.: The X9 Project :. (Released) — Page 6

Moth3r said:

I remember seeing a similar chroma lag before; it was caused by a colourspace conversion (YV12 to YUY2) before IVTC. However, in that case the chroma was combed and not like in those screenshots.

It does look like the result of a noise reduction filter.

Interesting, had no idea, there really are a multitude of things that can go wrong in the process if you're not careful. The example frame I posted was indeed IVTC'd but no colorspace conversion took place, the X9-video are being stored in YUY2 in this case, it's unfortunately in the source and I agree that it looks like a result of noise reduction, combined with the bleed and shift of the chroma it becomes pretty nasty looking sometimes. Unfortunate as an X9 cap of these LD's are truly a great gift otherwise.

Arnie.d, thanks for replying, weird... what capture cards are generally recommended, I guess the prices varies pretty much between cards, any good sites or threads for these sort of things?

msycamore: Is the chroma shift consistent? That is, is it always displaced by the same number of pixels?

Also, is it present in X9 TESB and ROTJ as well?

AntcuFaalb said:

msycamore: Is the chroma shift consistent? That is, is it always displaced by the same number of pixels?

From the way it was described as a chroma smear rather than a chroma shift, I'd guess that the displacement depends on how fast the object is moving.

Moth3r said:

AntcuFaalb said:

msycamore: Is the chroma shift consistent? That is, is it always displaced by the same number of pixels?

From the way it was described as a chroma smear rather than a chroma shift, I'd guess that the displacement depends on how fast the object is moving.

Ah, that makes sense!

Could this be caused by the comb filter not adapting (2D->3D and vice-versa) quickly enough?

Does Dark/Sega JSC exhibit chroma smearing as well? No, right?

Dark/Sega has pulldown frames stored in YV12 space, so after IVTC you'll find some combed chroma like Moth3r described.
That's why I chose selectEvery(10) in my IVTC scripts for those
instead of using selectEvery(5) a.k.a. pulldown().
That way the telecine 3's can avoid the problem, however half of the 2's are still affected no matter what.
So every fourth frame in the progressive stream will be suspect.

Darth Mallwalker said:

Dark/Sega has pulldown frames stored in YV12 space, so after IVTC you'll find some combed chroma like Moth3r described.
That's why I chose selectEvery(10) in my IVTC scripts for those
instead of using selectEvery(5) a.k.a. pulldown().
That way the telecine 3's can avoid the problem, however half of the 2's are still affected no matter what.
So every fourth frame in the progressive stream will be suspect.

Thanks for the information, Darth Mallwalker!

Do you know if DJ captured in YV12?

AntcuFaalb said:

msycamore: Is the chroma shift consistent? That is, is it always displaced by the same number of pixels?

Don't know, haven't studied it. But there is indeed both a smear and a shift present. Some of the chroma shift and bleed is there in the actual SW LD but not all of it.

AntcuFaalb said:

Also, is it present in X9 TESB and ROTJ as well?

ESB had the same artifacts seen in SW, so I assume Jedi is affected by the same thing, but you'll have to check to be sure.

AntcuFaalb said:

Does Dark/Sega JSC exhibit chroma smearing as well? No, right?

No idea, don't own it.

OK, using Dark/Sega's chroma with X9's luma works wonderfully.

Observations:

1. X9's letterbox is 8 pixels higher than Dark/Sega's.
2. X9 is exactly 7 fields ahead of Dark/Sega.
3. Either Dark/Sega looks too dark or X9 looks too bright.

(Note: I've only played with side #1, so these observations may not apply to the whole video.)

The following is a sample image of X9's luma + Dark/Sega's chroma. I'm too tired to post more than one, but I'll be sure to tomorrow if anyone's interested...

The colors are still shifted.

Or whatever you want to call it. You can see how the color information is further to the right than it should be. I tried going into Photoshop once and shifting it to the left, but it only made it look kind-of better.

Asaki said:

The colors are still shifted.

Or whatever you want to call it. You can see how the color information is further to the right than it should be. I tried going into Photoshop once and shifting it to the left, but it only made it look kind-of better.

 

Really? I'm not seeing it...

Can you post something from that image to demonstrate the chroma shift?

Also, note that the colors in X9 are smeared, so changing them to being simply shifted would be an improvement!

Guys, I may have messed something up when dumping the first Dark/Sega VOB to AVI. I'm going to look into it to see if I did.

Here's the frame from Dark/Sega. Is the chroma shifted in it as well?

For the sake of comparison, here's the frame from X9.

AntcuFaalb said:

Can you post something from that image to demonstrate the chroma shift?

 Well it's not a very good shot for pointing it out...

I don't think I still have the X9, or I'd find you a better example. I know someone put up a shot of the binary sunset that showed it off very well, along with a second shot where they tried to fix it.

I'm sorry to say it guys, but I think the Dark Jedi raw (DJ sent me a sample) capture looks better than the X9 raw capture.

NOTE: SEGAFLIP HAD NOTHING TO DO WITH THIS CAPTURE, SO PLEASE DISREGARD THE FOLLOWING CAPTION, AS IT SHOULD READ: "Dark Jedi Raw".

You_Too said:

You're right. After layering them on each other in photoshop and comparing, there is no doubt that the X9 has the cleanest luma.

I'd be more than willing to spend my time doing this, but I'd definitely need the raw, uncropped, 29.97fps captures from Dark_Jedi.

Why?

• raw => duh!
• uncropped => so that we don't lose resolution until the final stage(s)
• 29.97fps => so that I can apply a software comb filter to get rid of the rainbowing

Software comb filters always need to be applied before IVTC'ing.

^I doubt those still exist. DJ was never fond of them, and the set is years old. I can't even find the thread for the original release! He did attempt a V2 but claimed it was a fruitless effort and deleted it shortly after (which was, of course, his call.)

bkev said:

^I doubt those still exist. DJ was never fond of them, and the set is years old. I can't even find the thread for the original release! He did attempt a V2 but claimed it was a fruitless effort and deleted it shortly after (which was, of course, his call.)

Believe it or not, I've been in contact with Dark_Jedi, and that sample I posted above is from his V2!

He sent me a raw Lagarith-encoded AVI sample from his JSC V2 a few days ago.