New Sample Footage

Alex Leith · January 24th, 2007, 04:47 PM

Quote:

Originally Posted by Steve Mullen

This does sound reasonable too. But, why would P be cored more than I?

Um... Oh yeah! That would seem to be a flaw in that theory... especially as if it was just the coring setting it would have been an easy fix for V1E owners - especially given that the image looks so great in "I".

Tony Tremble · January 24th, 2007, 05:04 PM

It is quite amazing that now people are beginning to see the problems with progressive mode even on the V1U. The artefacts are the same as I reported for 25P except for some reason they are slightly reduced.

I am amazed that you didn't spot these Steve.

I have a pretty good understanding as to the problem and I know it has nothing to do with software decoders at all. I am not going to post what it is because the post will just get deleted and I'll get accused of spreading FUD. Which I think you will all find out that I haven't in due course.

You might feel that there isn't too much of a problem looking at those images posted by Brett but wait until the lighting goes against you and you'll find another problem surface. I'll refer you to my post with comparison pics of roof detail on the 25P issue.

Whether you can live with the oil paint effect is your call but here in PAL land where the issue is magnified for whatever reason I can't.

TT

Piotr Wozniacki · January 24th, 2007, 05:05 PM

Now I am shocked: the prog_int_test.m2t file shows the kind of softness in the progressive 30p mode _EXACTLY_ like that in my V1E 25p mode!

I can't believe my eyes! Tony, can you see it?

What is Sony thinking? Why have they stopped marketing the V1E as a true progressive camcorder and still trying to sell the V1U as a fully functional version, if _BOTH_ suffer from the same flaw?

I'm compeletely lost. But you know what? I've got a feeling they'll soon come up with a software fix to both versions. It can be done, they just need more time.

Edit:
After a second thought, I can see some logic to it. So far, most NTSC reports about how great the progressive mode is were based on 24p, which is different in its nature than the 30p (which is in relation to 60i the same as 25p to 50i). Tony, both of us suspected quite a time ago that perhaps both 25p and 30p might be affected. And voila - they are (with the effect slightly less pronounced in the 30p). Sony is still selling the NTSC version as progressive simply because it has 24p, which works. I bet they are working on both 25p and 30p; whether they will succeed? Well, let's keep fingers crossed.

Ken Ross · January 24th, 2007, 05:57 PM

Wow, I just don't see much of a difference in Brett's jpegs. Yes, there are some differences, but they're very minor IMO. I think Tony's shots of the roof were more dramatic and may have pointed out the lighting issue in this problem.

Piotr Wozniacki · January 24th, 2007, 06:08 PM

Ken, just watch the prog_int_test.m2t clip closely - once it switches from interlaced to progressive (I assume 30p), the roof details get softened (and the brown walls even more so). There's no doubt about it; I tested my V1E in exactly the same manner: same scene, several seconds in 50i then several scenes in 25p; the same kind of difference in sharpness, only just a little more pronounced.

Alex Leith · January 25th, 2007, 02:40 AM

In fact you can see the paint effect in the other clips that Brett posted, too.

It seems mostly to be limited to areas of similar colour - like skytones, or flat walls.

In the villiage clip it's there in the sky, clouds, distant mountainside, and shadow areas of the branches - although less noticable because they don't have much variation in them naturally, so the "coring" or "noise reduction" or "hideous MPG artefacts" (or whatever is causing this phenomenon) isn't quite so obvious - but it's definately there. And there is dancing noise all over edges.

Whether this is "terribly" noticable or not - it doesn't seem to be an acceptable image when compared to the "I" footage. P seems to totally wipe out vast swathes of detail.

Tony Tremble · January 25th, 2007, 03:15 AM

Alex

Whether the oil paint effect is noticeable is down to the scene on a case by case basis but the dancing noise is there all the time. The problem this causes is reducing the efficiency of the next encoding pass be that back HDV, compression to DVD or for broadcast etc etc. You'll be storing up problems for distribution with all that dancing noise.

It's not just how it looks it is the effect on the whole post production chain that must be taken into consideration.

TT

Brett Sherman · January 25th, 2007, 04:24 AM

Quote:

Originally Posted by Piotr Wozniacki

After a second thought, I can see some logic to it. So far, most NTSC reports about how great the progressive mode is were based on 24p, which is different in its nature than the 30p (which is in relation to 60i the same as 25p to 50i). Tony, both of us suspected quite a time ago that perhaps both 25p and 30p might be affected. And voila - they are (with the effect slightly less pronounced in the 30p). Sony is still selling the NTSC version as progressive simply because it has 24p, which works. I bet they are working on both 25p and 30p; whether they will succeed? Well, let's keep fingers crossed.

Just based on my informal tests I didn't notice any difference between 24P and 30P, they both exhibited the same effect. I don't have examples to show. But I'm pretty confident that 24P is no better.

