Resolution requirements for lenses - Page 4 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 269Likes
Reply
 
Thread Tools
post #91 of 1307 Old 06-21-2017, 05:02 PM
AVS Forum Special Member
 
Highjinx's Avatar
 
Join Date: Dec 2002
Location: Australia
Posts: 3,451
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1242 Post(s)
Liked: 510
Since the beam spot on the lens only changes by 1/2 a pixel H/V in e-shift mode. That area is capable of faithfully reproducing a 1920x1080 image when in non-eshift mode. Since the photons of each flash do not crosspaths, why do we need a better lens?

Perhaps we should be considering photons traveling back to the lens from the screen, colliding with photons emanating from the lens, causing resolution loss.
Dave Harper likes this.

May the success of a Nation be judged not by its collective wealth nor by its power, but by the contentment of its people.
Hiran J Wijeyesekera - 1985.
Highjinx is offline  
Sponsored Links
Advertisement
 
post #92 of 1307 Old 06-21-2017, 05:25 PM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Quote:
Originally Posted by Highjinx View Post
Since the beam spot on the lens only changes by 1/2 a pixel H/V in e-shift mode. That area is capable of faithfully reproducing a 1920x1080 image when in non-eshift mode.
"Faithfully reproducing" is part of the issue. No lens perfectly reproduces the image sent through it. There is always scatter, it is just a matter of how much it scatters. There can be more scatter while retaining the pixels in 1920x1080 than reproducing the things that are 1/4 the size of a 1080p pixel to human vision. Scatter matters to human vision, so it doesn't matter whether those scattered photons hit the back of the eye at the same millisecond or so fast that the brain sees those scattered photons as showing up at the same time. With the same percentage of scatter the blur to what a human sees is the same whether the photons hit the eye in the same millisecond or one millisecond apart.
Quote:
Originally Posted by Highjinx View Post
Since the photons of each flash do not crosspaths, why do we need a better lens?
Because it doesn't matter whether they cross paths. Even Dave's expert said that if we are dealing with totally incoherent light (where we are normally talking about light that is pretty incoherent, other than direct lasers) then you need just as good a lens for eShift frames going through the lens at different times as if they go through at the same time.

If you want to understand the subject matter you can't treat light as just particles. The wave properties of light are much more relevant to what we are talking about here.

Between these 4 cases:

A: One 2.7k image.
B: Two different 2.7k images with half pixel offset and going through the lens at different times.
C: Two different 2.7k images with half pixel offset and going through the lens at the same time.
D. One 4k image.

Do you think the first 3 require the same quality of lens? Dave's expert said B and C require the same quality of lens if the light is totally incoherent. This is due to the wave properties of light.
Quote:
Originally Posted by Highjinx View Post
Perhaps we should be considering photons traveling back to the lens from the screen, colliding with photons emanating from the lens, causing resolution loss.
In this case I think you can ignore minor things like how the lens temperature curves are slightly different or photons as particles colliding. They are more like waves in water that can pass right through each other.

--Darin

Last edited by darinp; 06-21-2017 at 05:44 PM.
darinp is offline  
post #93 of 1307 Old 06-21-2017, 05:44 PM
AVS Forum Special Member
 
Highjinx's Avatar
 
Join Date: Dec 2002
Location: Australia
Posts: 3,451
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1242 Post(s)
Liked: 510
Quote:
Originally Posted by darinp View Post
"Faithfully reproducing" is part of the issue. No lens perfectly reproduces the image sent through it. There is always scatter, it is just a matter of how much it scatters. There can be more scatter while retaining the pixels in 1920x1080 than reproducing the things that are 1/4 the size of a 1080p pixel to human vision. Scatter matters to human vision, so it doesn't matter whether those scatter photons hit the back of the eye at the same millisecond or so fast that the brain sees those scattered photons as showing up at the same time. With the same percentage of scatter the blur to what a human sees is the same whether the photons hit the eye in the same millisecond or one millisecond apart.
Because it doesn't matter whether they cross paths. Even Dave's expert said that if we are dealing with totally incoherent light (where we are normally talking about light that is pretty incoherent, other than direct lasers) then you need just as good a lens for eShift frames going through the lens at different times as if they go through at the same time.

If you want to understand the subject matter you can't treat light as just particles. The wave properties of light are much more relevant to what we are talking about here.

Between these 4 cases:

A: One 2.7k image.
B: Two different 2.7k images with half pixel offset and going through the lens at different times.
C: Two different 2.7k images with half pixel offset and going through the lens at the same time.
D. One 4k image.

Do you think all 4 require the same quality of lens? Dave's expert said B and C require the same quality of lens if the light is totally incoherent. This is due to the wave properties of light.
In this case I think you can ignore minor things like how the lens temperature curves are slightly different or photons as particles colliding. They are more like waves in water that can pass right through each other.

--Darin
My uninformed logic tells me:

A & B requires the same lens quality.
C requires a higher quality lens, but this action is not desirable as there will be too much interference, resulting in detail loss/optical distortion.
D Higher quality lens if beam spot on rear element is the same.

May the success of a Nation be judged not by its collective wealth nor by its power, but by the contentment of its people.
Hiran J Wijeyesekera - 1985.
Highjinx is offline  
Sponsored Links
Advertisement
 
post #94 of 1307 Old 06-21-2017, 05:59 PM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Quote:
Originally Posted by Highjinx View Post
C requires a higher quality lens, but this action is not desirable as there will be too much interference, resulting in detail loss/optical distortion.
This is where it is really important to think of light as a wave.

I'll give an example. If there was a huge tidal wave travelling fast from an earthquake toward the west coast it could be only a couple of feet high out in the ocean. If the US created an equal tidal wave to send toward this one, what do you think would happen? The first tidal wave would still hit the west coast and the one the US created could hit Hawaii or Japan. Water probably isn't a perfect example, but hopefully it helps a little bit to think about that.

Waves can pass like strangers in the night. That is why it doesn't matter with totally incoherent light whether you do B or C. That is why Dave's expert said:
Quote:
"But they could pass through at the same time as long as the light was perfectly incoherent."
This is saying that with perfectly incoherent light B requires just as good a lens as C.

With perfectly incoherent light the waves mess with each other randomly. Sometimes they negate each other and sometimes they add together, but when you have all wavelengths it averages out. This is why you don't have to worry about speckle much with most of these projectors, but do with projectors with direct lasers. Those direct lasers have just one wavelength and the interactions are not just random anymore. They can create peaks and valleys that our eyes end up seeing as speckle.

One way to get rid of speckle is to vibrate the screen. This helps average those wave interactions out.

The light we are dealing with usually has quite a few wavelengths for each of the 3 colors and so I think this wave coherence thing is a very minor factor for both speckle and lens quality requirements.
Quote:
Originally Posted by Highjinx View Post
A & B requires the same lens quality.
I understand that this seems logical to a lot of people and even Dave's expert made this mistake. However, if it was pointed out to him that his claim that B & C require the same quality of lens with perfectly incoherent light means that A & C would have the same lens quality requirements, I think this expert would quickly change his position about A vs B.
Quote:
Originally Posted by Highjinx View Post
D Higher quality lens if beam spot on rear element is the same.
I think we are all in agreement about D requiring a better lens than A. D vs B & C is a much more complicated discussion and that is one reason I think we should get to the right answers for the first 3 relative to each other before we try to figure out what it means to have 8 million pixels where half are shifted versus 8 million native pixels.

One more thing about this:
Quote:
Originally Posted by Highjinx View Post
there will be too much interference
Even with native projectors there would be a ton of interference if light was acting like large particles like some people seem to see it. A lenses job is to focus the plane of the chip to the plane of the screen. That doesn't mean however that all pixels go through the spots in the lens that correspond to their spots on the chip. Not all the DLP mirrors have the same angle when they are on. They may be close, but in my experience they are not all the same. And they don't have to be because the lenses job is to focus them to the correct location on the screen. Two adjacent mirrors could have their light pass at least twice before getting to the screen and it wouldn't make all that much difference because of the wave properties of light. In short, the light from the pixels doesn't have to be parallel coming off the chips to get a focused image on the screen.

