Hi!

Been reading about the new Epsons and find it good that the e-shift is at lower pricelevels this year. JVC got some (price)competition at least. We´ll se some shootouts soon.


Quote from JVC: "E-shift 4 Technology is capable of inputting full-spec 4K video signals including 4K60P 4:4:4 thanks to the optimised engine and the newly adopted high power lamp. These improvements result in an even higher definition picture with more precision"

This is my question: How the heck is the e-shift handling the 4k signal? As a skeptic consumer I would like to know how. Since the chip is working with 1080 I assume the 4k signal i scaled down to Full HD (1080). That means loosing 75% of the pixel information. Or is it "handeling" all the 4kpixels and making an average 1 pixel of the four (4K).
If so. Imagine 4 pixels (4k) with different colour variations and even HDR information.... shrinked into one (average) pixel that will be e-shifted 0,5 pixels...

Is that how they "handeling 4k"? If, so it wont be any different compared to bluray. (Except some smudged HDR info)

Can anyone please tell me if there´s more to the "handeling 4k-signals" in the e-shift technology??

Regards!

The JVC website describes it pretty well. It takes the 4k image and splits it into two - 2k images. It shifts one of the images 1/2 pixle up and 1/2 pixle over and flashes them at double speed. There is no loss of information, but the shifting does soften the image slightly. It will never beat a true 4k projector in terms of image detail, but it looks darn good.

Actually there is a 50% loss of information as the JVC and Epson projectors display 4M unique pixels while full resolution 4K/UHD has 8 Mpixel. There is NO down conversion of a 4K/UHD image to 1080p with these pixel shifting projectors. Rather only half the pixels in the original 4K/UHD image are needed to create the pixel shifted image. The results fall somewhere in between the 2M pixel 1080p image and the full 8M pixel UHD image.

Check out this thread and see for yourself a fairly good indication of the level of quality the e-Shift JVCs can display along with a native 4K Sony projector when fed UHD blurays and also supersampled 1080p Blurays upscaled to 4k and fed to both projectors via madVR.

The quality difference is incredibly close.

Shootout - JVC RS500 (X7000) & Sony 320ES

https://r.tapatalk.com/shareLink?url...3&share_type=t

The quality difference is incredibly close........and that's comparing stills.....with moving images the difference would be minuscule if any!


That Javs dude and Bandyka have done a great job.....wish they shot some 4k video as well though!

Thanks!

Alas I dont have access to a good enough 4k capable DSLR. Video would be pretty uselss through. Photos are going to be higher quality, taking video of a projector is a compromise.

Thanks for math lesson (feeling a little dumb now); 2k x 1k x 2 panels = 4 million pixle vs 2k x 4k = 8 million pixle. Brilliant.

Actually, no. There is no information lost. Half the displayed information is simply shifted by the dwell -- time difference -- between the two displayed "frames". If our visual cortex could not tell the difference between a 2K image and the e-shifted 4K image then you could make an argument that "information is lost." but in my understanding of the process no bits are thrown away; information is NOT lost.

There are 8M pixels in a 3840x2160 frame. There are only 4M pixels in a pair of 1920x1080 frames. 8M is larger than 4M, by a factor of 2.

If you really wanted to use an eshift to get all 8M pixels, you would need to flash 4 times, with each pixel 3/4 black surround, and 1/4 reflective. Of course that would waste 3/4 of the light, along with other likely issues.

True, but would give an indication of once motion was introduced and viewed from 'normal' viewing angles, how similar the end result is between E-shift and native 4k is, with 4k source material.

You guys have show this already with your stills....the videos would have added another dimension of entertainment!

So the difference between e-shift and the native ones is: E-shift has to shift between "50% of the UHD signal" in each of the two frames. Native have only one big 4k frame so to speak?
All colours and HDR info is displayed in 2K (two fast e-shifting 1080 frames) in an e-shift. But you can NOT display all 4 pixels in the UHD signal. (That would require 4 e-frames)

?

The softening of the image that you listed is from posts you have read. What they were referring to was E-shift softening an image when upscaling 1080P content, not when the projector was receiving 4k content.