Brett Sherman · January 25th, 2007, 04:28 AM

The question for me is what Sony is going to do about it. It seems like they have a problem of the same magnitude as the split screen with the first JVC HD 100's. They'll either have to fix the problem with some sort of firmware update or get a working model and exchange it for the bad ones. I don't have time today to call them, but I will sometime soon. It would be nice if Sony acknowledged the problem and made a commitment to fix it. This is going to be bad PR for them.

Mikko Lopponen · January 25th, 2007, 05:14 AM

Quote:

Originally Posted by Tony Tremble

The problem this causes is reducing the efficiency of the next encoding pass be that back HDV, compression to DVD or for broadcast etc etc. You'll be storing up problems for distribution with all that dancing noise.

The dancing noise is grain that the smoothing algorithm hasn't removed. But it does remove a lot of other noise so this actually will improve the encoding. It's easier to predict movements of smoother surfaces. The noise around edges would be there anyway even without the filter.

Tony Tremble · January 25th, 2007, 05:40 AM

Quote:

Originally Posted by Mikko Lopponen

The dancing noise is grain that the smoothing algorithm hasn't removed. But it does remove a lot of other noise so this actually will improve the encoding. It's easier to predict movements of smoother surfaces. The noise around edges would be there anyway even without the filter.

Mikko

Look closer there is much more noise in progressive than interlaced. I have loads of footage of the phenomenon and can clearly see an increase in noise in areas where it is relatively noise free in interlaced. This noise creates a shimmering effect in areas of high contrast detail and makes detail like winter trees against a blue sky extremely problematic. There is much more ringing in these circumstances.

The issues are much more than just coring believe me. When the light drops on a subject an over-active noise filter sets about destroying even more detail. This does not happen to the same degree in interlaced mode as it does in progressive.

Coring can be an effective too to improve the efficiency of the compression scheme but in this case it is not the only problem. The coring threshold is quite steep so you can literally have clumps of high detail "pop" out of a smoothed area.

You will also find edge detail is more aliased in progressive mode too due to over sharpening.

TT

Ken Ross · January 25th, 2007, 06:09 AM

Tony, it's funny that you mention noise in winter trees against a sky. This is precisely where I find the noise to be the worst in the interlaced mode on the FX7. Edge enhancement noise can be clearly seen along the edges of bare branches against the sky. As I've said before, I see the same issue with wires against a sky.

Now if I shoot the same scene with my little Canon HV10, there is no noise in that same area. Yes, the Sony will appear brighter & a bit sharper, but with the added phenomena of edge noise. Again, this does not appear in most scenes, but it's clearly there in some.

So IMO some artifacts are not a byproduct of just the progressive mode.

Thomas Smet · January 25th, 2007, 07:32 AM

To me so far it almost seems to react strange to subtle gradients. That may be why we notice this in areas of flat color such as a blue sky or wall. Maybe there is a lot of banding from the DSP. Just a thought.

Stephen van Vuuren · January 25th, 2007, 11:33 AM

Brett:

Can you shoot a 24p vs 30p vs 60i clip so we can be sure that is the issue? Say of the workshop? I'm a little uneasy that some cameras may have the issue in 24p.

Ken Ross · January 25th, 2007, 06:42 PM

Quote:

Originally Posted by Thomas Smet

To me so far it almost seems to react strange to subtle gradients. That may be why we notice this in areas of flat color such as a blue sky or wall. Maybe there is a lot of banding from the DSP. Just a thought.

Tom, make sure what you're seeing is in no way related to your display device. Many digital displays will produce precisely this type of banding with areas of similar color.

January 24th, 2007, 05:05 PM	#48
Piotr Wozniacki Inner Circle Join Date: Aug 2006 Location: Poland Posts: 4,086	Now I am shocked: the prog_int_test.m2t file shows the kind of softness in the progressive 30p mode _EXACTLY_ like that in my V1E 25p mode! I can't believe my eyes! Tony, can you see it? What is Sony thinking? Why have they stopped marketing the V1E as a true progressive camcorder and still trying to sell the V1U as a fully functional version, if _BOTH_ suffer from the same flaw? I'm compeletely lost. But you know what? I've got a feeling they'll soon come up with a software fix to both versions. It can be done, they just need more time. Edit: After a second thought, I can see some logic to it. So far, most NTSC reports about how great the progressive mode is were based on 24p, which is different in its nature than the 30p (which is in relation to 60i the same as 25p to 50i). Tony, both of us suspected quite a time ago that perhaps both 25p and 30p might be affected. And voila - they are (with the effect slightly less pronounced in the 30p). Sony is still selling the NTSC version as progressive simply because it has 24p, which works. I bet they are working on both 25p and 30p; whether they will succeed? Well, let's keep fingers crossed. Last edited by Piotr Wozniacki; January 24th, 2007 at 07:04 PM.