--Darin

Last edited by darinp; 06-21-2017 at 06:22 PM.
darinp is offline  
post #95 of 1307 Old 06-21-2017, 07:54 PM
Advanced Member
 
AJSJones's Avatar
 
Join Date: Dec 1999
Location: Sonoma County, CA, US
Posts: 886
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 100 Post(s)
Liked: 24
G'dam Safari just wiped out my post before I could post it. Here's a briefer version

The timing of when images go through the lens is a red herring. If a pixel (let's say a square wave white pixel on a black background) goes through the lens it will be blurred from X to X+ some blur (let's say 5%) and we will see that if it's enough blur (particularly if change from a good to poor lens - as per the Olympus capture example). That will happen no matter where (shifted or not) or when (which frame). If we use smaller pixels, say, X/2 = Y, the same will happen but the blur will now be a bigger % of the pixel size because the MTF is lower at a higher spatial frequency (so perhaps Y +10%). If it's a marginal lens, that will be perceived as a greater loss than for a better lens to Y+7% - and that is still true no matter when or where the pixel goes through the lens (on optical axis or ½ pixel off-axis if we're being picky - MTF falls as you go out from the optical axis but that's a separate issue) So a single pixel in the first and second of two frames of 2.7k will be equally blurred no matter when they go through, whether together or separately.

(With the potential for coherence-related interference± from laser light there is the potential for a problem but for normal images, not so much)
RLBURNSIDE likes this.
AJSJones is offline  
post #96 of 1307 Old 06-21-2017, 08:41 PM
AVS Forum Special Member
 
Highjinx's Avatar
 
Join Date: Dec 2002
Location: Australia
Posts: 3,451
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1242 Post(s)
Liked: 510
All this makes me want to give up projection and buy a Sony CLEDIS

May the success of a Nation be judged not by its collective wealth nor by its power, but by the contentment of its people.
Hiran J Wijeyesekera - 1985.
Highjinx is offline  
post #97 of 1307 Old 06-21-2017, 09:06 PM
 
Dave Harper's Avatar
 
Join Date: Feb 2000
Location: Paradise on Earth
Posts: 6,554
Mentioned: 62 Post(s)
Tagged: 1 Thread(s)
Quoted: 3159 Post(s)
Liked: 1723
Here's the real question Darin. Why the heck does it even matter if I or you or my expert is right or wrong when discussing whether 2 sub-frames pass through the lens at the same time, because that doesn't even happen in anything home theater that you or I or any manufacturer we know of, uses and is not relevant to the initial question posed, which I responded to correctly and you refuted incorrectly???

Here below is all (in the scenario that you posted) that matters in these here parts, and coincidentally, all that I ever initially said and responded to, which you then jumped in with your smug attitude and refuted (which was wrong, as I said!):

A: 2K or 2.7K native resolution projector ONLY (no eShift, i.e. - 1080p projector).

B: 2K or 2.7K native resolution projector WITH eShift. (Sub-Frames going through at separate points in time, i.e. - 1080p LCD/DiLA or DLP XPR eShift)

C: 4K native resolution projector ONLY (JVC or Sony 4K DiLA/SXRD, no eShift, which doesn't exist anyway that I know of).

Therefore:
- Scenarios A and/or B have the same lens requirements.

- Scenario C has a higher resolution/MTF lens requirement than scenarios A and B.

This is what you refuted with me initially when I said basically this very thing in my very first post in that thread, replying to @DavidHir and @sage11x 's posts, asking about improved lenses and if they're needed on the UHD60/65, quoted here for your review and pleasure:

Quote:
Quote:
Originally Posted by DavidHir View Post
Is the lens improved over the HT2050 (which is actually quite good for the money)?
Quote:
Originally Posted by sage11x View Post
I would suspect so. The BenQ's get a lot of credit for the glass lens but they have a reputation of suffering some (mild) chromatic aberration. I've yet to see one that is objectionable but it's there. My guess is the lens on the Optoma models would need to be pretty damn good to successfully display a 4k image. When you consider the features vs the price It looks as though most of the budget went towards the light path.
Quote:
Originally Posted by Dave Harper View Post
But consider this, and I don't think I've really researched it yet, but it popped into my head as I read this, so I'm bringing it up.......

If this is an eShift projector, which it is, then at any given point in time there's only half the image shining through the lens, unlike true native 4K like the Sonys, which shoot all 8.3 million out of it at the same time. I'm sure it's more than half really, due to the speed of the switching, but that's the idea I'm thinking anyway. Maybe 2/3rds?

I think @Javs knows lenses pretty well since he is really into pro type cameras, MTF, etc. Maybe he can comment or knows?
Quote:
Originally Posted by darinp View Post
This one gets a lot of people.

In reality it doesn't matter how many pixels go through the lens at one time. The lens requirements are built around the final composite image, not the image at an instant in time that a human cannot detect.

I tried to explain this in a thread here.

https://www.avsforum.com/forum/24-dig...l#post39018834

BTW: You are in good company. The vast majority around here thought that the JVC lenses only needed to be good enough for 1080p since that is all that goes through them at one time, but I am confident I am right that it is the full composite image that determines the lens requirements. Hopefully the examples I gave there help explain why that is.

--Darin
As you can see, you were wrong in your initial contention, so therefore......why can't you be a "stand up guy" and admit you were wrong when you first posted that incorrect info?

It was only AFTER this point, when I and Ruined were trying, doing and proving you wrong (which we did, because you were contending that eShift required a better lens because the "composite image" was going through the lens, which it isn't, the sub-frames go through at different times!), did you start throwing all of that in the weeds, off subject and irrelevant info about Es and 3s and going through the lens at the same time (which does not and is not happening, so it is therefore moot here!).

I also have since said that I certainly may have been confused at what you were getting at and that I could be wrong about whether they would interfere if they go through at the same time, but I also said I didn't particularly care because that is not what is happening anyway, nor will it ever I suspect from any manufacturers!

So please, this horse has been beaten dead a hundred times over. If all you want to hear is that I was wrong about whether two sub-frames at 1920x1080 or 2.7K XPR going through the lens at the same time would interfere, then fine I was wrong, but I don't particularly care because that isn't what is happening anyway!

I was asking my expert the proper questions, thank you very much, and we are in the process of compiling the scenario using his software. If you had any patience and cared what others were saying, doing and thought, then maybe I and others would have responded sooner to your utterly ridiculous rants here and on the other threads.

So please, keep my name out of your posts from now on and stop this personal vendetta against me just because we proved you wrong on the initial question and answer which I and Ruined first gave you correct answers to, which you incorrectly refuted!
Dave Harper is offline  
post #98 of 1307 Old 06-21-2017, 09:48 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,849
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6791 Post(s)
Liked: 6418
You know what.. hell must have froze over because I agree with Dave Harper.

The composite image does not manifest at the lens entry element. In the real world, as technology exists today, E-Shift has a delay associated with displaying composite frames, they do not project at the same moment in time, if they did it would be a blur similar to looking at frame packed 3d projector without the glasses on, you actually go backwards since the pixels are genuinely blurred together and your bran can't even differentiate pixel A from B, and see those in their 100% MTF state let alone form a composite of the two to create a higher resolution 'image' in our heads of what's on the screen.

If Darin is correct, then it also suggests by that logic that using lens shift would also require higher quality lenses, but this is false since a 1080p pixel can only be one size to the entry element on the lens no matter how much shifting is happening. Also, lenses are not sensors broken up into a grid, so where the image is on the surface of the lens is of no relevance. What is relevant is lens MTF for a given pixel size in relation to sensor/panel height and width, if you are using 1080p pixels of a certain pixel pitch per mm, and you want 100% MTF (Full resolution) of that pixel then the lens needs to be of a particular line pairs per millimetre rating, end of story. If you shift by 1/2 pixel in any direction, you are still sending the very same size single 1080p pixel or 2.7k pixel or what have you, thus the lens needs not be any greater.

Its also wrong to assume that via the e-shift process there is now a new pixel with new information formed from the two shifted pixels, there is not, those are not individually addressable, and as such there can be no MTF requirements for them. We are still stuck with 2x 1080p frames where the JVC is concerned, and those are real pixels entering the lens element at separate instances, and thus we can attribute an MTF requirement to them. You cannot place an MTF requirement on a trick that your brain plays with itself, which is what E-Shift does.

