NEW RANGE JVC 2014 - Page 37 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #1081 of 4136 Old 10-20-2013, 12:19 PM
AVS Special Member
 
Highjinx's Avatar
 
Join Date: Dec 2002
Location: Australia
Posts: 1,770
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 23
Quote:
Originally Posted by noah katz View Post

Unless they totally overengineered it (not likely), their current iris mechanism is likely totally unsuitable in response time and durability.
Why would they struggle to develop formulae for something as complex as psychovisual requirements when they could just use lookup tables?

As I said NOT using the same Iris(I mean't another iris in close proximity)...and a modified motor mechanism.

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 #1082 of 4136 Old 10-20-2013, 01:03 PM
AVS Addicted Member
 
mark haflich's Avatar
 
Join Date: Dec 2000
Location: brookeville, maryland, usa
Posts: 19,933
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 443 Post(s)
Liked: 488
I wouldn't assume the iris in the lenses they have used for several years would not be fast enough even if they have been used in the past to be set at a variety of constant positions. Normal through the lens finder cameras have irises in their detachable lenses that work lightening quick.

Mark Haflich

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

call me at: 240 876 2536
mark haflich is offline  
post #1083 of 4136 Old 10-21-2013, 02:19 AM
AVS Special Member
 
Highjinx's Avatar
 
Join Date: Dec 2002
Location: Australia
Posts: 1,770
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 23
The more I think about this the simpler it seems to implement, lens irises have been around for decades, adding a motorised control is fairly new, but nothing that cannot be done fairly simply without patent issues.

How difficuilt would it be to have say a 100 slides of captures of different scenes with dissimilar 'Picture Levels', the engineers have a group of golden eyes doing the observing, while aperture size and gama manupilation is performed for each of the slides to acheive best image quality.

This data becomes the foundation for aperture size control/gama modification for the various picture levels via the software.

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 #1084 of 4136 Old 10-21-2013, 02:37 AM
AVS Special Member
 
Seegs108's Avatar
 
Join Date: Jul 2007
Location: Schenectady, New York
Posts: 4,149
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 340 Post(s)
Liked: 242
If it were as easy as you think every manufacturer out there would have it down by now. It is a lot more complex than you think it is. You're grossly over simplifying the process. The implementation needs to be smart and know not to act sometimes. It cannot actually change on a frame to frame basis every time. If this were the case it would be WAY too noticeable. It needs to look at several frames in advanced to get an average picture level of the "scene" thats about to be projected. It needs to say "okay there are x number of frames with y amount of bright pixels in them and then there are x amount of frames where there are very few to no bright pixles in them. How should I proceed to adjust the contrast/gamma/ect?" " OR "There are x amount of frames with y amount of not so bright pixels in them. Can I close the iris further because they aren't as bright? But wait the frame after has a lot of very bright pixels, do I not close the iris so much? Will the iris action be too obvious?" To make it smart and to know how to react appropriately it needs to be given instructions on what to do with common, and I'm sure Sony has given it instructions on what to do with not-so common, changes in contrast. Which is what makes it so smooth and well implemented. If you were to watch those 100 frames (slides) at 24 frames per second and the iris adjusted individually for each frame you'd have one obnoxious looking scene with the iris bouncing all over the place.

----------------------------------------------------------------------------

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Seegs108 is online now  
post #1085 of 4136 Old 10-21-2013, 02:59 AM
AVS Special Member
 
Highjinx's Avatar
 
Join Date: Dec 2002
Location: Australia
Posts: 1,770
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 23
Quote:
Originally Posted by Seegs108 View Post

If it were as easy as you think every manufacturer out there would have it down by now. It is a lot more complex than you think it is. You're grossly over simplifying the process. The implementation needs to be smart and know not to act sometimes. It cannot actually change on a frame to frame basis every time. If this were the case it would be WAY too noticeable. It needs to look at several frames in advanced to get an average picture level of the "scene" thats about to be projected. It needs to say "okay there are x number of frames with x amount of bright pixels in them and then there are x amount of frames where there are very few to no bright pixles in them. How should I proceed to adjust the contrast/gamma/ect?" To make it smart and to know how to react appropriately it needs to be given instructions on what to do with common, and I'm sure Sony has given it instructions on what to do with not-so common, changes in contrast. If you were to watch those 100 frames (slides) at 24 frames per second and the iris adjusted individually for each frame you'd have one obnoxious looking scene with the iris bouncing all over the place.

