How PsF video from the V1 is different than "p" or "F" video

[Bob Grant]

Originally Posted by Piotr Wozniacki

Bob, my ATI graphics card / Fujitsu-Siemens monitor combo enable me to playback (with VLC, from an NLE, or any software that uses video overlay in general) with *BOTH* the digital DVI and/or the analogue Component. There is no much difference between them, apart from the fact that the component HDTV output from the graphics card is more dynamic ("punchy") than the DVI (but of course they can be tuned to match in the ATI Catalyst control center).

So, when I say it's OK with deinterlaing off and awful with bobbing on, I mean both interfaces - the DVI and component.

UPDATE: After a second thought, I must add that the above is a bit more complicated, actually. The thing I'm sort of uneasy about is: why do I need to enforce bobbing in VLC to recreate the twitter even through component, which itself should be doing to the captured 25PsF clip exactly the same it's doing to the live V1E output when fed through it? It beats me...

Piotr,
first off, thanks so much for all your hard work on this.

I don't really have an answer as to why you have to use Bob on the component outputs to recreate the problem, could it be that without it the video card is sending frames?
As far as I know that cannot be done, a component connection only supports fields but I'm far from well educated on what component connections are capable of. However if that were the case it would explain a lot given that the camera itself can only output fields.

By the way, have you tried using the Median filter instead of GB in Vegas?

[Piotr Wozniacki]

Originally Posted by Bob Grant

By the way, have you tried using the Median filter instead of GB in Vegas?

Yes Bob I have - it offers finer control indeed, similar to the Gaussian Blur in Premiere, thanks for that hint!

[Steve Mullen]

Originally Posted by Piotr Wozniacki

I think I can say it safely now -- have helped us all establish the V1E/P cameras are not guilty of the "flaws" with their 25PfS video presentation. If I can recreate this flaw with a 1080-lines chart, it's clearly a display device problem (or more specifically: the way a display device is treating the 25PfS signal).

You are saying the line-twitter only appears when 25p is displayed and is not "in" the recorded video from a V1E. Correct?

Which means it should also be seen when 30p is displayed. But I've not seen it and there have been no reports of it from V1Us. Which means it can't be the PsF system.

Moreover, the video you sent me had more than twitter -- it had dancing dot aliasing noise on fine details. I don't have anywhere near this level of aliasing at 30p.

All this tells me there is must be something unique to the V1E.

My theory of aliasing from the A/Ds is wrong -- thank you Bob. But, I remain convinced that something to do with greater V rez in the E/P units is correct. And, I'm equally convinced it has to do with the need for E/P units to record 576-line rather than 480-line SD video. Because the SD signal is derived from the HD signal, it seems logical that the E/P units "have" greater HD V rez -- before the video is encoded and recorded.

Why? I'm not sure without understanding the insides of the V1 we'll ever know.

[Piotr Wozniacki]

Originally Posted by Steve Mullen

You are saying the line-twitter only appears when 25p is displayed and is not "in" the recorded video from a V1E. Correct?

Which means it should also be seen when 30p is displayed. But I've not seen it and there have been no reports of it from V1Us. Which means it can't be the PsF system.

Moreover, the video you sent me had more than twitter -- it had dancing dot aliasing noise on fine details. I don't have anywhere near this level of aliasing at 30p.

All this tells me there is must be something unique to the V1E.

My theory of aliasing from the A/Ds is wrong -- thank you Bob. But, I remain convinced that something to do with greater V rez in the E/P units is correct. And, I'm equally convinced it has to do with the need for E/P units to record 576-line rather than 480-line SD video. Because the SD signal is derived from the HD signal, it seems logical that the E/P units "have" greater HD V rez -- before the video is encoded and recorded.

Why? I'm not sure without understanding the insides of the V1 we'll ever know.

Steve, I tend to agree that the R50 V1 differs from the R60 model - your explanation of this difference may be accurate. I believe it is a quantitative rather than qualitative difference. I saw similar aliasing and mosquito noise in the 30p clips. I didn't see line twitter - but of course all 24p or 30p clips I ever saw were downloaded from the net and played back from the computer, and my own clips watched this way do not flicker, either.

