Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

The Issue of Ghosting

Phil Bates

Phil Bates

Well-known member
Joined
Apr 4, 2007
Messages
685
Reaction score
0
Points
0
Website
www.artbeats.com
I don't know about you, but ghosting is the biggest obstacle I am facing in creating quality 3D footage. I can control just about everything else; convergence, sync, matching geometry, matching color, and I can shoot with the precision necessary to avoid the typical pitfalls. But then, when I display the footage on a Samsung or Panasonic VT25 monitor the ghosting can literally ruin the shot. I've tried the new Panasonic replacement glasses and the XPAND generic glasses and the problem stays consistent.

I know one of the basic rules to shooting 3D is to avoid high contrast edges, AND if you can't avoid them, try to engineer them to sit near the screen plane. But, I am finding this rule to be so restricting, that even mundane landscape images can have high contrasting edges in the wrong places, especially if you like shooting at magic hour.

I am relatively new to 3D and many of you on this forum have knowledge that amazes me, so I have some questions maybe someone can answer.

1. Is ghosting as big a problem to you as it seems to be for me? What are your workarounds and solutions?

2. Where does the ghosting come from? Is it a phosphor lag in the monitor, lack of precision shutter timing or is it that the glasses are not opaque enough?

3. I've noticed in the theater that with Dolby 3D ghosting doesn't seem to be a problem, or at least it is so minimal it does not distract. So, it seems that high end display systems have figured this problem out. If so, should 3D content/stock footage be judged on how much ghosting it shows on a plasma monitor?

4. Do you see this as a temporary problem that future generations of TVs will fix? Is that time coming soon?

Thanks for your consideration,

Phil
 
I will be releasing a 3D test disc later this year. We include objective test patterns to measure the amount of ghosting. Hopefully this will help display manufacturers improve. To give you some examples:

1. Sony VPL-VW90ES (Front Projector) has .25% ghosting in R, G, and B. This is really low. This is the best I have measured.
2. JVC RS40/50 has up to 2% ghosting in at least one channel. I don't have my notes with me so I don't recall.
3. Panasonic VT20. 0% ghosting in blue, 1.4% ghosting in red, and 2% ghosting in green.
4. LG 3D panel has 13% ghosting in one of the channels. One of my patterns looks the same w/ or w/o glasses. This is as bad as I have seen. As a 3D display it is pure crap. As a 2D display, its not so bad. Impressive backlighting.
5. Sharp Panel has 2.75% in blue, 3% in red, and 8% in green. This is also pretty bad. Samsung and Sony are also poor.

Pixar and Dreamworks both use ghost busting algorithms on their content. Most of this works by looking for high contrast edges and altering them. I am not sure if Universal has the same technology, or if it is not as good. Dispicable Me has some pretty strong ghosting in it.

Have also measured Monster, Xpands, and JVC glasses. In one test, the Xpands worked better than the JVCs, but this may be screen dependent. Both glasses are made by XPands, but one has the polarizer rotated 90 degrees. I have not really seen much in the way of ghosting differences. Most has been with light output.
 
Stacey Spears said:
I will be releasing a 3D test disc later this year. We include objective test patterns to measure the amount of ghosting. Hopefully this will help display manufacturers improve. To give you some examples:

1. Sony VPL-VW90ES (Front Projector) has .25% ghosting in R, G, and B. This is really low. This is the best I have measured.
2. JVC RS40/50 has up to 2% ghosting in at least one channel. I don't have my notes with me so I don't recall.
3. Panasonic VT20. 0% ghosting in blue, 1.4% ghosting in red, and 2% ghosting in green.
4. LG 3D panel has 13% ghosting in one of the channels. One of my patterns looks the same w/ or w/o glasses. This is as bad as I have seen. As a 3D display it is pure crap. As a 2D display, its not so bad. Impressive backlighting.
5. Sharp Panel has 2.75% in blue, 3% in red, and 8% in green. This is also pretty bad. Samsung and Sony are also poor.

Pixar and Dreamworks both use ghost busting algorithms on their content. Most of this works by looking for high contrast edges and altering them. I am not sure if Universal has the same technology, or if it is not as good. Dispicable Me has some pretty strong ghosting in it.

