3D PinP problems

[Qframe]

I’m at a loss too!

I have the latest Edius 4.5c and a fairly new NX and we have both tested the same files. I’m trying to think of a non common factor here.

I know of one other person (in Australia) who has the same problem but could offer no answers.

I suppose I will have to find a work-around… a bit of a pain though.

[7hbg]

Drew

I was experimenting with my own files and I can now say that I can reproduce your problem 100% of the time and I can even tell when it will appear:

It is not the graphic file you use for 3DPiP that is at fault, but the contrast-details of the underling video that determines the degree of ‘distortion’ on the edges of the graphic. Imagine you have a clear sheet of glass in front of you with some text on it, and you are spinning it. As you spin it, the edges of the glass will distort the background. If the background has a lot of sharp details you will see the distortion very clearly, if the background is smooth or is monotone the distortion is impossible to see. This is the analog equivalent of what is happening here in Edius. It is still fault of 3DPiP, but you can avoid it by placing your graphic over smooth or monotone areas.

Here is a sample project that demonstrates what I just described: 3DPiP5.zip

Play the first sequence in a loop. You will see the distortions of the background grid where one would expect the edges of the graphic.
Next, play the second sequence in a loop (no sharp details there) and you will not see any distortions.

Yes it is a fault of 3DPiP, but as far as I can tell it is distorting background video beyond the graphic and the degree it is visible is dictated by the amount of contrast-details at the point of the graphic’s edge.
(I hope this helps developers in Kobe…)

[Qframe]

Thanks Rob!

Finally someone else can see it too!

Yes it is dependant on the Luminance changes in the background image. I was having trouble explaining this in my earlier posts.

I hope they can fix this in future updates because as TV commercial producer the effects we use on a day to basis all rely on the 3DPiniP filter for DVEs. This “noise” is unacceptable for broadcast applications.

[7hbg]

Drew

I was experimenting with my own files and I can now say that I can reproduce your problem 100% of the time and I can even tell when it will appear:

It is not the graphic file you use for 3DPiP that is at fault, but the contrast-details of the underling video that determines the degree of ‘distortion’ on the edges of the graphic. Imagine you have a clear sheet of glass in front of you with some text on it, and you are spinning it. As you spin it, the edges of the glass will distort the background. If the background has a lot of sharp details you will see the distortion very clearly, if the background is smooth or is monotone the distortion is impossible to see. This is the analog equivalent of what is happening here in Edius. It is still fault of 3DPiP, but you can avoid it by placing your graphic over smooth or monotone areas.

Here is a sample project that demonstrates what I just described: 3DPiP5.zip

Play the first sequence in a loop. You will see the distortions of the background grid where one would expect the edges of the graphic.
Next, play the second sequence in a loop (no sharp details there) and you will not see any distortions.

Yes it is a fault of 3DPiP, but as far as I can tell it is distorting background video beyond the graphic and the degree it is visible is dictated by the amount of contrast-details at the point of the graphic’s edge.
(I hope this helps developers in Kobe…)

Brandon (GrassValley_BH)

Can we consider this issue ‘reported’ or do we need a new thread in the ‘User-Reported Issues’?

[Khoi Pham]

As a producer of TV commercials I’m constantly giving the 3D PinP (Edius 2,3,4) filter a work out due to the lack of a DVE in Edius. I mainly use text produced in Photoshop or Quick Titler (both with alpha) and move them in all directions.

When doing this there is a visible thin line that appears where the outside edge of the moving graphic is.

Following numerous complaints from clients I discovered I could eliminate the line by turning on “soft borders”. But this creates a new problem. Another soft line, the width of the soft parameter, appears on the edge of the graphic. It is only faint but is very noticeable over some backgrounds.

Does anyone know how to get rid of this?

I had the same problem, I get rid of this by going into the layout tool and just move it a little bit to the right or left and that line just dissapeared.

[GrassValley_BH]

Brandon (GrassValley_BH)

Can we consider this issue ‘reported’ or do we need a new thread in the ‘User-Reported Issues’?

Not to sound lazy here, but you're "closest" to the issue, so I think you'd do a much better job reporting it that I would. Please submit it into the User-Reported Issues section. Thanks!

[7hbg]

Not to sound lazy here, but you're "closest" to the issue, so I think you'd do a much better job reporting it that I would. Please submit it into the User-Reported Issues section. Thanks!

So… it has been a week now since I have submitted a ‘User-Reported Issue’ on this subject.
Does it really take over a week to approve a bug report?

