New Sample Footage

[Alex Leith]

Originally Posted by Steve Mullen

...if there isn't a consensus -- then there isn't any problem with the V1U.

Possibly not true, given that it's generally felt that the "effect" is intermittent (and depends on the image shot).

We certainly have something weird going on in the footage and images provided by Brett.

I also think it's images 1 and 3 that are P, but I wouldn't stake my life on it - and I'll agree with you that in this case it's hard to tell.

[Steve Mullen]

Originally Posted by Philip Williams

Perhaps you could look at the M2T file and two images of the red workshop that Brett posted for us and let us know if you think those are fine.

Looking at the JPEGs I see exactly what I've always said about 30P vs 60I. The P has about 2 clicks less Sharpness than does I. Look at the grass. Look at the siding. (By the way -- I really don't believe these were shot with +9. Why would anyone shoot at anything above NORMAL.)

I posted this observation almost 3 weeks ago -- if not longer. It's obvious! You can see the lack of detail in my posts and Bretts. Look at my pix A B C. Look at the wood grain! It's obvious!

And, while the P softness is not always there -- likely because if there is a lot of contrast it increases apparent sharpness -- I'd hardly call it "intermittent." I'd say it's almost consistently slightly soft.

And, I'm not surprised since the capture method for P is different from that for I. Every I frame is sourced from 2 frames of 1920x1080 while P is sourced from 1 frame of 1920x1080. I'm not sure how many times I have to explain this. I would expect this difference. No one should expect I and P to be identical.

The solution is simple -- try a bit more Sharpness with P than you use for I. If that adds to much EE, then the answer is equally simple. Just accept the fact that the way Sony gets P results in less fine detail that the way it gets I. Once accepted -- you either live with it or buy a different camcorder.

[Philip Williams]

Originally Posted by Steve Mullen

Looking at the JPEGs I see exactly what I've always said about 30P vs 60I. The P has about 2 clicks less Sharpness than does I. Look at the grass. Look at the siding.
<snip>

Well this is getting even weirder. So I'm looking at the two screen caps with the red work shed. I'm really trying to work with this reduced sharpness theory but I'm actually noticing now that the tree branches, especially in the right side of the frame, look quite a bit sharper and crisper in the progressive frame. There are actually some very fine branches that wash out in interlace mode that are clearly visible in the progressive frame. The increased resolution and detail in some parts of the progressive frame just does not align with the theory that progressive mode has reduced sharpness over interlaced. I think it stands to reason that whatever the sharpness settings are for progressive vs interlaced, it should be even across the entire image. I'm very clearly seeing now that the red siding on the shed - especially the front with the single door - is extremely blurred out in progressive and the details in the branches are actually increased.

I think I'm going to pull this clip into After Effects tomorrow and take a closer look.

Thanks for responding Steve.

[Stephen van Vuuren]

Originally Posted by Philip Williams

Well this is getting even weirder. So I'm looking at the two screen caps with the red work shed. I'm really trying to work with this reduced sharpness theory but I'm actually noticing now that the tree branches, especially in the right side of the frame, look quite a bit sharper and crisper in the progressive frame. There are actually some very fine branches that wash out in interlace mode that are clearly visible in the progressive frame.

Good eye. That effect is visible on LCD as well. I think the reduced sharpness is unlikely. I played with the clip in Vegas 7 using levels and channel blend. The red shed lines that appear to soften in progressive are actually the more underexposed portion of frame. They contain much more blue and red, much green.

Most of the branches appear to stay same or sharpen slighty in progressive whereas the shed detail disappears.

My guess is that it's noise reduction (or possible a gamma/matrix issue) that is causing somewhat underexposed (lower IRE) areas (seems to be more blue and red channel affected) areas to get excessive noise reduction applied, basically treated detail as noise and smoothing it out.

However, areas of the frame that it identifies correctly as detail, and we see the slight progressive sharpness improvement we expect.

This jibes from my observations of Tony's images - areas of low detail in darker areas of frame lose detail and/or watercolor effect.

Of course, I'm not a camera engineer, but I think with these clues, some good charts, a vectorscope and controlled manipulations of camera settings, this issue (which is clearly real) can be illuminated.

I can see why testing might have missed it - my guess is well lit, higher contrast progressive images look fabulous as DSE's footage did.