Have also measured Monster, Xpands, and JVC glasses. In one test, the Xpands worked better than the JVCs, but this may be screen dependent. Both glasses are made by XPands, but one has the polarizer rotated 90 degrees. I have not really seen much in the way of ghosting differences. Most has been with light output.
Click to expand...
Wow, Stacy. Thanks for that information. I don't suppose you've tested the Panasonic VT25? I've read that that particular model has the least ghosting of any plasma, but would be interested to know if your test bears that out.

Thanks again,
Phil
 
I don't suppose you've tested the Panasonic VT25?
Click to expand...

There are two difference between the VT20 and VT25. One is the bezel and the other is thje serial port. Performance wise, they are the same. My number 3 includes the VT20 data. The GT series has true 709 color gamut, which the VT20/25 does not. The VT35 is coming soon. Not sure what may have changed.
 
Stacey Spears said:
The VT35 is coming soon. Not sure what may have changed.
Click to expand...

Deeper blacks
Faster phosphor, reduced ghosting
New lighter, more closed 3D glasses
CalMAN > Ethernet > ISF Calibration
 
anti-Ghosting

anti-Ghosting

I added anti-Ghosting to my 3D formatting commands years ago.

It works well for high contrast things like wire frame with shutter glasses, less well with rendered images, but does help somewhat.

There are some steps to do it:

1) Reduce the contrast of both images so that you can get "blacker than black" and "whiter than white" after the mask is combined with the image.

2) Make a mask from the R image for the L image, and from the L image for the R image.

3) Adjust the gamma of the masks so that the highlights are more pronounced maybe.

4) Subtract the mask from the image.

The result looks like a double image when you look at just the R or L with a negative ghost on it. When you view in Stereo 3D with the shutter or other glasses the mask cuts some of the normal ghost. For some images it seems to work well, but with a loss of overall contrast depending on how heavy of a mask you apply, in some cases if the Ghosting is very bad you might need 30% mask so that reduces the range of tones, since the mask will not work if you clip at all.

The best solution is to reduce the ghosting in the first place, mechanical shutter glasses and mechanical shutters on the projectors would do that, but people want electronic or passive things, and most of those "leak" to some degree so you will see Ghosts since your eyes are very sensitive to small brightness values, they are not linear, so a 1% leak will look more like a 5 to 10% leak if its in the highlights.

One reason its hard to get the mask right is that the iris in the eye and the eye in general change depending on where in the projected image you are looking and your state of mind etc. as well as how long you look at the image and if the last shot was brighter or darker, so your eye is not a stable element in the system, therefore there is no way to predict the exact mask needed for all viewers at all times... And with some passive pola glasses when you tilt your head a bit the amount of ghosting changes as well...

To adjust the mask percentage, view through the 3D glasses, or off the glassesless display in 3D stereo mode, then cover one eye and adjust the mask until the ghosting "nulls" at a point where you don't see the negative ghost much, you can vary the gamma of the mask to help in finding the right "null" values...

I did not experement with optimizing the values for each eye to take into account unbalanced specular reflections, nor did I have options to "animate" the control values for the mask frame to frame, which might help...
 
Dan Hudgins said:
I added anti-Ghosting to my 3D formatting commands years ago.

...

4) Subtract the mask from the image.
Click to expand...
Dan, I am having trouble subtracting the mask. I am using AE and there is no Subtract transfer mode. "Difference" doesn't seem to work. I also tried inverting the mask and applying the Multiply mode, but that tends to darken the highlights. Can you be more specific on how this would work in AE?

Thanks,
P
 
is it possible for you to post a still frame (left / right) from a scene you believe has a lot of ghosting?

besides the ghosting caused by display equipment, and high contrast aside, two things that can also cause ghosting

1) Excessive positive or negative parallax, combined with contrast (I'm not suggesting that your shooting with too much parallax, but it's worth taking a look into)
2) Having too much of a "gap" between elements in the foreground and the background. I.e nothing of reference in the mid ground... i'd call it occlusion ghosts.

Regards.
 
You must log in or register to reply here.
Back
Top