It would be impossible for any manufacture to be precise for every frame out there in movie land. There are very few if any successive frames where bright would go to dark and then bright again. I doubt each frame is looked at in detail per se, but rather a response to an average voltage level. To gather the data required one would need to ramp up the picture level per successive slide and adjust the gamma where there is a benefit.

It's don't feel it's too complex, the image is being dimmed where required to assist the perception of deeper black and gamma boosted in the region required to compensate for brightness compression. The JVC's high native CR will enable it to operate its IA from much lower levels than the Sony, although the 10x multiplier is claimed I would think it is in response to the marketing game. Though it is available I doubt it would be required(10x)

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 #1086 of 4136 Old 10-21-2013, 03:14 AM
AVS Special Member
 
Seegs108's Avatar
 
Join Date: Jul 2007
Location: Schenectady, New York
Posts: 4,149
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 340 Post(s)
Liked: 242
Quote:
Originally Posted by Highjinx View Post

It's don't feel it's too complex,

Feel free to contact JVC to let them know how easy it is. Tell them they better not screw it up because it's too easy to fail. rolleyes.gif

I'm sure they will have a great implementation, but please don't sit there and tell me it isn't complex (or cheap) to do one correctly. If that was the case everyone would have great DI's. Let me check.....nope, most don't.

----------------------------------------------------------------------------

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Seegs108 is online now  
post #1087 of 4136 Old 10-21-2013, 04:13 AM
AVS Special Member
 
coderguy's Avatar
 
Join Date: Dec 2006
Posts: 9,984
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 147
Math coding is rarely ever short and easy, even the simplest math can sometimes be a MAJOR pain in the rear in code. Some things I can do on a calculator in 10 steps can require 1000 steps in code. Problem is in code things interact and are constantly changing, so the numbers are dynamic and in real-time. Whereas in person, you are just dealing with static numbers in one instance of time.

An IRIS functions on a curve as a response over time, could you theoretically pass data linearly to it in a bunch of conditions and spit out a positional number, yes you can do it that way. As a matter of fact, I'm sure many of the failed IRIS code tried to do it this "easy" way. The problem becomes you're conditions start colliding and the code becomes impossible to debug and tweak, because instead of having a consistent numbering and response system, you just have a conditional mess loosely connected together that starts producing unpredictable results. You add a condition to fix one bug, and that condition affects your other condition, then you end up with an AI problem bordering on Chaos Theory (no thank you) and end up with the butterfly effect.

The proper way is to use curves based on change (calc derivatives) and then modify those with tables of polynomials or variables that correct things to maintain a certain preferred response. Once you have your generic response curves setup, you can then tweak those with additional conditions at the extremes or to handle abnormal response.

Here is an example of something that would seem to be 10 lines of code, that ends up as 1000...

This code below does NOTHING else but convert Metric to English and back and forth, nothing else. Should only be 10 lines of code, right?
I wish...

IRIS code would be much much worse than this. This isn't even 2% of my math code from the calculator, and my calculator's code isn't as complicated as designing an IRIS would be. This is just a snippet.



//Converts 3.51m to a label like 12 ft 8 in
function convert_M_TO_FTin(meters) {
var meterPos = meters.indexOf(" m");

m = meters.substring(0, meterPos);
var inches = (inPercm * 100) * m;
var ft = Math.floor(inches / 12, 0);
inches = inches % 12;
inches = inches.toFixed(2);
inches = Math.round(inches)


return ft + " ft " + inches + " in";
}

//Converts 3 m 51 cm to a label like 12 ft 8 in
function convert_M_cm_TO_FT_in(meters,cm,round) {
cm = parseFloat(cm);
cm = cm + (meters * 100);

var inches = .393701 * cm;
var ft = Math.floor(inches / 12, 0);
inches = inches % 12;
inches = inches.toFixed(2);

if (round == true) {
inches = Math.round(inches);
inches = inches.toFixed(1);
inches = Math.round(inches);
}

if (inches == 12) {
inches = 0; ft += 1;
}

var ftIn = new Array(ft, inches);
return ftIn;

}

function convert_FT_in_TO_M_cm(FT, inches) {
inches = parseFloat(inches);
inches = inches + (12 * FT);
var cm = inches * cmPerInch;
var m = Math.floor(cm / 100, 0);
cm = cm % 100;
cm = cm.toFixed(1);
cm = Math.round(cm);

var MetersCM = new Array(m, cm);
return MetersCM;

