http://www.dvinfo.net/conf/showthread.php?t=90148
cool

love CineForm Btw

Tom, I guess I still don't get where this leads us? Even if it turns out that F-Mode uses PsF, that would mean that HDV 24F ought to appear as 47.952 fields/sec or 23.976 frames/sec in the NLE -- unless 2:3 or 2:3:3:2 pull-down was applied, which of course would be very inefficient and I personally think highly unlikely.

Even if 24F is sent by 1394 as PsF at 23.976, it started progressive and is read by F-Mode-aware NLEs as 23.976 so it really doesn't matter and we're all just speculating anyway. If it helps us end-users to know the data protocol beyond what's described in the manual and by the documentation of the NLEs we use, let me know and I'll see if I can get some additional factual details (no promises, as a lot of the detail may still be propriety secrets).

EDIT: Well, Salah posted while I was typing. There it is. Kudos for using the SEARCH feature...shoulda done that myself, since I'm usually first in line to remind people to do so!

Thanks Brian. I credit Barlow Elton for getting us on track though. Your observations are excellent as well.

Me too, I'd like to thank everybody for the valuable and most interesting and educating contributions! When I posted the original question I never thought I'd raise so much dust / questions and disagreements. It's incredible how much I learned from this. Thanks again!

Dino

There are lots of standalone capture utilities that report fps numbers that are wrong. I've watched a couple of clips that were 23.97 reported as 29.97 in different software.

Good explanation, Tom. I just know what it looks like, never knew how it got there.

The last few posts seem confusing to me, so probably will be to those reading this thread in the future looking for facts. To clarify, Canon and Cineform (one of the first third party companies to support Canon F-Mode) are both on record as stating that HDV F-Mode via 1394 is a purely progressive data stream.

Some freeware applets or other software report might possibly report otherwise.

24f footage (xh-a1) will not cut together on the same timeline as 24p footage (HV20).
The 24f is captured (in FCP by the 1080/24p preset) and works great on a 24p timeline without render.
The 24p footage cuts on a 1080/60i timeline fine. But if you want to take it to a 24p time line you must capture it using the Apple Int. Codec and then reverse telecine with Cinema Tools to remove the pulldown. Then it will work on a 24p timeline. BUT not with 24f footage without rendering the 24p footage b/c the 24f is captured with an HDv codec, not AIC. If you try and capture the 24f stuff with the AIC, it captures at 29.97 not 23.98.
Crazy.

Ryan,
There are differences among editors. 24F cuts right with 60i for me on the same timeline, no rendering.

You only have to look at the chroma samples to figure out what the video is.

mpeg2 uses two different methods of 4:2:0 color space depending on if the video is encoded as interlaced or progressive. With progressive the chroma pixels are 2x2 pixels in size for a clean result only not as detailed. Interlaced on the other hand has to alternate the chroma samples every other line because every other line is a different moment in time. If the samples didn't alternate every other line then field A would always look good while field B would always look very bad. By alternating the chroma samples the encoder and decoder treat the video as 1440x540 and then sample at 4:2:0. The video is then interweaved back into 1440x1080. It is a huge pain in the rump but it works great for interlaced video.

The reason why it works well for interlaced video is because only one field is shown at a time anyway. Progressive inside of a interlaced encoded file however can have chroma problems. This is because the chroma is treated as interlaced so it alternates but when viewing as true progressive the lines form the same frame so you do not want the chroma samples to alternate.

All of the F mode footage I have seen has had 2x2 pixel chroma samples without any sign of the chroma ever being in the interlaced form.

Even 24p DVD's can show this. The video is 24p but there are repeat field flags that tell the DVD player to output 60i instead of 24p. The disk itself still has only the 24p video. A normal DVD player will honor those flags and play 60i. To that player it only sees 60i video and deals with it as 60i. A progressive scan DVD player hooked up to a progressive scan display will output the 24p frames.