[GrassValley_BH]

So… it has been a week now since I have submitted a ‘User-Reported Issue’ on this subject.
Does it really take over a week to approve a bug report?

I've been really really busy so unless Kenneally or one of the other mods get to it, yes, it can...

[jdroadie]

Mike, I hate to be a bother but this 3d PIP thing is in fact a showstopper for us choosing edius...this saddens me....is there any word of said rewrite? are you talking the whole app or just the 3d pip? Maybe 4.7? if its the whole app, then I guess we'll have to go back to premiere, what a pity....if they are tired of apologizing maybe sooner? I can't believe GV can put out a product called "broadcast" with the weak DVE....I am amazed the user base has put up with it so long....

[GrassValley_KH]

3D PiP is expected to be revisited for the next major version, as it is part of the Vista-compliancy mandate.

You're quite welcome to use another solution that fits your needs - each to their own.

[Belphegor]

I'm happy to hear 3D PiP will be revisited in future.

I remind to everybody that 3D PiP is a keyer and not a Filter

This is the story behind it.

3D Pip was developed some years ago during the good old DVStorm days when competitors crafted better and improved editing solutions regarding 3D NLE effects.

3D PiP was initially written as a plugin to be used in Premiere enabling DVStorm card to do what Canopus missed from Rex card era: the ability to map video in a 3D space.
DVStorm and Rex cards have not any 3D chip onboard, like actual NX cards, differently from other competitors cards.

So, a solution able to map video in 3D space had to be crafted only in software, and this had to be powerful and smart enough to deliver Realtime (remember that we were in Pentium 4 2,8 Ghz era) and not to force editors to perform rendering.
So, this software solution forced engineers to write something that had not to be a Transition between two clips but a Filter to be applied to each clip allowing editors to compose a multilayered stack granted of 3D features.
Unfortunately, something went wrong, probably due to Alpha handling issues in DVStorm card under Premiere or in DVStorm card in itself.
This happened while Canopus engineers were also developing their own NLE called Edius and then migrated 3D PiP in it (marketing Edius and pumping it with this new feature).
So, to overcome to DVStorm limitations, 3D PiP was implemented in Edius as a Keyer and not as a Filter.
The Keyer, in the same way of Chroma/Luma Keyers, has some limitations in DV arena due to compression of DV aspect ratio: there are some visible thin black/grey lines that DV spectrum does not allow to key properly.

This is a limitation of DV video and it's a general issue.
Look cloesly how other NLE perform a Chromakey: you'll see the same thin black/grey line in the boundaries of the keyed track.
The truth is that every NLE Keyer delivers artifacts like the one reported in this first post.

I hope you don't blame me if I talked about this.
Blame should be due to Canopus/Grass Valley that in these years did do anything to improve 3D PiP keyer to solve the quirks and to empower it allowing us to get rectangular or oval PiPs with heights and lenght independently.

This is stoty mates...

[GrassValley_KH]

Well, the story is a little different...but it's all rather moot at this point.

As Brandon has mentioned, the 3D Composer effects found within Xplode Pro 4.x feature far more flexibility in terms of shapes, and indeed, quality.

[Khoi Pham]

Yep Dave I did all that.

But have a look at this JPG attached. It is a still captured from the timeline.

I guess you missed my post, but go into the Layout tool and move it a few pixel to the left or right and that line will be gone.

[jdroadie]

"I remind to everybody that 3D PiP is a keyer and not a Filter"

This is absolutely correct for said DVE, I mean 3d pip...In the switcher world, which GV knows as good as anyone, this feature is accomplished on the keyer channels....The thin line does come out, I can do PSD's with glow and shadow(nice job keeping layer effects)using PS presets....no line as long as boder/shadow is off, you can even use lights...thing is PIP keyer is missing only one thing: the ability to map its shadows and borders to the actual alpha of what is in it...instead those are mapped to the entire video frame(alpha?)....it seems kind of simple like a a bug fix you'd get in a 4.6x fix...of course if you want it to be backward compatible to that hardware mentioned it may need a checkbox,

I know the developers are busy with the format of the week, and Windoze Vista development, ....just a major bummer....everything else in Edius is easily worked around.

I am only a trial user here (yeah I know a lot of ragging) but Edius is a pretty nice package for the money. My customers like lots of stuff moving and it seems most productive to do this in the NLE when possible. Looking forward to the Edius at NAB.

[GrassValley_BH]

We'll get there... just not today. Thanks for the feedback - both good and bad is good for us in the long run.