i, yet the chroma is actually encoding as if the data was 30p -- 4:2:0 only has one line of chroma value for each scan line pair. As pairs of scanlines can come from different frames due to the pulldown, this can lead to chroma cross talk. In a 60i sequence, 5 frames are extracted to 4 producing 24p, only one of these 4 frames has a chroma signal with cross talk (need a diagram here.) Check out frame 161, the red of her bottom lip on the right hand edge has some chroma shear or crosstalk. The next frame frames are fine, yet frame 165 has some (harder to see) crosstalk. This is not the fault of the pulldown, it is enherent in 60i encoding of 24p at 4:2:0 -- this is the one case where 4:1:1 would be preferable. More proof can be obtained by viewing the image as luma only, you will no longer see the crosstalk. Fortunately our eyes is such a slave to luma, this not apparent in the final viewing conditions. Now for keying work this can be a pain, for that you should use HDMI from an Intensity card (once Blackmagic has worked out the HV20) and capture live. Also another reason of those on the fence to get NEO HDV or HD. :) The HDMI output is 4:2:2, and therefore doesn't have any chroma frame crosstalk.

Should a write this up with diagrams? May I use your images (with credit of course)?

[Marc Ramirez]

Thanks for the detailed explanation. I am happy to know that everything is working (as it should) so that I can move forward and start the editing process for my clips. HDMI out will be a great solution assuming someone addresses this with a portable device.

You have my permission to use this clip. A diagram would be interesting and beneficial, but is by no means required. I do have some background in digital video as I took a digital video electrical engineering course at Stanford. So, if you are going to make it for demonstration purposes to show others, I would like to see it as well. Thanks again for your help. Looks like I'll be upgrading to the full version when my trial runs out. Thanks again, Marc

[Ario Damghani]

I'm trying to use HDLink to convert HV20 M2T's > cineform (direct capture crashes in Vista). I select pulldown removal, and get 24fps but the video looks interlaced on motion. It only looks right when I also select "deinterlace 1080i and dv sources" and pulldown at the same time.

I shot in 24p, why do I still need to check the deinterlace box? Am I loosing quality by doing this?

Also - is there a noticable quality difference between cineform medium and high quality?

Thanks

[David Newman]

Ario,

You should not need to deinterlace progressive footage, and yes that will lower the quality. Post a short M2T clip and I will verify that you have a 24p source. Use yousendit.com or similar. We know this works, we just need to determine why it is not working for you.

Medium vs High, that is up to you; both are designed to look good.

[Ario Damghani]

Here's the raw M2T file that I recorded in 24p mode on my Canon HV20 in TV mode (low shutter speed intentional for slow effect). I have to turn on the "remove 1080i interlace" and remove pulldown functions to make it look right when converting to cineform. If I don't check remove interlace and convert - I get bad interlacing.

http://anivo.com/m2t/interlaced24p.m2t (right click save)

[David Newman]

OK, seems that shutter at 1/6 of second is throwing off the pulldown removal. Please test with 24p rather than 6p. :) We will investigate the ultra slow frame rate.