All of this can only even be discussed if we assume one thing. The lens in the projector can resolve 1080p to an MTF of 100%. If it is below that, then the argument is not about needing a better lens for e-shift, it is what MTF level lens is acceptable to resolve 1080p? And would E-Shift require a lens which can resolve 80% MTF 1080p or 100%. My opinion, 100% MTF in 1080p is required for E-shift to look its best. I also believe if you have a lens which can resolve 150% MTF of 1080p (Read as closer to 4k MTF levels) you will not see an optical improvement in E-Shift sharpness since we already resolve 100% of 1080p.
Spizz and Dave Harper like this.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is online now  
post #99 of 1307 Old 06-22-2017, 02:26 AM
 
Dave Harper's Avatar
 
Join Date: Feb 2000
Location: Paradise on Earth
Posts: 6,554
Mentioned: 62 Post(s)
Tagged: 1 Thread(s)
Quoted: 3159 Post(s)
Liked: 1723
Quote:
Originally Posted by Javs View Post
You know what.. hell must have froze over because I agree with Dave Harper...

Spoiler!

I think you'd find that we agree on a great many things, and Seegs too, but our personalities and different approaches just seem to get in the way.
Dave Harper is offline  
post #100 of 1307 Old 06-22-2017, 08:11 AM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Javs,

Do you agree with Dave's expert about what would happen if both eShift frames went through the lens at the same time:
Quote:
"But they could pass through at the same time as long as the light was perfectly incoherent."
That is, do you agree that the lens requirements for B in the following are as high as they are for C with totally incoherent light, or do you think that C requires a much higher quality lens than B due to the 2 eShift frames going through the lens at the same time?

A: One 2.7k image.
B: Two different 2.7k images with half pixel offset and going through the lens at different times.
C: Two different 2.7k images with half pixel offset and going through the lens at the same time.
D. One 4k image.

I wonder what people who think that sending the eShift frames at different times than the same time avoids large interference between the light from the 2 eShift frames think would happen if a person shined a flashlight on the ground and somebody else shined another flashlight through the first beam and perpendicular to it. This isn't like 2 water hoses where the second stream would throw the first one off. If people don't understand light as waves they will have trouble understanding this subject matter.

And no, we can discuss with real world lens that are not 100% MTF just like we can discuss lens requirements for 1080p projectors vs 4K projectors even if manufacturers never spend the money to deliver 100% MTF lenses. Cost is relevant to the best lens for a product, just as performance is relevant. A lens that is adequate for just native 1080p or 2.7k may not be adequate for eShift or 4K, regardless of whether they deliver 100% MTF for anything. Not sure they could make a 100% MTF lens even if cost was no object.

I'm not even sure what you mean by 100% MTF. In proper usage of MTF I believe that would mean that the contrast ratio between white lives and black lines would be infinite. Is that what you mean? MTF is a measurement, not what somebody thinks they see. Real world lenses light up the black area between white lines.
Quote:
Originally Posted by Javs View Post
If Darin is correct, then it also suggests by that logic that using lens shift would also require higher quality lenses...
No, not even close to true. It is not the fact that a different part of the lens was used that matters, it is that the lens's job has increased to resolve smaller things for the perceived inages. We could call them subpixels. With eShift the subpixels have correlation. That complicates things, which is one reason to figure out the answer between B and C before trying to figure out the answer between those and D.

If the lens moved so the same part of the lens was used for both eShift frames the lens requirements would be the same as not moving. The key is that there is smaller detail to resolve. Just like my E3 example. The lens has to not blur the fine detail between the E and the 3 in order for a human to see the characters properly, yet that detail never went through a single lens. Just like the eShift case, the E3 example in no way implies that doing a small amount of lens shift requires a much better lens.
Quote:
Originally Posted by Javs View Post
but this is false since a 1080p pixel can only be one size to the entry element on the lens no matter how much shifting is happening. Also, lenses are not sensors broken up into a grid, so where the image is on the surface of the lens is of no relevance.
This part is basically correct. It doesn't change the fact that sending 2 eShift frames at different times requires just as good a lens as sending them at the same time, with incoherent light.

--Darin

Last edited by darinp; 06-22-2017 at 08:35 AM.
darinp is offline  
post #101 of 1307 Old 06-22-2017, 09:49 AM
AVS Forum Addicted Member
 
stanger89's Avatar
 
Join Date: Nov 2002
Location: Marion, IA
Posts: 23,130
Mentioned: 34 Post(s)
Tagged: 0 Thread(s)
Quoted: 4156 Post(s)
Liked: 2384
Quote:
Originally Posted by Dave Harper View Post
Therefore:
- Scenarios A and/or B have the same lens requirements.

- Scenario C has a higher resolution/MTF lens requirement than scenarios A and B.
The problem for me is we keep referencing "requirements" without ever defining what that requirement that a lens must meet is. That really seems to be the key issue here.

Lets simulate some situations, borrowing a screenshot of a 1x1 vertical pattern from the Samsung AB900 from here:

http://www.videovantage.com/?p=819

First lets simulate a "bare minimum" lens, one that can only resolve the peak white/black of a pixel, but not the shape (applied a Gaussian blur of ~1 pixel dimension)
We would see something like this for a 1x1 vertical line pattern:


Now if we shift that, we see something like this:


I don't know about everyone else, but to me, it doesn't look like the resolution was increased by the shifting in that case, quite the contrary in fact. In the native case (first picture) you can see a small amount of pixel detail, where as that detail is obliterated in the shifted version.

Now what happens if we say the requirement for pixel shifting is higher, lets try again with half the blur radius (better "lens")
Unshifted:


Shifted


To me, the better here allows the shifted image to show more detail in this case. IMO, this shows that pixel shifting most definitely does require a better lens, since the "lens" (Gaussian blur) is such that there's really no detail to the original pixel. Intuitively this makes a lot of sense, the resolution on screen, shifted, is to be double the unshifted, and pixel shifting, by design, requires clearly resolved pixels since it's the overlap that splits the native pixels into 4, increasing the resolution.

However..... What about the real world?

Unshifted Samsung AB900:


Shifted Samsung AB900:


Clearly the Samsung's lens is already well, well better than the "minimum requirement" to resolve pixel shifting detail. I think this is really where the debate is falling. On the one hand, the minimum requirement for a lens for an e-shift system is clearly (see first two examples) higher than a non-eshift. However in the real world, we (consumers and projector manufacturers) in the pursuit of maximally sharp images, have already raised our requirements for lenses so far above that minimum requirement (for pixel shifting to be beneficial), that the question is basically irrelevant. All the pixel shifting machines out there today can clearly resolve the native pixel grid of their chips, that means their lenses are already more than good enough to show an improvement when pixel shifting is added.

In realty,
Attached Thumbnails
Click image for larger version

Name:	No Blur.jpg
Views:	202
Size:	18.9 KB
ID:	2199137   Click image for larger version

Name:	Blur native.jpg
Views:	195
Size:	9.1 KB
ID:	2199145   Click image for larger version

Name:	Blur Shift.jpg
Views:	199
Size:	8.1 KB
ID:	2199153   Click image for larger version

Name:	No Blur Shift.jpg
Views:	203
Size:	17.3 KB
ID:	2199161   Click image for larger version

Name:	Less Blur Shift.jpg
Views:	199
Size:	10.5 KB
ID:	2199177  

Click image for larger version

Name:	Less Blur native.jpg
Views:	200
Size:	11.7 KB
ID:	2199185  
RLBURNSIDE likes this.
stanger89 is offline  
post #102 of 1307 Old 06-22-2017, 09:56 AM
Advanced Member
 
Eternal_Sunshine's Avatar
 
Join Date: Jul 2005
Posts: 512
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 32 Post(s)
Liked: 33
This thread is really interesting, but mainly for psychological reasons. It shows (again) that it is really, really hard to consider points that run counter to your own intuition and pre-formed opinions. I experience this every time someone explains to me that time isn't linear.

Either way, 4K e-shift needs a really good lens to work well, a better lens than 1080p needs to work well. darinp explains it, AJSJones explains it, the Olympus example – which none of the naysayers reacted to – explains it, with data to boot.

