Panasonic pushing motion portrayal [scanning backlights]

[Ahigh]

At the top of the HD lines, there’s the 55-inch TC-55AS680U. It has a native 240Hz refresh.

Of course this could be simple frame insertion. But who knows?

http://hdguru.com/panasonic-2013-tv-lineup-4k-and-lcd/

Continuing along,

The TC-60AS650U, TC-55AS650U, and TC-50AS650U are the next step down (above), and have “1500 Backlight Scanning” for improved motion resolution.

The 60-inch TC-60AS640U has “240 Backlight Blinking.” It’s bundled with a home theater system.

The AS530 Voice Guidance to the AS640’s Voice Assistant, but also has “240 Backlight Blinking.” There are four models: TC-60AS530U, TC-55AS530U, TC-50AS530U, and TC-39AS530U.

The only 2014 series that doesn’t have Life+ is the A400 series, which has three models: TC-50A400U, TC-39A400U, and TC-32A400U, which all have “120 Backlight Blinking,” which is presumably a 60 Hz panel with a flashing backlight.

Interestingly, many models are available now at Panasonic.com.

http://shop.panasonic.com/shop/viera-televisions-led

[Ahigh]

Last year, as I mentioned elsewhere on this forum, Panasonic was the only "big-boy" display manufacturer (IE: non PC gamer display) that I noticed demonstrating backlight strobing to eliminate retinal blurring. It was a very small little area and it seemed they weren't making it obvious what you were seeing, but I knew. It looks like they are trying to find a niche here. I'm just curious how effective they will be able to market this stuff if they are pushing in areas near and dear to our hearts here at Blur Busters.



They advertise a segmented set of LED backlights that strobe in sequence. Much of this has been out for a while, but I am curious where they are putting these efforts and how effective they are for games. I just haven't heard much back from users really like I have for other sets on this forum.

Strobing the backlight from top to bottom as the signal arrives reduces the latency compared to waiting to present image data at the top of the screen until data is also present at the bottom. So this is definitely a better way to strobe the backlight (in stages).

But I don't know if Panasonic is doing anything but frame insertion. I highly doubt they are even doing what Sony is doing with their MotionFlow Impulse that does not do frame insertion, just strobes the signal you give it.

We really need a big TV manufacturer to take 1080p120 and strobe at 120Hz. But I seriously doubt anyone is paying enough attention yet to anything except non-interactive displays (IE: television broadcasts).

[Chief Blur Buster]

Strobing the backlight from top to bottom as the signal arrives reduces the latency compared to waiting to present image data at the top of the screen until data is also present at the bottom. So this is definitely a better way to strobe the backlight (in stages).

Theoretically, scanning backlights are better, but in real life, they are actually not as good at motion blur elimination as strobe backlights. It can reduces lag if done properly, but doesn't reduce blur as much because of the backlight diffusion problem. This is covered in several places:

TFTCentral: http://www.tftcentral.co.uk/articles/co ... htm#theory
Blur Busters: http://www.blurbusters.com/faq/creating ... t/#problem

I have done some high speed camera tests, and this confirms the problem. The rest of the screen is dimly lit even as a backlight segment brightly illuminates part of the LCD. So this reduces the theoretical amount of motion blur elimination. Only pure-strobe backlights allow unbounded improvement (motion clarity limited only by strobe length). And strobe backlights are cheaper to engineer, provided the LCD is fast enough to finish refreshing before the next refresh cycle.

[Ahigh]

As usual, you are the man, Mark. I'm going to visit the Panasonic booth and ask them some questions about their backlight scanning methods after reading some of your research. Not that I expect much, but it would be great to find out if Panasonic is looking at these issues that you're pointing out.

[Chief Blur Buster]

As usual, you are the man, Mark. I'm going to visit the Panasonic booth and ask them some questions about their backlight scanning methods after reading some of your research. Not that I expect much, but it would be great to find out if Panasonic is looking at these issues that you're pointing out.

Very good scanning backlights, with very good diffusion control, are extremely hard to beat. But accomplishing that, AND low latency, AND the scanning complexity, is something extremely hard to do cheaply, and as I've found out via experimentation.

It's far easier to modify a 3D-ready LCD panel to do good-quality strobing, and the 3D-readiness (active shutter) of the panel means the panel is capable of sufficiently clean/finished refreshes for backlight strobing, by the nature of shutter glasses 3D requirements...

A really good scanning backlight can be made preferable to a strobe backlight, but it is apparently difficult to do a good game-friendly, low-latency ultra-low-persistence scanning backlight... So if Panasonic did a really good job, it would work well in Game Mode. Please ask Panasonic about scanning backlight modes in Game Mode. Bring your high speed camera if possible.

[Neo]

Backlight scanning for TVs has been around for a while. About the time "240 Hz" was starting to be the buzzterm it either meant actual 4x frame interpolation or "backlight blinking" "clear motion" etc. As usual, haters at the time merely brushed it off as a "fake" gimmick.

