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

[Mikko Lopponen]

Originally Posted by Piotr Wozniacki

EDIT: I played the clip with VLC and like Steve says: the Bob deinterlacer re-introduces heavy horizontal lines twitter, which is present when playing back from camera. This means it definitely (and not just probably) is a display-device related problem.

The bob-deinterlacer in vlc seems to be a bit worse than the bob method in for example powerdvd. It seems to exaggerate twitter.

[Piotr Wozniacki]

Originally Posted by Mikko Lopponen

The bob-deinterlacer in vlc seems to be a bit worse than the bob method in for example powerdvd. It seems to exaggerate twitter.

Yes it does, but what's important it clearly shows that what is so dreadful when feeding 25p into component of my monitor (either live, or from tape with the V1) is just the monitor trying to "deinterlace the deinterlaced", and thus it's NOT the camera's fault.

[Piotr Wozniacki]

Originally Posted by Steve Mullen

Thank you. Using VLC to play via HDMI to my hdtv I see three things.

1) on P: aliasing on thin horizontal lines that appears as "dancing dot noise.

I'm not sure if we're talking the same thing; by noise I mean the dancing noise that is equally visible in my 25p as well as 30p (or was it 24p?) V1 versions (eg. in the "church", "shoeshine" or "village" clips posted earlier on this forum).

Bottom line so far is:

- the line twitter is NOT the camera fault (monitor trying to deinterlace with bob or similar technique)

- the dancing noise is common to both E and U versions of the V1.

I think we can safely say that both E and U versions of the V1 give the same results, also in progressive mode - one can like it or not, but no need to differentiate between them any longer.

[Steve Mullen]

UPDATE on my tests with v1u:

30p and 24p: Aliasing in the form of "dancing dots" (looks like noise) on horizontal lines/edges and on VERY thin diagonals. 99% eliminated at 5. Becomes very visble at 11.

Actual resolution seems the same on tree limbs and wood grain textures in all modes. I'm not sure why my posted pix and others looked softer when the video doesn't. I fear I may have switched to Cine (at 5) AND switched 24/30 on. If so, my bad.


Can you send me a 3 sample and a 5 sample to add to my HD DVD.

[Piotr Wozniacki]

Steve, do you mean the same scene but with sharpness at 3 and at 5? OK I'll do that, but have to wait for a better weather (it's now snowing in Poland).

Do you know about some tool for trimming HDV m2t files without a need for recompression? I used to have some shareware doing just that but only for SD video files...

Generally, do we agree that the only problem now is this "dancing dots" noise, and that it's the same in both the E and U versions of V1's progressive?

If so, we've just made an important step forward in understanding some myths about this camera.

[Steve Mullen]

Originally Posted by Piotr Wozniacki

I'm not sure if we're talking the same thing; by noise I mean the dancing noise that is equally visible in my 25p as well as 30p (or was it 24p?) V1 versions (eg. in the "church", "shoeshine" or "village" clips posted earlier on this forum).

Bottom line so far is:
I think we can safely say that both E and U versions of the V1 give the same results, also in progressive mode - one can like it or not, but no need to differentiate between them any longer.

The amount of aliasing, however, seems very different. And, after seeing your line twitter -- I'm not so sure there is not a real difference between the units.

I have finally come up with an explanation of why R50 models would have more aliasing than R60. It's really simple, but I want to think a bit more. If I'm correct -- it confirms Sony recommendation to use 3 with p. But, it also confirms the downside of doing so. It even explains why Sony created the first firmware for r50.

No matter the explanation -- the work you are doing is critical because it points to what r50 users can do.

[Thomas Smet]

Steve,

How come if I render video from 3D studio Max as 1440x1080 30p anamorphic and then encode that to HDV which in turn really makes it 30psf it doesn't show that many problems with aliasing? I mean a 3D render with have a lot more detail then any HD camera could do. I even encoded two different segments. One as a raw 60i and the other with flags set. I'm not sure if the flags made it to the tape on my HC1 that I used for the test but in the end it was still 1080i video that was progressive in nature even though it was encoded as 60i. I even added film grain to simulate noise and everything seemd to look pretty good even on my HDTV that just bobs.

[Steve Mullen]

A good 3D app will have the option to apply a anti-aliasing filter. Perhaps it turned on.

I think we've moved past the thread topic and really now looking at R50 vs R60. It's looking like -- contrary to my thinking -- that the R50 has more aliasing & line twitter than the R60. At the same time -- it appears that the type of deinterlace makes a difference during playback.

Is this related PsF? I really don't know.

[Piotr Wozniacki]

Steve, I can't wait for the explanation, though - please provide us with it!

In the meantime, I've uploaded another file showing more or less the same scene with:
-50i, sharpness 7
-25p, sharpness 7
-25p, sharpness 5
-25p, sharpness 3
Here is the link:
http://rapidshare.com/files/17910475...arpness7-3.m2v

From my viewpoint, with sharpness set to 3 (as Prime Support suggests for the PAL users), the dancing dots (or aliasing) is almost eliminated. Also, I have shot a couple of quite nice clips with this setting and in some conditions, the video is not too soft (combined with the cinegamma and other settings, but also with the particular contents, you can get quite "filmic" look to it). However, when you watch the scene I provided with the link right after the same scene shot with higher sharpness, it looks absolutely too soft. I have to add though it's still on par with the default XH sharpness.

I'm looking forward to what others think of it.