Think of it this way. If the projector was to display a single line across the screen that was one pixle wide with E-shift off, it would display a line one pixle wide. Take that same line (regardless if it is 1080p or 2160p source) with E-shift on and it will flash two lines across the screen that are shifted by 1/2 pixle. What we see is a single line that is 1.5 pixles wide. That is why the image is softer when e-shift is on. I personally can't tell the difference unless it is a fixed image and I get up close to the screen.

The softness comes from up scaling 1080p to 2160p. Not sampling 2160p for 2 flashes. When sampling a true 2160p input this does not happen. Granted 50% of the information is missing, but the eye/brain makes up for that.

Almost impossible to tell the difference to the human eye from normal viewing distances.

Javs thread on E-Shift vs Sony 4k is a terrific example of that: https://www.avsforum.com/forum/24-dig...ony-320es.html

....and that's scrutinizing the still images under a magnifying glass, at normal viewing distances with moving images.....??!!

It is (potentially) more complicated than that. Since the two subframes overlap, they cannot display 8 million individually addressable pixels, nor can they display 4 million individually addressible pixels.

However since each subframe overlaps the other by 50%, they effectively multiply each other. The second subframe breaks each pixel into 4 parts, and those 4 parts, while not individually addressable, can have "unique" colors/values, so there are really 8 million "unique" pixels with an e-Shift system. The trick is, what do you put in those two subframes to generate the resulting image? If I were designing an e-Shift system, I would have software that would model/understand the interaction of those two subframes, would evaluate the source, and then would figure out what the two subframes would need to look like to best approximate the original source. For some content, you can get very, very close to 8Mpixel worth of information on screen, but of course there are limitations, like you can't generate single pixel scale, on-off patterns. Of course who knows what JVC actually does.

Calling e-Shift "4 Mpixel" is a convenient way to describe the difference between eShift and native 4K, but it is oversimplifying things.

Those overlapped pixels generate equivalent to 8 MPixel density but not 8 Mpixel information content. Single pixel wide test patterns have shown that JVC simply uses 4 Mpixels from the original image to create the two subframes with 2 Mpixels each with the other 4 Mpixels from the original image being not used.

Correct, it's not actually 8 Mpixel of information, I did not mean to imply it was.

The test patters you're referring to, do they actually show what JVC does, or do they simply show what it doesn't do, i.e. that it doesn't actually do full 8 Mpixels, individually addressible?

I don't know if my logic is flawed, but e-shift appears to cover a lot more area than we have thought it covers previously.

https://www.avsforum.com/forum/attach...700017&thumb=1

The solid blue lines and the dotted purple lines represent e-shift phases, the pink squares are 4k pixel areas. The red areas are the only area not covered by the shift phases.

If this is correct, a heck of a lot more than 4 mega pixels worth of area is sampled! .....I'd still rather have native 4k...but this is pretty darn close!

Attached Thumbnails

 

That picture represent my question about "how it really works?". But I don´t belive the red dots represent "the only missing information in e-shift since the original 4k native is downscaled and processed (analyzed). It´s a bit tricky to think about the two different techniques overlayed at the same time. A bit confusing. But, I have tried to do it more simple to discuss the image with some cordinates.

For example. If we look at the 4k pixels displayed in D1, D2, E1 and E2. (Devided by white lines). Which of these four pixels creates the bigger (blue square) 1 e-shift pixel within the mentioned area. Major part of that blue one is in the D2 area. So is that pixel the major contributor of an "average" processed pixel from the original 4 pixels.

This was what I ment by questioning if the e-shift is an average pixel of the four or just one of them (D2) in this exampe.

Regarding the red dots, I assume these areas are constant white and creating a slim pattern not visible from normal viewing distance. (say 45 degrees viewing angle)

Regards.

Attached Thumbnails

 

No the original 4K signal is not downscaled, two 1920x1080 snap shots are taken of the original 4k image from two 1/4 pixel shifted positions.

The red areas would be similar to interpixel gaps.

Correct, not downscaled has been repeated over and over again in this forum. I guess it is in the logic that people think a 1080P panel would have to downscale a 4K signal. E-shift is quite ingenious.