I'm sure the same thing is happening with the F footage. When played from the camera it works like a DVD where the video is 24p with flags set that output the video as 60i through analog connections. For firewire transfer some decoders only know how to read the stream as 60i (like nornal DVD players) while the special decoders can remove the flags (like a progressive scan DVD player.)

same timeline no problem.

The progressive scan DVD player should not need to remove the flags because it's not outputting compressed mpeg, just fields or frames.

But more interesting are you comments about chroma sampling. You've said that chroma samples alternate for interlaced video, whether encoded as fields or progressive segmented frames. So the alternating is good for i but bad for p because the chroma is treated as interlaced.

How do you make that observation? Are you viewing with a hex editor or what other method do you use to observe the chroma samples in the mpeg file?

Also, you seem to be saying that 24F is essentially the best 24 progressively encoded format, because for the small penalty in resolution drop from the 2x2 chroma sampling, you avoid the artifacting caused by alternating chroma fields.

Extending that explanation to describe the current cams, I would infer that only 24F is a "true" progressive format, and that 24P from the HV20 and Sony V1 are the segmented frames carried within 60i. In other words, pulldown flags for 24F frames, but field pulldown to weave 1440x540 into full frames for 24P for HV20 and V1.

Your explanations do accurately describe what has been observed, artifact free but lower resolution 24F images, and artifacted but higher resolution 24P from HV20 and V1. If true it just seems contradictory that the marketing emphasis for 24P is the "true progressive" format with true progressive chips, yet the format with the progressive-from-interlaced chip trickery is the one outputting the progressively encoded frames.

Do you see it that way?

http://members.aol.com/ajaynejr/vidbug2.htm

Here is one link that talks about CUE (Chroma Upsampling Error)


http://en.wikipedia.org/wiki/Chroma_subsampling

Here is a link that shows the three forms of 4:2:0 color. Scroll down to the section on 4:2:0. Look at the images of the green text on the red background. You can see why progressive 4:2:0 is much better for keying. Progressive 4:2:0 is actually not all that bad for keying since it is very clean. The problem with 4:1:1 DV was that the blocks of chroma were 4x1 pixels in size. That means on edges you could end up with gaps or holes that were 4 pixels wide. Interlaced 4:2:0 also has a problem due to the alternated lines which can leave odd jagged edges. Progressive 4:2:0 really is a much better method for accurate chroma sampling with progressive images.

The JVC series of cameras also use progressive encoding so they are true 24p as well.

24F has really nothing to do with the detail in the image. The Canon series of cameras still do something with the interlaced chips to make a progressive image in the DSP. From there it goes through a whole new process where it is encoded as progressive frames. So in terms of encoded video yes it is true progressive. Again that really has nothing to do with detail. You could take a 32x32 pixel image and blow it up to 1920x1080 and encode it as a progressive sequence but it still wouldn't look as good as a raw 1920x1080 image encoded.

So yes the video from the HV20 is true 24p video in the DSP but then it gets encoded as 60i which uses the interlaced form of chroma sampling. It really isn't all that bad however. Progressive 4:2:0 is better of course but some decoders and some great tools do a really good job at swapping the chroma lines so they can upsample to 4:2:2 fairly well. With the HV20 there is also the bonus of capturing HDMI live into a 4:2:2 format such as Cineform, jpeg or uncompressed. When the video is 4:2:2 you no longer have to worry about 4:2:0 interlaced chroma patterns because every line has it's own chroma anyway. The HDMI output is still 24p inside of a 60i stream but once you remove the pulldown you will be left with perfect 4:2:2 24p frames. The higher end Canon cameras start with a much easier to use form of HDV for 24p but when dealing with uncompressed you still have to worry about removing the pulldown. The uncompressed output is also still from the magic performed on the interlaced chips so the small resolution loss is still there.

Very intersting! I wonder what will be the case with the PAL HV20 25p?

nope ...PSF is a differing animal(frame blend from ccds)
m streams are fun