[Mikko Lopponen]

Originally Posted by Piotr Wozniacki

Yes it does, but what's important it clearly shows that what is so dreadful when feeding 25p into component of my monitor (either live, or from tape with the V1) is just the monitor trying to "deinterlace the deinterlaced", and thus it's NOT the camera's fault.

Maybe that's because the camera outputs 50i at its components not 25p?

[Steve Mullen]

Originally Posted by Piotr Wozniacki

Also, I have shot a couple of quite nice clips with this setting and in some conditions, the video is not too soft (combined with the cinegamma and other settings, but also with the particular contents, you can get quite "filmic" look to it). I have to add though it's still on par with the default XH sharpness.

I think you've found your answer -- since your goal is a filmic look you are altering the video in several ways.

Even in the USA, Sony recommends a softer look for Cine. And, given the BBC's claim that switching to 24p is not enough because video is still too sharp -- going softer seems to be part of what folks do, or should do, when using P video for a film look.

If I'm correct that at "3" R50 units have the same V rez as the R60 units at "5" -- the only loss is H. rez. Since V rez is the most critical, using "3" is no big deal.

The critical question -- is how will you view your productions. If using a computer, then you can even use "5." If going back to HDV, then "3" is correct. Maybe "4" would be an OK compromise.

[Piotr Wozniacki]

Not that I like the idea of having to turn down the sharpness on a HDV camera in order to obtain a watchable video, but I can add that under certain conditions, it is possible to keep sharpness low (eg at 3) to avoid aliasing, but still get a good picture. One thing to observe is the aperture; one should try never to close it too much or diffraction can add to low sharpness, further spoiling the picture. The same scene I posted, still with sharpness at 3 but with ND filter engaged and iris open up from around 8 to around 5.6, can be found here:

http://rapidshare.com/files/18138889...rpness3_ND.m2v

The perceived sharpness/resolution goes up even though the sharpness setting was untouched.

[Piotr Wozniacki]

Originally Posted by Steve Mullen

I think you've found your answer -- since your goal is a filmic look you are altering the video in several ways.

Even in the USA, Sony recommends a softer look for Cine. And, given the BBC's claim that switching to 24p is not enough because video is still too sharp -- going softer seems to be part of what folks do, or should do, when using P video for a film look.

If I'm correct that at "3" R50 units have the same V rez as the R60 units at "5" -- the only loss is H. rez. Since V rez is the most critical, using "3" is no big deal.

The critical question -- is how will you view your productions. If using a computer, then you can even use "5." If going back to HDV, then "3" is correct. Maybe "4" would be an OK compromise.

Steve, if its true that at "3" R50 units have the same V rez as the R60 units at "5" , and indeed some 20% more bandwidth must be squeezed out of the R50 hardware than from R60, it seems still possible than Sony might come with a new firmware fix for PAL models. It'd be unough to filter vertical resolution by some 20%, while keeping the horizontal rez and default sharpness intact, or keep the V rez intact while cranking up the H rez to its absolute limits, and then establish the default sharpness of 7 being the same as the current 5 setting in the 25p mode. The purpose of the above being that lowering sharpness to avoid dancing dots would only result in V rez reduction, keeping the H rez on par with the interlaced mode. Canon XH-series have separate "H DTL FREQ" and "DTL HV BAL" settings for just that. What do you think?

[Tony Tremble]

Originally Posted by Piotr Wozniacki

Steve, I can't wait for the explanation, though - please provide us with it!

In the meantime, I've uploaded another file showing more or less the same scene with:
-50i, sharpness 7
-25p, sharpness 7
-25p, sharpness 5
-25p, sharpness 3
Here is the link:
http://rapidshare.com/files/17910475...arpness7-3.m2v

From my viewpoint, with sharpness set to 3 (as Prime Support suggests for the PAL users), the dancing dots (or aliasing) is almost eliminated. Also, I have shot a couple of quite nice clips with this setting and in some conditions, the video is not too soft (combined with the cinegamma and other settings, but also with the particular contents, you can get quite "filmic" look to it). However, when you watch the scene I provided with the link right after the same scene shot with higher sharpness, it looks absolutely too soft. I have to add though it's still on par with the default XH sharpness.

I'm looking forward to what others think of it.

Hi Piotr

Are you saying the XH-A1 image is as soft as the V1E at sharpness setting 3?

That would be an extraordinary claim. Perhaps you could clarify that remark?

I have downloaded and viewwed your footage to see if there is any difference between a "fixed" V1 to one shipping directly from the factory with upgraded firmware. Well no. The image still turns to mush at sharpness setting 3 in line with Sony's recommended progressive setting.

Regards

TT

[Piotr Wozniacki]

Tony,

Have you also downloaded the clip showing clearly how opening aperture up can sharpen the picture, even at the sharpness setting at 3? For you convenience, here is the link: http://rapidshare.com/files/18138889...rpness3_ND.m2v

All I want so say is that by carefully tuning the parameters, it is posiible to obtain a very nice picture with sharpness=3, completely free of the dancing dots (the horizontal, contrasty line twitter have already been established beyond doubt as a viewing device-induced artefact).

As to the Canon - yes, unless you crank its sharpness up (to +3, or even higer), it looks much softer than the V1E at the default 7 setting, which leaves the room to decrease the V1's sharpness while still being on par with the A1, at least with the V rez of its progressive mode (unfortunately, de-sharpening affetcs also the H rez - hence my "wish" to Sony in the previous message). Of course, the best compromise could be setting sharpness at around 5, which is completely viable in certain conditions, where the contents doesn't have too many contrasty lines and thus the aliasing artefacts are not so strong.