Anyway, for "normal" HT folks all that matters is what the reviews say wrt perceived sharpness of shipping models. How the manufacturers get there is neither here nor there.
Eternal_Sunshine is offline  
post #103 of 1307 Old 06-22-2017, 10:04 AM
AVS Forum Addicted Member
 
stanger89's Avatar
 
Join Date: Nov 2002
Location: Marion, IA
Posts: 23,130
Mentioned: 34 Post(s)
Tagged: 0 Thread(s)
Quoted: 4156 Post(s)
Liked: 2384
Quote:
Originally Posted by Eternal_Sunshine View Post
Either way, 4K e-shift needs a really good lens to work well, a better lens than 1080p needs to work well. darinp explains it, AJSJones explains it, the Olympus example – which none of the naysayers reacted to – explains it, with data to boot.
I kind of think the issue is the parties are coming at this from two different directions. On the darinp side, and intuitively, given that the shifted image has a higher resolution, means that a better lens is required. The flip side of the argument is that (take JVC for example) e-Shift doesn't really need a better lens than is already in place on those JVCs. But that's because those lenses are already fantasic, not because pixel shifting doesn't have a higher minimum requirement, it's just that the lens is already well above that requirement.

So, for a XPR DLP, odds are the XPR function doesn't really change the lens that's already on those machines, because the lens is probably already more than good enough.
Eternal_Sunshine likes this.
stanger89 is offline  
post #104 of 1307 Old 06-22-2017, 10:22 AM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Quote:
Originally Posted by stanger89 View Post
I kind of think the issue is the parties are coming at this from two different directions. On the darinp side, and intuitively, given that the shifted image has a higher resolution, means that a better lens is required. The flip side of the argument is that (take JVC for example) e-Shift doesn't really need a better lens than is already in place on those JVCs. But that's because those lenses are already fantastic, not because pixel shifting doesn't have a higher minimum requirement, it's just that the lens is already well above that requirement.
I am not disagreeing about a lens already being "good enough" for eShift. They may be. However, if the lens is say a 95 out of 100 in quality for 1080p images, it will not be 95 for eShift images from 1080p panels. It may be 95 for 1080p and 92 for eShift, which are both adequate.
Quote:
Originally Posted by stanger89 View Post
So, for a XPR DLP, odds are the XPR function doesn't really change the lens that's already on those machines, because the lens is probably already more than good enough.
Those lenses may already be better that is required for 2.7k, but that doesn't change that moving to higher resolution has higher lens requirements, whether that be adding eShift or keeping the chips the same size and going to 4k.

I doubt anybody here would disagree that native 4k lens requirements are higher than native 2k lens requirements, and that is true whether current lenses are good enough for 4k or not. Same with eShift except that some people who don't seem to understand how waves work very well think that sending the eShift sub-frames at different times somehow means that you can show higher resolution to human vision without changing the lens requirements from what was necessary with just the native resolution.

I will say that Dave's experts has to understand wave properties fairly well, but they still screwed up. If they were asked my pretty simple question I think they would see the mistake they made. However, Dave seems really adamant about spending his time trying to keep us from seeing his expert's answer to my question.

--Darin

Last edited by darinp; 06-22-2017 at 11:10 AM.
darinp is offline  
post #105 of 1307 Old 06-22-2017, 10:52 AM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Quote:
Originally Posted by Dave Harper View Post
Here's the real question Darin. Why the heck does it even matter if I or you or my expert is right or wrong when discussing whether 2 sub-frames pass through the lens at the same time, because that doesn't even happen in anything home theater that you or I or any manufacturer we know of, uses and is not relevant to the initial question posed ...
First of all, what you are doing here is not intellectually honest. You and your expert both commented on the hypothetical of what it would mean if the eShift sub-frames were sent at the same time instead of at different times. At that time you seemed to revel in what you thought was proof that you were right about that hypothetical from your expert. Now it seems like you are afraid of what the answer would be if you asked your expert my question. Or maybe you already asked it and don't want readers to know your expert's answer, since it would support what I have been saying from the beginning.

As far as the general question about why something hypothetical matters, because that is one of the ways intelligent people learn. I highly doubt your expert owns an eShift projector and yet they commented both on eShift projectors with the light going through at different times versus going through at the same time, and you posted their answer at the time that you thought they were proving you right. Even if eShift projectors didn't exist people could discuss what it would mean if somebody built one. If this were a college test I would have an A and you wouldn't be doing so well, especially if the teacher asked what would happen for a particular example and you wrote that nobody had built one yet, so it didn't matter what would happen.

You are spending way more time trying to justify not asking your expert my question and posting their answer here than it would take you to take a picture of my question and send it to your expert. If your expert is as smart as you think they are then it should take them less than 5 minutes to come up with an answer. I'll even try to lay it out so it is easy for you to take a picture of it and text it to your expert.
Code:
With totally incoherent light do A, B, and C
all have the same lens requirements:

A: One 2.7k image.
B: Two different 2.7k images with half pixel offset
    and going through the lens at different times.
C: Two different 2.7k images with half pixel offset
    and going through the lens at the same time.
D. One 4k image.
Quote:
Originally Posted by Dave Harper View Post
Here below is all (in the scenario that you posted) that matters in these here parts, and coincidentally, all that I ever initially said and responded to, which you then jumped in with your smug attitude and refuted (which was wrong, as I said!):

Quote:
Originally Posted by Dave Harper View Post
If this is an eShift projector, which it is, then at any given point in time there's only half the image shining through the lens, unlike true native 4K like the Sonys, which shoot all 8.3 million out of it at the same time.
Your English there seems pretty simple to me. Your position was clearly stated to mean:

"Since the pixels with eShift images only go through the lens half at a time the lens requirements with 4k input are only the same as for the native resolution. If all 8 million eShift pixels went through the lens at the same time (like native 4k) with 4k input then the lens requirements for eShift would be higher (like native 4k)."

Are you really trying to make readers here believe that this was not the position you were taking? You posted multiple times that it was the time difference that mattered and you clearly posted about what you thought would happen if both eShift sub-frames went through the lens at the same time, even if you want to try to fool people into thinking you never took a position about that.
Quote:
Originally Posted by Dave Harper View Post
This is what you refuted with me initially when I said basically this very thing in my very first post in that thread, replying to @DavidHir and @sage11x 's posts, asking about improved lenses and if they're needed on the UHD60/65, quoted here for your review and pleasure:
...
Quote:
Originally Posted by darinp View Post
In reality it doesn't matter how many pixels go through the lens at one time. The lens requirements are built around the final composite image, not the image at an instant in time that a human cannot detect.
I didn't mention factors I consider minor, like lens temperature curves or that the light isn't totally incoherent (since some wavelengths between blue and green are blocked and so are some wavelengths between green and red), but other than those minor things my statement is still true. Even your expert seemed to agree that with incoherent light the lens requirements would not change between sending both eShift sub-frames at the same time or at different times. The only reason that we haven't gotten a clarification from your expert is because you seem dead set on making sure that the readers here never see your expert's answer to my question. That is not the behavior of a stand up person, no matter how many times you claim to qualify.
Quote:
Originally Posted by Dave Harper View Post
As you can see, you were wrong in your initial contention
No, I was not wrong and you are trying to keep readers from finding out the truth by doing whatever you can to keep readers from seeing your expert's answer to my question.
Quote:
Originally Posted by Dave Harper View Post
... why can't you be a "stand up guy" and admit you were wrong when you first posted that incorrect info?
I would admit if I was wrong, but I'm not.

Readers may not follow all the technical stuff to see if I was right or you are right, but here is one thing I will point out that is often a good indication of who is misinforming.