Out of the methods of post-processing the 25PsF materials tried so far (selective Gaussian Blur in vertical direction in Vegas and Premiere, antialiasing filters etc) I found the v=2.5% GB in Premiere was most effective (99% of line twitter removed while the V-rez reduced only slightly and the H-rez untouched). I have tested even more solutions and now think an even better one is the Canopus Flicker Reduce filter; I'm uploading my "garage" clip with sharpness 7, encoded in Edius with this filter.

I will add the link to this clips in a while; please compare it with your results from FCP and share your opinion.

UPDATE: Here is the link to the abovementioned clip:
http://rapidshare.com/files/20590009...ker_reduce.m2t

Steve and Bob:
It's very important that you check it, as the Canopus Edius Anti Flicker filter effectiveness on eliminating twitter and flicker from the 25PsF video is astonishing! In fact, I have added it to my clip with Bob's 1080 lines chart - the result is an absolutely clean video, even with bobbing enabled in VLC! I wish I knew how this filter differs from all I tried previosuly in how it actually is doing the magic.

[Bob Grant]

Originally Posted by Steve Mullen

You are saying the line-twitter only appears when 25p is displayed and is not "in" the recorded video from a V1E. Correct?

Which means it should also be seen when 30p is displayed. But I've not seen it and there have been no reports of it from V1Us. Which means it can't be the PsF system.

Moreover, the video you sent me had more than twitter -- it had dancing dot aliasing noise on fine details. I don't have anywhere near this level of aliasing at 30p.

All this tells me there is must be something unique to the V1E.

My theory of aliasing from the A/Ds is wrong -- thank you Bob. But, I remain convinced that something to do with greater V rez in the E/P units is correct. And, I'm equally convinced it has to do with the need for E/P units to record 576-line rather than 480-line SD video. Because the SD signal is derived from the HD signal, it seems logical that the E/P units "have" greater HD V rez -- before the video is encoded and recorded.

Why? I'm not sure without understanding the insides of the V1 we'll ever know.

Are you certain the dancing dots are aliasing?
The amount of them is very dependant on light levels and they taper off away from the edges. This looks to me like what's left after the level sensitive DNR that Sony have said they're using.
What also makes me doubt it's aliasing is it's apparent on both vertical and horizontal edges. What confuses me though is they're quite different in nature to typical chroma noise, they're bigger than one pixel and they're only black.
I've had to push the camera hard to make them really noticeable though.

[Steve Mullen]

Originally Posted by Piotr Wozniacki

Steve, I tend to agree that the R50 V1 differs from the R60 model -- your explanation of this difference may be accurate. I believe it is a quantitative rather than qualitative difference. I saw similar aliasing and mosquito noise in the 30p clips. I didn't see line twitter --

1) The fact that your twitter is so bad and I don't have any -- tells me there is a real difference between units. It is only quantitative.

2) Now about making HD DVDs. I tried creating a real 24p file. The burning application would not accept the fame-rate. So, the only thing I can burn is 24p with 2-3 pulldown added!

3) Which raises the question, given how difficult it is to remove 2-3 pulldown from source video using FCP, is it worth the trouble just to have a 24p timeline?

A) One is supposed to not cut-to a judder frame. OK -- what happens if one does? (You can learn to spot the judder frames in the timeline so one can avoid them -- if it's really important.)

B) What happens when you edit and break the 2-3 cadence at each edit point? Since the cadence in the output file will not a perfect cadence -- your HDTV's deinterlacer will not be able to follow it. Which means 24p will be deinterlaced incorrectly. Does this matter? Does it matter when 80% of HDTVs can't correctly sense the 2-3 cadence even on a test signal?

I'm wondering if it's worth all these issues just to get 24fps motion judder. It's not like anything you shoot is going to fool anyone into thinking your video is film. And, it's not like very many people shooting with a $4K camcorder are going to transfer to film.