In the original Dolby Vision HDR papers the authors mention that the dual-modulation backlight also has potential to do low persistence scanning. With such spatial precision I would hope the current Dolby Vision system offers this for temporal precision. Perhaps even dynamic micro-scanning?!?! The paper: http://www.cs.ubc.ca/~heidrich/Papers/Siggraph.04.pdf

Chief, I found some technical papers at that LinkedIn group I noted previously. They might make good additions to your Science index. WARNING: BIG pdfs.

Video Processing for LCD-TVs Frank van Heesch
http://alexandria.tue.nl/extra2/692181.pdf (131 MB)

Flat Panel Display Signal Processing Analysis and Algorithms for Improved Static and Dynamic Resolution Michiel A. Klompenhouwer
http://alexandria.tue.nl/extra2/200612229.pdf (20 MB)

Been lurking at the LI group and some interesting thoughts on technical issues from a broadcast angle:
http://www.linkedin.com/groups/Balancin ... mp_5048880
http://www.linkedin.com/groups/Tech-Com ... mp_5048880

[Chief Blur Buster]

Backlight scanning for TVs has been around for a while. About the time "240 Hz" was starting to be the buzzterm it either meant actual 4x frame interpolation or "backlight blinking" "clear motion" etc. As usual, haters at the time merely brushed it off as a "fake" gimmick.

Early scanning backlights were extremely inefficient, so they couldn't meet the 4x improvement. And they often combined it with motion interpolation (see Existing HDTV Scanning Backlights), which added input lag. And weren't suitable for video games.

Today, full-strobe backlights are able to reach darn near theoretical limitations of motion-blur-reduction -- strobe length is the exact persistence and is directly proportional to motion blur. For LCD's that are able to refresh fast enough (typically stereoscopic 3D-capable LCDs), strobe backlights are far simpler and more effective than scanning backlights, and you don't have backlight diffusion (light leakage between adjacent scanning backlight segments) that interferes with theoretical motion blur elimination.

Blur Busters actually started in 2012 as http://www.scanningbacklight.com and my old Scanning Backlight FAQ still exists. This before LightBoost got discovered. The relevant forum thread about Blur Busters History is currently in Area 51

Newer scanning backlights are vastly better, and are more suitable for slower-responding LCD's than strobe backlights, however, strobe backlights are the future because it is the only way to permit unbounded motion clarity improvement (e.g. it's now scientifically possible to have an LCD with less motion blur than a short-persistence CRT -- it is a simply a light output issue).

[Neo]

Yep, the "240 Hz" scanning was added to a 120 Hz 2x interpolated display hence the flames about being "fake 240 Hz" displays.

[Neo]

Ohhhh I just had an idea regarding scanning. Need a new post for it.

What if a frame was actually segmented into multiple bands and each was rendered as a slice of a unique frame and flashed as a type of sub-frame? If there are 10 bands a complete frame would be rendered like a shutter roll (if viewed combined). The gpu would render 10% of a frame then start over at the next band. Like a controlled tearing artifact. It's basically like how the old-timey vacuum tube video cameras would capture and then render on crts. Has this been tried yet? Any artifacts? Hmmm...

[Chief Blur Buster]

Ohhhh I just had an idea regarding scanning. Need a new post for it.

I can move this post (and my reply) into a separate thread quite easily; just need a thread for it. If you post a new thread, I can move this reply over upon a button click.

What if a frame was actually segmented into multiple bands and each was rendered as a slice of a unique frame and flashed as a type of sub-frame? If there are 10 bands a complete frame would be rendered like a shutter roll (if viewed combined). The gpu would render 10% of a frame then start over at the next band. Like a controlled tearing artifact. It's basically like how the old-timey vacuum tube video cameras would capture and then render on crts. Has this been tried yet? Any artifacts? Hmmm...

I am trying to interpret what you mean, so are you talking about GPU side or display side?

GPU segmented rendering

They already do this in some SLI modes. It's called Split Frame Rendering. Different GPU's render different segments of the screen. This complicates a lot of things, some parts of the frame can be more complex than others, and that means a GPU may be behind others. Likewise, advanced effects might do memory-accesses on other parts of the screen. Alternate Frame Rendering is more common because it's simpler and generates bigger frame rate gains, I believe.

Display side segmented scanning

Multi-scanning has been done for a long time.
A lot of jumbotrons use this technique, for example.
Dual-scan LCD's have existed for a long time.

The problem is stationary tearlines dividing the segments, because of the offset timebase of the scanned bottom edges of segments, versus scanned top edges of segments. It is a problem that also showed up on the Sony Crystal LED prototype (which used multiscanning, probably in an attempt to solve the light-output problem of passive-matrix LED displays).

If you must do multi-scanning, divide the display in vertical strips instead. That way, the scanning is all synchronous, and you don't get the stationary tearlines occuring at the mutliscan boundaries.