I want the readers to have more information. I want them to have your expert's answer to my question. You clearly do not. It is even possible you already know your expert's answer to my question and would rather try to mislead people here than let them see it.
Quote:
Originally Posted by Dave Harper View Post
I was asking my expert the proper questions ...
There is a very simple way to find out. Ask them my questions and see if they still have the same position as they did from your questions. Not hard at all, but I understand it would require you being enough of a stand up person to let readers here see their answer.
Quote:
Originally Posted by Dave Harper View Post
and we are in the process of compiling the scenario using his software
That takes a lot longer than simply asking my question and seeing if they really maintain that A, B, and C have the same lens requirements, so why try so hard to keep readers here from seeing the truth about that? Also, software is only as good as the person using it. Your expert should be able to answer my question and calculate for all 4 of the examples in my question.
Quote:
Originally Posted by Dave Harper View Post
So please, keep my name out of your posts from now on ...
This is not how things work around here. There is a reason posts stay on here for years. People can still learn from them. One thing that can be helpful is if there is any question in the future about how trustworthy somebody is, a person can look at how trustworthy they were in the past. For example, if somebody isn't enough of a stand up person to provide the answer to a question that should be pretty simple for their expert. If a person misinforms and plays games to keep readers misinformed then that is useful information for years.

If you want to be known as trustworthy then ask your expert my question and post their answer here.

I have shown my trustworthiness here for 15 years on multiple subjects, including contrast ratio where "experts" would claim I was wrong, then readers would experiment and find out I was right. If we really take this one to conclusion, then people will see that your original conjecture was incorrect. You can stall and try to keep people from seeing the truth, but I will point that out if there is any disagreement in the future, especially if you try to play the same games you are playing here with trying to keep readers from seeing the whole truth about what your expert believes about whether the lens requirements are the same between those 3 cases.

--Darin

Last edited by darinp; 06-22-2017 at 11:07 AM.
darinp is offline  
post #106 of 1307 Old 06-22-2017, 05:58 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,849
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6791 Post(s)
Liked: 6418
Quote:
Originally Posted by darinp View Post
I am not disagreeing about a lens already being "good enough" for eShift. They may be. However, if the lens is say a 95 out of 100 in quality for 1080p images, it will not be 95 for eShift images from 1080p panels. It may be 95 for 1080p and 92 for eShift, which are both adequate.

Those lenses may already be better that is required for 2.7k, but that doesn't change that moving to higher resolution has higher lens requirements, whether that be adding eShift or keeping the chips the same size and going to 4k.

I doubt anybody here would disagree that native 4k lens requirements are higher than native 2k lens requirements, and that is true whether current lenses are good enough for 4k or not. Same with eShift except that some people who don't seem to understand how waves work very well think that sending the eShift sub-frames at different times somehow means that you can show higher resolution to human vision without changing the lens requirements from what was necessary with just the native resolution.
--Darin
Those lenses may already be better that is required for 2.7k, but that doesn't change that moving to higher resolution has higher lens requirements, whether that be adding eShift or keeping the chips the same size and going to 4k.

That's just the thing, E-Shift as it functions today is not a higher resolution to the lenses than native since it travels through the glass at different times. Your brain does the rest. Take a photo of the screen at high enough shutter speed and you should see each shifted frame, they will be 2.7k pixel size each.

You are fixed on the postulation of what if it did travel through the lens at the same time, I say it does not in any current physical real world scenario, so, its totally irrelevant, and that's also what Dave is saying, yet you are seriously fixated on what if it did, Why is that, when its not the current state of technology, its transcended reality and now become a philosophical scenario in a sense.

I doubt anybody here would disagree that native 4k lens requirements are higher than native 2k lens requirements, and that is true whether current lenses are good enough for 4k or not.

Of course I would say a native 4k pixel requires a better lens, but as I said before, its the MTF requirements that are now higher to resolve as close as possible to 100% of the original sharpness of the pixel.This dynamically changes based on pixel size to the input element of the lens. Denser real individually addressable pixels require higher MTF lenses.
Same with eShift except that some people who don't seem to understand how waves work very well think that sending the eShift sub-frames at different times somehow means that you can show higher resolution to human vision without changing the lens requirements from what was necessary with just the native resolution.

I think where you may be tripping up, is the subframes are not created a NEW pixel, thus, the lens requirements remain the same. If there was a new pixel (2x smaller in each direction) with information that differs from either of the original subframes then I would agree with you. Please answer the following then for me in concise simple responses:

1. Do you believe that with the current state of technology, via E-shift, if I sent Frame A through then lens, then a millisecond later Frame B through the lens, while both essentially consisting of 1080p frames, do you need a better lens than what would otherwise be considered a high quality lens for resolving 1080p?

2. If the projector is only flashing one frame, either frame A or B at any one point in time, when do they become higher resolution? Is the projector doing that, or is it your brain blending them together?
Dave Harper likes this.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is online now  
post #107 of 1307 Old 06-22-2017, 06:37 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,849
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6791 Post(s)
Liked: 6418
Quote:
Originally Posted by darinp View Post
And no, we can discuss with real world lens that are not 100% MTF just like we can discuss lens requirements for 1080p projectors vs 4K projectors even if manufacturers never spend the money to deliver 100% MTF lenses. Cost is relevant to the best lens for a product, just as performance is relevant. A lens that is adequate for just native 1080p or 2.7k may not be adequate for eShift or 4K, regardless of whether they deliver 100% MTF for anything. Not sure they could make a 100% MTF lens even if cost was no object.

I'm not even sure what you mean by 100% MTF. In proper usage of MTF I believe that would mean that the contrast ratio between white lives and black lines would be infinite. Is that what you mean? MTF is a measurement, not what somebody thinks they see. Real world lenses light up the black area between white lines.

No, not even close to true. It is not the fact that a different part of the lens was used that matters, it is that the lens's job has increased to resolve smaller things for the perceived inages. We could call them subpixels. With eShift the subpixels have correlation. That complicates things, which is one reason to figure out the answer between B and C before trying to figure out the answer between those and D.

If the lens moved so the same part of the lens was used for both eShift frames the lens requirements would be the same as not moving. The key is that there is smaller detail to resolve. Just like my E3 example. The lens has to not blur the fine detail between the E and the 3 in order for a human to see the characters properly, yet that detail never went through a single lens. Just like the eShift case, the E3 example in no way implies that doing a small amount of lens shift requires a much better lens.
This part is basically correct. It doesn't change the fact that sending 2 eShift frames at different times requires just as good a lens as sending them at the same time, with incoherent light.

--Darin
And no, we can discuss with real world lens that are not 100% MTF just like we can discuss lens requirements for 1080p projectors vs 4K projectors even if manufacturers never spend the money to deliver 100% MTF lenses.

Why, because I think this is the single most important aspect of this entire discussion, you ask do you need a better lens for E-Shift, I counter with, if you had a lens which could resolve all the detail there is to resolve in 1080p, and e-shift consists of two 1080p images, correlated or not, they are simply 1080p sized pixels which are shades of a colour fundamentally, they are neither E or 3 as in your examples, they are simply shades of a colour, now, given that, and given they are still 1080p sized pixels, and we have a lens that can resolve all there is to resolve in 1080p, why on earth would you need a higher quality lens?

Especially so when the pixel travel through the element at different times, there is no combining happening at any one point in time, your brain does that, even a camera can barely accurately be synced to capture precisely the moment it happens.

A lens that is adequate for just native 1080p or 2.7k may not be adequate for eShift or 4K, regardless of whether they deliver 100% MTF for anything. Not sure they could make a 100% MTF lens even if cost was no object.

Define adequate? You need to talk in MTF terms in relation to the panel size in line pairs per mm to do this. I submit to you that most lenses are not even adequate for 1080p in order to even begin to have this discussion, until such point as they are 100% MTF lenses at 1080p for the panel size, then the real issue is just that. Everything else is hypothetical. This discussion only really makes sense if you could resolve 100% of the 1080p frame in the first place, impossible or not.

If you never even resolve 100% of the 1080p pixel in the first place, how can you judge anything beyond that when still effectively using 1080p subframes which are not even fully resolved? That is my position, and I cannot move past this until we can hypothetically at least discuss this as if the 1080p pixels are in fact 100% resolved in the first place.

The real discussion in this thread should be:

Of what MTF resolution does a lens need to be to show a discernible resolution and sharpness increase when comparing a 1080p frame to a resultant combined e-shifted frame.

Now to put my hat on, and jump to the other side of the fence for a moment, coming at this from another angle...



Now with the above image, I can related to this.