[Piotr Wozniacki]

Originally Posted by Bob Grant

Are you certain the dancing dots are aliasing?
The amount of them is very dependant on light levels and they taper off away from the edges. This looks to me like what's left after the level sensitive DNR that Sony have said they're using.
What also makes me doubt it's aliasing is it's apparent on both vertical and horizontal edges. What confuses me though is they're quite different in nature to typical chroma noise, they're bigger than one pixel and they're only black.
I've had to push the camera hard to make them really noticeable though.

I think the dancing dots, marching ants or mosquito noise (whatever you call it) is indeed aliasing. I'm encouraging you again to download the s=7 clip I encoded with Edius aniflicker filter; not only is the twitter gone but even with bobbing imposed and through component, the noise around the garage roof bottom edge and on the trunk is almost gone!

[Douglas Spotted Eagle]

Originally Posted by Steve Mullen

1)
I'm wondering if it's worth all these issues just to get 24fps motion judder. It's not like anything you shoot is going to fool anyone into thinking your video is film. And, it's not like very many people shooting with a $4K camcorder are going to transfer to film.

Around 20% of the films submitted to Sundance this year were shot on DVX100 or similar camcorders and 24p. Virtually all of the SlamDance entries, and same for the XDance entries were shot with sub 10K camcorders shooting 24p or resampled to 24p. There were several entries I viewed that were shot on the Canon XLH1.

I'm not a big fan of 24p for most things, but there is no denying the elephant in the room. Marcus van Bavel or one of the other transfer houses such as DFL could likely provide an extremely accurate number of how many sub 10K camcorders are used in the digital media submitted to their transfer houses.
There is at least enough work to keep them in business.

[Bob Grant]

Here's a link to a test clip I shot a few weeks ago:

http://www.yousendit.com/download/T2...Z1A1bmcwTVE9PQ

I think it pretty clearly shows what I've been talking about regarding noise.
Sorry it's a big file (147MB) and pretty boring stuff so only download it if you've got bandwidth to spare.
This is deliberately badly shot to try to 'provoke' the camera.
Make of it what you will.
And sorry it's on Yousendit who require you to sign up to download.

[Bob Grant]

Originally Posted by Piotr Wozniacki

I think the dancing dots, marching ants or mosquito noise (whatever you call it) is indeed aliasing. I'm encouraging you again to download the s=7 clip I encoded with Edius aniflicker filter; not only is the twitter gone but even with bobbing imposed and through component, the noise around the garage roof bottom edge and on the trunk is almost gone!

I've downloaded it and yes it looks very clean. I don't have a HD monitor here capable of displaying interlace but all VLC modes show it cleanly.

I did a little research and it would seem that you can send frames down a component connection. This might explain what was worrying you before. The camera is sending fields, your video card unless told to bob is sending frames.

[Steve Mullen]

post deleted for errors

[Steve Mullen]

Originally Posted by Bob Grant

I did a little research and it would seem that you can send frames down a component connection. This might explain what was worrying you before. The camera is sending fields, your video card unless told to bob is sending frames.

If you have 720p you always send frame sat up to 60p. And with HDMI you can send 1080 frames at up to 60p.

[Bob Grant]

Originally Posted by Steve Mullen

I was talking about the V1 specifically. I just got an email from someone who found out the hard way that Vegas will not write the V1's 24p back to the V1. Same as Piotr found with 25p.

And since I've found I can't burn a 24p or 25p HD DVD -- that leaves a camcorder with a near useless HD progressive mode. This is not the case for the DVX100 and JVC 720p24 camcorders.

One can explain Apple/Avid not supporting the V1 because they have never promised a time frame. But Sony not supporting Sony??? It sure can't be that Sony software couldn't get access to a prototype. And, it can't be that Sony software was unable to get the technical documentation from Sony so they could support 25p/30p PTT. And, it's not like it's rocket science to ADD 2-3 pulldown.

Sony is spending millions on marketing 24p and 25p, but without full support of the V1 by software, it's wasting its own money. It's crazy to tell a person go buy CineForm and Premiere Pro. But, it's equally crazy they can't output their HD producton to HDV. And, given who the client is -- it's really really bizzarre.