January 24th, 2007, 06:08 PM	#50
Piotr Wozniacki Inner Circle Join Date: Aug 2006 Location: Poland Posts: 4,086	Ken, just watch the prog_int_test.m2t clip closely - once it switches from interlaced to progressive (I assume 30p), the roof details get softened (and the brown walls even more so). There's no doubt about it; I tested my V1E in exactly the same manner: same scene, several seconds in 50i then several scenes in 25p; the same kind of difference in sharpness, only just a little more pronounced. Last edited by Piotr Wozniacki; January 25th, 2007 at 01:49 AM.

January 25th, 2007, 02:40 AM	#51
Alex Leith Major Player Join Date: Mar 2005 Location: Northampton, England Posts: 500	In fact you can see the paint effect in the other clips that Brett posted, too. It seems mostly to be limited to areas of similar colour - like skytones, or flat walls. In the villiage clip it's there in the sky, clouds, distant mountainside, and shadow areas of the branches - although less noticable because they don't have much variation in them naturally, so the "coring" or "noise reduction" or "hideous MPG artefacts" (or whatever is causing this phenomenon) isn't quite so obvious - but it's definately there. And there is dancing noise all over edges. Whether this is "terribly" noticable or not - it doesn't seem to be an acceptable image when compared to the "I" footage. P seems to totally wipe out vast swathes of detail. __________________ Alex

January 25th, 2007, 11:33 AM	#59
Stephen van Vuuren Space Hipster Join Date: Oct 2001 Location: Greensboro, NC Posts: 1,508	24p vs 30p Brett: Can you shoot a 24p vs 30p vs 60i clip so we can be sure that is the issue? Say of the workshop? I'm a little uneasy that some cameras may have the issue in 24p. __________________ stephen v2 www.insaturnsrings.com

January 24th, 2007, 05:04 PM	#47
Tony Tremble Regular Crew Join Date: Oct 2006 Location: United Kingdom Posts: 451	It is quite amazing that now people are beginning to see the problems with progressive mode even on the V1U. The artefacts are the same as I reported for 25P except for some reason they are slightly reduced. I am amazed that you didn't spot these Steve. I have a pretty good understanding as to the problem and I know it has nothing to do with software decoders at all. I am not going to post what it is because the post will just get deleted and I'll get accused of spreading FUD. Which I think you will all find out that I haven't in due course. You might feel that there isn't too much of a problem looking at those images posted by Brett but wait until the lighting goes against you and you'll find another problem surface. I'll refer you to my post with comparison pics of roof detail on the 25P issue. Whether you can live with the oil paint effect is your call but here in PAL land where the issue is magnified for whatever reason I can't. TT

January 24th, 2007, 05:57 PM	#49
Ken Ross Trustee Join Date: Jan 2005 Location: New York Posts: 1,945	Wow, I just don't see much of a difference in Brett's jpegs. Yes, there are some differences, but they're very minor IMO. I think Tony's shots of the roof were more dramatic and may have pointed out the lighting issue in this problem.

January 25th, 2007, 03:15 AM	#52
Tony Tremble Regular Crew Join Date: Oct 2006 Location: United Kingdom Posts: 451	Alex Whether the oil paint effect is noticeable is down to the scene on a case by case basis but the dancing noise is there all the time. The problem this causes is reducing the efficiency of the next encoding pass be that back HDV, compression to DVD or for broadcast etc etc. You'll be storing up problems for distribution with all that dancing noise. It's not just how it looks it is the effect on the whole post production chain that must be taken into consideration. TT

January 25th, 2007, 04:28 AM	#54
Brett Sherman Major Player Join Date: Mar 2006 Location: Mount Rainier, MD Posts: 428	The question for me is what Sony is going to do about it. It seems like they have a problem of the same magnitude as the split screen with the first JVC HD 100's. They'll either have to fix the problem with some sort of firmware update or get a working model and exchange it for the bad ones. I don't have time today to call them, but I will sometime soon. It would be nice if Sony acknowledged the problem and made a commitment to fix it. This is going to be bad PR for them.

January 25th, 2007, 06:09 AM	#57
Ken Ross Trustee Join Date: Jan 2005 Location: New York Posts: 1,945	Tony, it's funny that you mention noise in winter trees against a sky. This is precisely where I find the noise to be the worst in the interlaced mode on the FX7. Edge enhancement noise can be clearly seen along the edges of bare branches against the sky. As I've said before, I see the same issue with wires against a sky. Now if I shoot the same scene with my little Canon HV10, there is no noise in that same area. Yes, the Sony will appear brighter & a bit sharper, but with the added phenomena of edge noise. Again, this does not appear in most scenes, but it's clearly there in some. So IMO some artifacts are not a byproduct of just the progressive mode.

January 25th, 2007, 07:32 AM	#58
Thomas Smet Trustee Join Date: Mar 2004 Location: Milwaukee, WI Posts: 1,719	To me so far it almost seems to react strange to subtle gradients. That may be why we notice this in areas of flat color such as a blue sky or wall. Maybe there is a lot of banding from the DSP. Just a thought.