I suppose the yellow pixel is technically new information since its the culmination of the two subframes, and thus there is a new colour there. I have to recognise this. Now this all might be a misunderstanding of comminication, if it is, then let me clarify my position.

The two sub-frames are 1080p pixels, I suppose 100% MTF is impossible because what would be 100% MTF for 1080p would also technically be 100% MTF for 4k since it would define the edge of the pixel as infinitely sharp, and thus is impossible in the real world.

Its clear there is a new yellow pixel there which did not exist before, but as it stands with current technology this yellow pixel does not exist in the lens, its in our brains. From this angle, I maintain that the lens be as close to 100% MTF as humanely possible in the first place for 1080p to present this yellow 'illusion' in its sharpest form. I guess that aligns with your position Darren?

For me its as simple as this:

Do you agree that the lens requirements for B in the following are as high as they are for C with totally incoherent light, or do you think that C requires a much higher quality lens than B due to the 2 eShift frames going through the lens at the same time?

A: One 2.7k image.
B: Two different 2.7k images with half pixel offset and going through the lens at different times.
C: Two different 2.7k images with half pixel offset and going through the lens at the same time.
D. One 4k image.

Answer: A lens can ALWAYS be better. Since in the real world we never even resolve 2.7k or 1080p what have you at 100% MTF then the lens can ALWAYS be improved.
'Lens Requirements' is a moving target unless we define this before having this discussion, the discussion has no base or 'control' point of reference.
Dave Harper likes this.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is online now  
post #108 of 1307 Old 06-22-2017, 07:13 PM
AVS Forum Addicted Member
 
stanger89's Avatar
 
Join Date: Nov 2002
Location: Marion, IA
Posts: 23,130
Mentioned: 34 Post(s)
Tagged: 0 Thread(s)
Quoted: 4156 Post(s)
Liked: 2384
Quote:
Originally Posted by Javs View Post
Now with the above image, I can related to this.

I suppose the yellow pixel is technically new information since its the culmination of the two subframes, and thus there is a new colour there. I have to recognise this. Now this all might be a misunderstanding of comminication, if it is, then let me clarify my position.
And that the basis for pixel shifting, and why it actually works (as you've shown so clearly in your screenshots), the second subframe effectively splits each pixel in four new "pixels", or individual areas of color (not individually addressable of course). Thus they can reproduce a sharper image (more high frequency information) than just a single unshifted frame.

Quote:
The two sub-frames are 1080p pixels, I suppose 100% MTF is impossible because what would be 100% MTF for 1080p would also technically be 100% MTF for 4k since it would define the edge of the pixel as infinitely sharp, and thus is impossible in the real world.
Yes exactly, and where back to what's "requirements" mean. Of course if we assume perfect reproduction of the 1080p subframe, then no improvement is necessary for the shifted version, but then again, as you say, no improvement would be necessary for 4k either. But we live in the real world, so we're back to, what do we mean when we say "requirement". I'll refer you back to my post of simulated images.

Quote:
Its clear there is a new yellow pixel there which did not exist before, but as it stands with current technology this yellow pixel does not exist in the lens, its in our brains. From this angle, I maintain that the lens be as close to 100% MTF as humanely possible in the first place for 1080p to present this yellow 'illusion' in its sharpest form. I guess that aligns with your position Darren?
I think this is what Darrin was getting at with his response to me. Modern 1080p machine lenses are pretty good, so they might achieve 95% of perfection unshifted, and only 92% when shifted. In the real world that's probably irrelevant but it is a difference.

The key thing to remember though is the final shifted image is physically twice the resolution of the native one. Sure it doesn't all get to the screen at the same time, but it does get to the screen. Think of it like a CRT in a way. Technically a CRT only displays one pixel at a time, so it only needs a lens good enough for 1 pixel right? But if it can't focus that pixel to a fine enough degree, you won't be able to combine those pixels on screen to create a high resolution image. Doesn't an HD CRT need a better lens than an SD one?

Same is true of e-shift, you have to be able to focus the native pixels to a high enough degree that each quarter physical pixel is resolve well enough for the resulting double resolution, while that's not strictly speaking required for the native resolution itself.
stanger89 is offline  
post #109 of 1307 Old 06-23-2017, 12:51 AM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
I'm really liking the last 3 posts.

Javs,

I think you learned some things as you were working them out during your posts, which is always a good thing IMO.

If you want me to go back and address each of the individual items I will, but I think it is best if I concentrate on one thing first.
Quote:
Originally Posted by Javs View Post
Now to put my hat on, and jump to the other side of the fence for a moment, coming at this from another angle...



Now with the above image, I can related to this.

I suppose the yellow pixel is technically new information since its the culmination of the two subframes, and thus there is a new colour there. I have to recognise this. Now this all might be a misunderstanding of communication, if it is, then let me clarify my position.
If the input was a 4K signal and it had the single pixel of yellow surrounded with red and green like you showed then a 4K projector should be able to show that. A 1080p only projector would have trouble showing that. BTW: I'm ignoring whether there is less chroma info.

A 1080p eShift projector has some correlation between the 4K sized pixels (sub-pixels in 2K space) due to overlap, but not total correlation. There are some things in 4K space that the eShift projector can display properly and some that it can't.

For your example the lens for both eShift and true 4K has the job of displaying the yellow pixel (or sub-pixel). The lens in 1080p mode never has that task. It only has to display things that are 4 times the size.

As far as the 100% MTF argument, I think we are in agreement that if a lens can perfectly display every single little detail (like the corners) in a 1080p pixel then it can display the smallest detail the human brain sees in eShift images, but that is true of native 4K images also. That is, if the question was whether projectors require a better lens with native 4K than native 1080p we could say that a perfect lens for 1080p would be good enough for 4K. That is one reason I tried to explain what I meant by needing a better lens in the first post if this thread, since nobody actually needs a better lens for entertainment purposes. We still want lenses that are good enough to show the smallest details from the source that the projector can display.

As far as whether sending the two eShift sub-frames at different times means the lens requirements are lower, I'm not sure if you remember in school where the teacher would say that some problems could be broken into individual pieces. With light of all wavelengths, this is a problem that can be broken into individual pieces. The reason is that light acts like waves, and just like how you can shine one flashlight through the beam of another flashlight without obscuring the image the first flashlight makes.

If you use light of all wavelengths then 2 eShift subframes can be sent through the lens at the same time without interfering with each other. The interference of the waves will cancel out to essentially zero.

I'll try using your example. Let's say that you want the green to be 30 nits, the red to be 10 nits, and the yellow to be 40 nits, but you wanted a black sub-pixel next to the yellow sub-pixel and the same size.

I'm simplifying, but let's say the lens blurs the 1080p pixels a little and sends an extra 20% of the light about 1/8th of a 1080p pixel into the pixels next to where the light is supposed to go.

Just considering the pixels in your example, if you send the 30 nits into the green pixel 6 nits will go halfway into that sub-pixel that you want to be black.

If you then send the red pixel next eShift sub-frame that black area will get 2 nits of red.

In this case your eye will see 8 nits of scatter.

If you send the 2 eShift sub-frames at the same then what happens? You eyes still see 8 nits of scatter that wasn't supposed to be there.

The job of the projectors we are discussing is to deliver images to humans and that is what matters for the lens requirements. And for human vision it doesn't really matter whether the 2 eShift sub-frames are sent at the same time or different times other than some minor factors, as long as the light isn't coherent, and even then it would likely be minor compared to the extra requirements from trying to display detail that is smaller than the native pixel size.

This is why even Dave's expert said it didn't matter whether the eShift frames were sent at the same time or different times.

As far as lines per mm and MTF, I think that eShift forces thinking about things in a new way. That might be one reason Dave's expert had trouble thinking about it and ended up claiming the rather ridiculous combination that with totally incoherent light sending the sub-frames at the different times requires just as good a lens as if you send them at the same time, but sending them at different times doesn't require any better lens than the native resolution. That is somebody who is getting paid to know about lenses who can't even come up with a reasonable combination of answers, so I can understand how those who aren't paid to do this stuff and aren't that into physics questions would have trouble with it.

Please let me know if you want me to try to address something else that you already posted or if you have any new questions/issues.