Sorry Steve but from what I've seen there's some glaring errors in what I assume you're saying.
I can convert 25PsF to 24p, frame for frame in Vegas. I can then insert pulldown and create a 60i m2t file, I posted a link to exactly this a few days ago. Only thing I haven't investigated is doing Advanced pulldown.
I can take the same thing back into a 24p timeline with Remove Pulldown enabled and get back the 24p. The only thing I come up short on is being able to set the P flag on the tape. This is hardly a new challenge as far as I know, DB, SP, DV and probably most other industry standard tape formats leave you in the same boat, including footage from the DVX100.

Same goes for 25PsF, OK, I've got to render it as 50i and print that to tape, so what? Same deal applies to every other tape format out there, from VHS to DB and HDCAM.

In the case of the camera it makes zero difference from what I can see, 25PsF or 50i on tape is going to come out the HDMI or component ports the same way. Realistically the flags aren't that much help, just saves you loging what it was shot in.

This is nowhere near the issues Canon's F mode creates, the only way to play the tapes is in a Canon camera. And once you get it on a T/L then what , you're going to have to PTT as 50i anyway unless you've got a CineAlta deck or a Canon camera on hand.
And then for broadcast, as far as I know there's no standard even being discussed for 1080p broadcasting and when there is it'll probably be 60p.

If you want to gripe about anything, well I for one think it'd be nice if Sony came out with a HDV deck that played it all, what's a post house to do when handed a 25F tape?

[Thomas Smet]

There is always 24p and 25p DVD. It is SD of course but even the sony site states that a great use of the V1 is to create true 24p DVD's. This has usually been the delivery format of those shooting 24p with the DVX100 or Canon XL2 camera.

As for not authoring a 24p HD-DVD that is the fault of the highly limited HD-DVD authoring software we have at this time. HD-DVD sadly only supports 1080 24p sitting inside of a 60i stream with flags sort of like how 24p DVD works. With proper authoring and encoding software we should all be able to at some point create a HD-DVD with the proper flags set. The limited specs I could find did say that 24p was supported for 720p but there was no 25p support. If 720p 24p doesn't work right now I would have to take a guess that it is a limitation of the current crop of HD-DVD authoring software.

If Vegas itself will not export export a 23.976 with pulldown file then the next best thing would be to export a true 23.976 file and use a stand alone encoder such as the Main Concept encoder to encode the mpeg2 file with the proper flags.

Since HD-DVD and Blu-Ray are still pretty new I'm sure most people dealing with 24p will continue to do it the way we have for the last few years which is to either deliver as a 24p DVD or deliver a image sequence to a film lab, both methods which are listed on the SONY website.


As for editing 24p without removing the pulldown, you have to be very carefull. If your project is 100% cuts only and you cut on the right frame then you should be ok. As soon as you add any type of rendered effects or animation those effects will get rendered as 60i while the video under will still be 24p inside of 60i. This is going to not only create a funky look with the effects having a different style of motion compared to the video but also really mess up any pulldown detection hardware.

[Steve Mullen]

Originally Posted by Bob Grant

I can convert 25PsF to 24p, frame for frame in Vegas. I can then insert pulldown and create a 60i m2t file, I posted a link to exactly this a few days ago. Only thing I haven't investigated is doing Advanced pulldown.
I can take the same thing back into a 24p timeline with Remove Pulldown enabled and get back the 24p. The only thing I come up short on is being able to set the P flag on the tape.

"Then I rendered to 1080 60i HDV but in the MainConcept mpeg encoder set the frame rate to 23.976 + 2-3 Pulldown and moved the quality slider to 15.

Vegas will render the file with a .m2t extension and although I haven't tried as yet you should be able to PTT this file."

Bob, did you try PTT? My friend said that the problem was the V1 wouldn't accept anyhing 24p. Perhaps, the camcorder does need certain flags set. Perhaps it thinks you have a ripped movie.

I'll delete my post as I think I have figured out a way to fool the HD DVD burning software to do 24. No point in doing this for 25p if the players can't play it.