[Tim Le]

Originally Posted by Philip Williams

Well this is getting even weirder. So I'm looking at the two screen caps with the red work shed. I'm really trying to work with this reduced sharpness theory but I'm actually noticing now that the tree branches, especially in the right side of the frame, look quite a bit sharper and crisper in the progressive frame.

Ha! I noticed the exact same thing! The progressive frame even seems to have more sharpening artifacts in certain areas of the frame. I made these 100% crops to illustrate Philip's point.

In the m2t, I also definitely see the dancing noise in the progressive versus interlaced.

I hate to say it, but for me, this camera is just too inconsistent. I really wanted to buy this camera for the smaller size and better focusing and exposure aids but Sony isn't doing a good job of convincing me, especially when the competing Canon XH-A1's street price has now settled to $3500. I really feel for the V1E folks who have been inconvenienced by this whole mess and I don't blame them at all if they vote with their pocketbooks and get an XH-A1 instead.

P.S. Another thing Sony, the focus distance display turning off after 3 sec is also another really bonehead thing to do!

[Bob Grant]

I'd suggest that increasing sharpness in a attempt to compensate might make things worse. Edge enhancement seems to always do a good job of enhancing noise and if this 'effect' is the result of DNR then edge enhancement is going to make the noise worse and the DNR bite harder. Perhaps this is why Sony were saying to turn detail down?

[Stephen van Vuuren]

Originally Posted by Bob Grant

I'd suggest that increasing sharpness in a attempt to compensate might make things worse. Edge enhancement seems to always do a good job of enhancing noise and if this 'effect' is the result of DNR then edge enhancement is going to make the noise worse and the DNR bite harder. Perhaps this is why Sony were saying to turn detail down?

I agree. If my and others admittedly sketch DNR theory is correct, sharpening would make it worse. What would be needed is a fix to the NR DSP firmware and/or ability to adjust/turn off NR

[Steve Mullen]

Originally Posted by Philip Williams

Well this is getting even weirder. So I'm looking at the two screen caps with the red work shed. I'm really trying to work with this reduced sharpness theory but I'm actually noticing now that the tree branches, especially in the right side of the frame, look quite a bit sharper and crisper in the progressive frame.

I too noticed this, but I feel these HIGH-CONTRAST lines are exactly the ones where edge enhancement would be applied to increase their apparent sharpness.

In other words, overall detail can be soft -- the siding, grass, the wood railing -- while at the same time EE is applied to edges that meet certain criteria. And, at the same time noise reduction -- perhaps only chroma noise reduction -- is applied to areas of the same color. Multiple alterations to the video occur at the same time. And, each alteration is dynamic. It is turned on and off as needed.

This is what DSP does to "enhance" the image. This is how Sony got Hi8 to look so clean compared to S-VHS. This is nothing new in camcorder design.

--------------

I agree that increasing Sharpness might not be the right solution because -- as I think about it -- it might not help the grass and siding at all. It all depends on exactly what increasing sharpness does. I'm now thinking it only affects EE and not detail. (Which means there may be no way to avoid the fact P looks softer than I.)

Thus, if Brett's video was shot at 9 -- the branches might look better if 7 had been used -- while the siding might look the same. Which raises the possibility that Sony defined 5 for Cine
because it reduces EE while not really reducing detail.

It would be nice to shoot a complex scene at different levels of Sharpness to determine what the optimal level is for I and for P.

[Tony Tremble]

Originally Posted by Bob Grant

I'd suggest that increasing sharpness in a attempt to compensate might make things worse. Edge enhancement seems to always do a good job of enhancing noise and if this 'effect' is the result of DNR then edge enhancement is going to make the noise worse and the DNR bite harder. Perhaps this is why Sony were saying to turn detail down?

Again spot on the money Bob. We 25Pers can see the effect of sharpening because our default setting of 7 is more like the 24Per's 11. That's why we are being told to turn down sharpness to 3 to bring it more in line with 7 on the V1u.

The consensus is people really can tell the difference between P and I with the exception of Steve it seems.

P is not necessarily softer that I only when certain conditions are met. In other cases I is sharper than P. It is a mixture throughout the frame. This is why in Brett's clip the trees are alive with noise and the side of the shed is obliterated. This get worse as the light drops.