--Darin

Last edited by darinp; 06-23-2017 at 01:08 AM.
darinp is offline  
post #110 of 1307 Old 06-23-2017, 03:51 AM
AVS Forum Special Member
 
Ruined's Avatar
 
Join Date: Jul 2002
Posts: 6,807
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2597 Post(s)
Liked: 1240
Quote:
Originally Posted by Javs View Post
The two sub-frames are 1080p pixels, I suppose 100% MTF is impossible because what would be 100% MTF for 1080p would also technically be 100% MTF for 4k since it would define the edge of the pixel as infinitely sharp, and thus is impossible in the real world.
Javs just ended the debate with this statement and was exactly what I was getting at. A lens that fully resolves 2716x1528 would display the pixels, pixel edges, and spaces between the pixels as incredibly sharp with minimal artifacts .

Thus you could evaluate a lens for sharpness and artifacts at 2716x1528 and if this lens is extremely sharp at this resolution with minimal aberrations it will work fantastic for either 2716x1528 or XPR 4k... No further evaluation is needed as if it has blur/abberations with XPR it will also have blur/aberrations without XPR, and arguing deeper than this based on hypotheticals not based in reality is simply going to net you a red herring.

In DLPs case this sharp lens for XPR would not work for native 4k, though, as the DMD is much larger for native 4k (1.38" vs 0.67") and thus native 4k on DLP requires a much larger and more expensive lens (bigger elements/more glass) . There is no native 4k DMD as small as the XPR DMDs and hence the lenses for native 4k on DLP are much more expensive than XPR 4k or 2716x1528 . Hence Darin your claim that XPR is just as expensive to implement as a native 4k for DLP or has the same requirements as native 4k in terms of lens is false. And before you accuse me of changing my argument this was in my very first post on the matter.
Dave Harper and coug7669 like this.

Last edited by Ruined; 06-23-2017 at 06:16 AM.
Ruined is offline  
post #111 of 1307 Old 06-23-2017, 08:28 AM
AVS Forum Special Member
 
Pete's Avatar
 
Join Date: Jan 2001
Location: New England
Posts: 2,964
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 262 Post(s)
Liked: 86
For the TI .67 4K chip, the "ideal" lens MTF would be 93 line pairs per millimeter. An "adequate" lens MTF for this chip is any point below 93 lp/mm that a projector manufacture feels is still impressive enough to sell projectors without breaking the bank.

Last edited by Pete; 06-23-2017 at 12:41 PM.
Pete is offline  
post #112 of 1307 Old 06-23-2017, 02:46 PM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Quote:
Originally Posted by Ruined View Post
Javs just ended the debate with this statement...
No, Javs did not and even he seems to understand that. By this line of thinking if the question was whether lens requirements were the same between 1080p and 4K with the same chip size Javs could end the argument with the same 100% MYF argument.
Quote:
Originally Posted by Ruined View Post
A lens that fully resolves 2716x1528 would display the pixels, pixel edges, and spaces between the pixels as incredibly sharp with minimal artifacts .
AsI basically said, real world lenses all scatter light, but if you had a lens that was way better than was required for 1080p that same lens can be good enough for eShift or native 4K, since if it can display every fine detail in larger objects (like 1080p pixels) it can do well with smaller objects (like the finer detail that eShift or 4K projectors are trying to project). That does not mean that if you scored that lens for its task that it would receive the same score. Even if you had a lens that you scored at 99.9% you might score it closer to 99% for eShift or true 4K.

Few rojectors ever require 99.9% quality lenses though. If a lens exceeds the requirements for a task it does not mean that the lens requirements aren't different between native and eShift or 1080p+eShift or 2.7k+eShift and native 4K, with the same chip sizes.
[QUOTE=Ruined;53827193]In DLPs case this sharp lens for XPR would not work for native 4k, though, as the DMD is much larger for native 4k (1.38" vs 0.67")[/qoute]It doesn't matter whether TI has 0.67" native 4K chips yet or not, intelligent people can discuss what lens requirements will be if they do. By your same "fully resolved" argument would you say that lenses that are good enough for native 2.7k are good enough for native 4K, at the same chip size?
Quote:
Originally Posted by Ruined View Post
Hence Darin your claim that XPR is just as expensive to implement as a native 4k for DLP or has the same requirements as native 4k
Please don't make things up. That is most definitely not true and I never said it was. As far as lens requirements between those 2 I said it is a more complicated discussion and people should learn to walk before they learn to run. For example, with my A, B, C, and D choices they should learn how the lens requirements differ between A, B, and C before trying to figure out how they differ between B, C, and D.
Quote:
Originally Posted by Ruined View Post
And before you accuse me of changing my argument this was in my very first post on the matter.
In your original post you said multiple things, including:
Quote:
Originally Posted by Ruined View Post
In this case, if you had two dlp projectors, one 2716x1528 0. 67" without XPR and one 2716x1528 0.67" with XPR, I don't believe you would need a more spec'd lens as:
...
2. There would be no difference in density of projected pixels off the DMD, as it is a mechanical process. There is no bleed from one frame to next and the size of the pixels remains the same regardless if shifted or not. The overlap happens on the screen.
Are you now saying you did not take a position in that post that would mean that B would only need as good a lens as A, while C would require a reasonably better lens than B, given the same chip size for all?

A: One 2.7k image.
B: Two different 2.7k images with half pixel offset and going through the lens at different times.
C: Two different 2.7k images with half pixel offset and going through the lens at the same time.
D. One 4k image.

Sure looked to me like you were taking a position about whether the eShift frame going through the lens at the same time or different times had much effect on lens requirements.

--Darin

Last edited by darinp; 06-23-2017 at 02:51 PM.
darinp is offline  
post #113 of 1307 Old 06-23-2017, 04:20 PM
AVS Forum Special Member
 
Ruined's Avatar
 
Join Date: Jul 2002
Posts: 6,807
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2597 Post(s)
Liked: 1240
A same as B, given lens fully resolves A with minimal abberations
C irrelevant as not occurring in reality
D much higher requirement for DLP due to much larger DMD. Still higher requirement for non dlp because the native pixels are smaller with fine lines in between them that must be rendered.

All as originally posted.

Bottom line is requirement for B never goes up unless you are willing to accept an inferior quality A, which thus would fail to fully resolve the native.
Dave Harper likes this.

Last edited by Ruined; 06-23-2017 at 04:39 PM.
Ruined is offline  
post #114 of 1307 Old 06-23-2017, 04:38 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,849
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6791 Post(s)
Liked: 6418
Quote:
Originally Posted by Ruined View Post
A same as B, given lens fully resolves A with minimal abberations
C irrelevant as not occurring in reality
D much higher requirement for DLP due to much larger DMD

All as originally posted.
D actually depends on pixel pitch, and lens cost is not representative of lens quality based on only the fact it is larger.

Larger DMD's if resulting in larger pixels actually require glass with less resolving power per mm of the surface. So technically the resolution requirements would be lower. (See below)

However, Im guessing the larger DMD's actually use the same smaller pixel pitch of the 0.69" and just manufacture larger panels with the same proiess. Meaning per mm of the panel there is tghe same amount of pixels, thus the lens MTF requirements are actually the same, but with the added cost of course of needing larger elements, but like I said thats not correlated with lens quality.

The lens in the JVC Z1 for eg, thats a true native 0.69" chip, with a pixel pitch of 4um, so, the lens needs to resolve far more pixel density per mm. They actually doubled the lens requirements over the current line of E-Shift models because the pixels are half as large.

Ill quote an older post of mine which goes over it:

Quote:
Yeah absolutely. This is a big issue in cinema cameras too. There is one reason why the larger the sensor size in film cameras the higher the quality image you get, its not just about letting in more photons, its also about reducing the load on the lens to resolve minuscule details.

The more glass surface area, or the bigger the space for photons to pass through, the less pure the lens has to be. In film, and in relation to lenses we call this MTF, Modulation Transfer Function. The Z1 lens based on its DMD (Sensor as you would call it in a film camera) requires about 120lp/mm of resolving power, meaning for every one mm of the DMD surface, there are 240 pixels!, that's 120 possible line pairs (Read this as contrast, or single pixel checker-board test pattern), the lens has to be able to transfer that pixel density of detail through its first few groups of lens elements before it is effectively enlarged. This is quite a task!.

For eg, the latest RED WEAPON Helium 8K cameras are still super 35mm in sensor size, they have a 5 micron pixel pitch (200 Pixels per mm / 100 Line Pairs Per MM), they require really great lenses. The Z1 requires even greater lenses since its 3.8 micron! 240pixels per mm / 120 line pairs.

The Arri Alexa 65 (6.5K) does not, since they just took 3x Alexa sensors and stitched them together, actually reducing the load on the 'glass' since the sensor merely tripled in size! The pixel pitch on the Alexa 65 is 8.3 microns (The current E-Shift panels are also 8 microns), which means there is only 121 pixels per mm on the sensor, MUCH less load required on the lenses, 60 line pairs per mm... In the film world they talk a lot about why the Alexa 65 is the better camera, and the reasons I just stated are at the top of the list! The resolving power of the larger sensor at 6K will destroy the RED camera with its much smaller sensor if the same lens is used. Simple as that.

What JVC have done is quadrupled the pixel density that what was on virtually the same overall sized DMD they used with the E-Shift machines, which is odd since they already had a 4K panel sitting around based on the same pixel pitch of the e-shift machines previously, its what they used in the older 4K cinema machines they sold, it was just larger, over 1 inch panel size. They also have an 8K panel which is near 1.5inch and the same pixel pitch! Why couldn't they use that, or improve on it??!!

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is online now  
post #115 of 1307 Old 06-23-2017, 04:48 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,849
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6791 Post(s)
Liked: 6418
Here is an interesting chart of the MTF performance of the Arri Alexa 65's actual sensor. This is before a lens is even attached and it quite interesting. Its probably the single sharpest digital film camera there is right now so I think its relevant.



You can see that while the cameras capture resoltion is actually 6.5K, the MTF performance at 4k is only roughly .35 which is 35% MTF.

You can also see by this graph that 100% MTF at higher resolutions is impossible since 100% MTF in my understanding actually represents a single solid colour such as black or white. As resolution increases, the modulation frequency does, but since the details become smaller, the modulation curve should actually decline in a smooth fashion.

I would say that for a lens to truly resolve a given resolution to what we would see as an acceptable standard would be 50% MTF for the given resolution.

Here is another graph which shows what the 'perfect' lens MTF curve would look like.


JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is online now  
post #116 of 1307 Old 06-23-2017, 04:49 PM
AVS Forum Special Member
 
Ruined's Avatar
 
Join Date: Jul 2002
Posts: 6,807
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2597 Post(s)
Liked: 1240
Quote:
Originally Posted by Javs View Post
D actually depends on pixel pitch, and lens cost is not representative of lens quality based on only the fact it is larger.

Larger DMD's if resulting in larger pixels actually require glass with less resolving power per mm of the surface. So technically the resolution requirements would be lower. (See below)

However, Im guessing the larger DMD's actually use the same smaller pixel pitch of the 0.69" and just manufacture larger panels with the same proiess. Meaning per mm of the panel there is tghe same amount of pixels, thus the lens MTF requirements are actually the same, but with the added cost of course of needing larger elements, but like I said thats not correlated with lens quality.

The lens in the JVC Z1 for eg, thats a true native 0.69" chip, with a pixel pitch of 4um, so, the lens needs to resolve far more pixel density per mm. They actually doubled the lens requirements over the current line of E-Shift models because the pixels are half as large.

Ill quote an older post of mine which goes over it:
Oh I'm aware of that. But since you have twice as many pixels in the 1.38" plus the much larger glass required for the larger DMD would be far more expensive than the 0.67 xpr lens.

IOW 0.67 actually more ppi but not enough to make up cost of larger glass elements required for 1.38 which still requires very highly resolving elements.

BTW for context this whole argument started with Darin's assertion that an 0.67" XPR DMD would require a lens just as costly as a native 4k lens for DLP, so XPR wouldn't be a cost saver or something to that extent. Hence my objection due to the large and highly resolving glass elements required for a 1.38" native 4k DMD. After that it devolved into a bunch of info with shift and virtual resolutions irrelevant to the original assertion, as the native 4k lens Darin is imagining in his example for an 0.67" DMD does not exist in reality since the 0.67" DMD tops out at 2516x1528.
Dave Harper likes this.

Last edited by Ruined; 06-23-2017 at 05:11 PM.
Ruined is offline  
post #117 of 1307 Old 06-23-2017, 05:11 PM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Quote:
Originally Posted by Ruined View Post
C irrelevant as not occurring in reality
What is it with the people who are fine with showing everybody here that they are incapable of answering a hypothetical question, while they claim they understand the subject matter.

Try that kind of stunt during an interview and we would probably appreciate it that you just showed we should never hire you.

Fiirst Dave commented on that hypothetical more than once, then looks like he realized getting straight answers to those questions from his expert would show that he was wrong. So Dave, who claims to be a stand up guy, started arguing that this situation he had used as part of his "proof" that he was right was no longer relevant and he wouldn't let readers here see his expert's follow up answers to these questions.

Now Ruined tries the same stunt. He may understand the subject matter well enough to know an actual answer about B vs C would make him look bad, since he would either have to take a clearly stupid position or agree with Dave's expert about B vs C.

Ruined,

So your final answer to B vs C is really going to be the one that supports that you don't know enough about physics to provide an intelligent answer?

If this were a college physics course I would have an A and Dave and Ruined would be wondering why they were doing so poorly.

--Darin
darinp is offline  
post #118 of 1307 Old 06-23-2017, 05:12 PM
AVS Forum Special Member
 
Ruined's Avatar
 
Join Date: Jul 2002
Posts: 6,807
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2597 Post(s)
Liked: 1240
We could also speculate how high Pegasus could fly without its wings being burned, but the answer would be just as grounded in reality and just as useful. Don't have time to debate things that are not actually happening.

Last edited by Ruined; 06-23-2017 at 05:26 PM.
Ruined is offline  
post #119 of 1307 Old 06-23-2017, 05:22 PM
AVS Forum Special Member
 
darinp's Avatar
 
Join Date: Jul 2002
Location: Seattle, WA
Posts: 4,735
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 543 Post(s)
Liked: 749
Quote:
Originally Posted by Ruined View Post
D much higher requirement for DLP due to much larger DMD.
Did you have trouble comprehending the question? It was clear; same chip size. You really would have lots of trouble with a college physics test.
Quote:
Originally Posted by Ruined View Post
We could also speculate how high Pegasus could fly without its wings being burned, but the answer would be just as grounded in reality and just as useful. Don't have time to debate things that are not actually happening.
You just spent more time than it would have taken to provide an actual answer if you actually understood the subject matter. I think you and I bith know that you don't understand the subject matter well enough to provide a reasonable answer to B vs C though.

Please stop acting like you understand this subject matter as long as you can't provide an actual answer to B vs C, since if you actually did understand the subject matter it would take hardly any time for you to answer that question.

--Darin

Last edited by darinp; 06-23-2017 at 05:26 PM.
darinp is offline  
post #120 of 1307 Old 06-23-2017, 05:28 PM
AVS Forum Special Member
 
Ruined's Avatar
 
Join Date: Jul 2002
Posts: 6,807
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2597 Post(s)
Liked: 1240
Quote:
Originally Posted by darinp View Post
Did you have trouble comprehending the question? It was clear; same chip size. You really would have lots of trouble with a college physics test.You just spent more time than it would have taken to provide an actual answer if you actually understood the subject matter. I think you and I bith know that you don't understand the subject matter well enough to provide a reasonable answer to B vs C though.

Please stop acting like you understand this subject matter as long as you can't provide an actual answer to B vs C, since if you actually did understand the subject matter it would take hardly any time for you to answer that question.

--Darin

Will also add Darin, the only thing you've accomplished in all this time is loss of respect as you continue to insult those who disagree with you, expert or not. The expert became an agent of misinformation when he disagreed with you. You keep posting and posting the same thing over and over and calling names and disrespecting until people don't even care to read it anymore. Sad.
Dave Harper and coug7669 like this.
Ruined is offline  
Sponsored Links
Advertisement
 
Reply Digital Hi-End Projectors - $3,000+ USD MSRP

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off