TT

[Tony Tremble]

Originally Posted by Stephen van Vuuren

I agree. If my and others admittedly sketch DNR theory is correct, sharpening would make it worse. What would be needed is a fix to the NR DSP firmware and/or ability to adjust/turn off NR

You are definitely on the right track. You have to think of a reason why there is more digital noise and therefore an increased involvement from the DNR circuit particularly in darker areas of the frame.

Work that out and you have the answer to why P has the processed look.

The reason is prosaic. So prosaic that it is being overlooked.

TT

[Pasty Jackson]

So isn't it more than likely that the encoder is just trying to compensate for too much detail? It would be extremly awesome if someone could post a series of labeled shots, all of the same subject, at every detail setting starting with the lowest. That would hopefully shed some light on what exactly the encoding engine is doing. Anybody up for the task?!

-pasty

[Alex Leith]

Interesting... let's hope we're on to something here...

[Steve Mullen]

Originally Posted by Pasty Jackson

So isn't it more than likely that the encoder is just trying to compensate for too much detail?]

I don't think the encoder is involved in this.

And, frankly it really doesn't matter WHY P is softer than I. All of us can see that it is. I reported this and posted pix weeks ago. There's nothing really new here.

Now given this fact, when Sony defined the Cine setting they had a choice:

1) increase EE and hope that fact that "the objects to which it is applied" will have increased apparent sharpness which will overcome the overall softness; or

2) decrease EE to keep it in balance with the overal image softness.

Obviously, given that 90% of those shooting Cine mode are after a film-look, they made the right choice -- decrease EE to 5. This is the right amount of EE for P as is 7 for I.

In fact, given that every "how to get a film look" begins with "set 24p" AND "reduce Sharpness/Detail" -- the fact P looks softer is exactly what most folks want in Cine mode. The whole point of Cine mode is to avoid the hyper video look. So there is no real problem with softer P. Even for shooting 30p web video -- softer compresses better.

Now obviously there will be scenes where the EE balance isn't ideal. But an audience doesn't look at tree branches. They don't look at the siding on a house. They look at faces. They follow a story.

Once you understand the role of EE (Sharpness) and understand that it needs to be different for I and P because I and P are inherently different in image detail -- you are home free with a V1U. Use Sony's settings as I have and you get great HD.

Of course, you are free to increase or decrease EE as needed. But, if you increase it too much you are going to make mess. And, if you reduce it too much, you are going to make mush.

As Chris said many posts back -- the point of a tool is to use it be creative. Now that we understand the V1U better -- we know what settings to use. Why not shoot some PEOPLE doing somthing interesting? Enough of the tree branches. This all reminds me of people shooting water with the Z1 and worrying about the compression artifacts they saw. That kind of "testing" was big news for a week -- now few remember the horror stories. You can always find situations that break digital cameras.

[Mikko Lopponen]

Originally Posted by Steve Mullen

Look for the oil paint FX. Look for the ants. Look for the macroblocks with no detail. Look for the ringing. :)

You deinterlaced those pictures. That means effective resolution is dropped by half even in the progressive pictures. Previous pics were not deinterlaced, but they had no motion so the interlaced pics also contained a lot of information and no visible interlacing.

Atleast the 1,2,3,4 pictures, the a,b,c shots look fine. Compared to the a,b,c shots the four first ones are visibly lower in resolution.

I'd have to guess A is interlaced. B and C are not.

[Alex Leith]

Originally Posted by Steve Mullen

And, frankly it really doesn't matter WHY P is softer than I. All of us can see that it is. I reported this and posted pix weeks ago. There's nothing really new here.

Although to be fair Steve, you did several times say that there wasn't a "problem". Now that we're closer to working out what the cause of the "problem" might be you're saying there is a problem and you noticed it first (which you did - but not as a "problem".)

But people don't want softer in cine mode... they want less edge enhancement.

What this camera is giving is excessive palette knifing under certain circumstances. Just because it is intermittant doesn't make it acceptable. It is unpredictable, and that makes it very dangerous to anyone who HAS to get the shots they want without the time, budget or opportunity to reshoot.

No-one is here to bash the camera. We're trying to work out what's going on, and under what circumstances it displays the excess softening. That way those who own the camera can have more certainty of not finding their footage looking like mush when they get back home.

We don't YET understand whether this IS actually the cause of the problem (and it is a "problem" until we understand it fully), because no-one has yet done a comparison, adjusting sharpness across P and I.