}

//Converts a full label like "12 ft 8 in" to "3.51 m"
function convert_ftin_TO_m(strFtIn) {
var ftPos = strFtIn.indexOf(" ft"); //because we already replaced ft to m
var ft = strFtIn.substring(0, ftPos);
var inches = parseFloat(strFtIn.substring(ftPos + 3, ftPos + 6));
inches = parseFloat(inches) + (parseFloat(ft) * 12);
var meters = ((cmPerInch * inches) / 100).toFixed(2);

return meters;
}

//Changes a label from FT in to M cm
function ChangeLabelTo_M_cm(label) {
label = label.replace(" ft", " m");
label = label.replace(" in", " cm");
return label;
}
function ChangeLabelTo_Ft_in(label) {
label = label.replace(" m", " ft");
label = label.replace(" cm", " in");
return label;
}

function ConvertALL_M_Toftin(ShouldRound) {

var meters; var cm; var ftFloor; var ftIn = new Array(0, 0);

meters = $("tbCeilingHeightFeet").value; cm = $("tbCeilingHeightInches").value;
ftIn = convert_M_cm_TO_FT_in(meters, cm, ShouldRound);
$("tbCeilingHeightFeet").value = ftIn[0]; $("tbCeilingHeightInches").value = ftIn[1];
$("divCeilingHeight").innerHTML = ChangeLabelTo_Ft_in($("divCeilingHeight").innerHTML);

meters = $("tbFeet_FloorToScreen").value; cm = $("tbInches_FloorToScreen").value;
ftIn = convert_M_cm_TO_FT_in(meters, cm, ShouldRound);
$("tbFeet_FloorToScreen").value = ftIn[0]; $("tbInches_FloorToScreen").value = ftIn[1];
$("divScreenToFloor").innerHTML = ChangeLabelTo_Ft_in($("divScreenToFloor").innerHTML);

$("divRightLabel").innerHTML = ChangeLabelTo_Ft_in($("divRightLabel").innerHTML);
$("tbLensDistance").value = (inPercm * $("tbLensDistance").value).toFixed(2);
$("divMount").innerHTML = $("divMount").innerHTML.replace(" cm", " in");
$("tbMountPoleLength").value = (inPercm * $("tbMountPoleLength").value).toFixed(0);

//Convert the Markers
$("divTopMarker").innerHTML = convertMarkers_ToFTin($("divTopMarker").innerHTML);
$("divCenterMarker").innerHTML = convertMarkers_ToFTin($("divCenterMarker").innerHTML);
$("divBottomMarker").innerHTML = convertMarkers_ToFTin($("divBottomMarker").innerHTML);

//Convert txtMinCeilingHeight to Feet/Inches
var txtMeters = $("txtMinCeilingHeight").value;
var txtFtIn = convert_M_TO_FTin(txtMeters);
$("txtMinCeilingHeight").value = txtFtIn;


$("divMountPoleLength").innerHTML = ChangeLabelTo_Ft_in($("divMountPoleLength").innerHTML);
}

function ConvertALL_ftin_To_M() {
var m; var cm; var mFloor; var cmFloor; var MetersCM = new Array(0, 0);

MetersCM = convert_FT_in_TO_M_cm($("tbCeilingHeightFeet").value, $("tbCeilingHeightInches").value);
$("tbCeilingHeightFeet").value = MetersCM[0];
$("tbCeilingHeightInches").value = MetersCM[1];
$("divCeilingHeight").innerHTML = ChangeLabelTo_M_cm($("divCeilingHeight").innerHTML);

MetersCM = convert_FT_in_TO_M_cm($("tbFeet_FloorToScreen").value, $("tbInches_FloorToScreen").value);
$("tbFeet_FloorToScreen").value = MetersCM[0];
$("tbInches_FloorToScreen").value = MetersCM[1];
$("divScreenToFloor").innerHTML = ChangeLabelTo_M_cm($("divScreenToFloor").innerHTML);

$("divRightLabel").innerHTML = ChangeLabelTo_M_cm($("divRightLabel").innerHTML);
$("tbLensDistance").value = (cmPerInch * $("tbLensDistance").value).toFixed(2);
$("divMount").innerHTML = ChangeLabelTo_M_cm($("divMount").innerHTML);
$("tbMountPoleLength").value = (cmPerInch * $("tbMountPoleLength").value).toFixed(0);
$("divMountPoleLength").innerHTML = $("divMountPoleLength").innerHTML.replace(" in", " cm");

//Convert the Markers
$("divTopMarker").innerHTML = convertMarkers_ToMeters($("divTopMarker").innerHTML);
$("divCenterMarker").innerHTML = convertMarkers_ToMeters($("divCenterMarker").innerHTML);
$("divBottomMarker").innerHTML = convertMarkers_ToMeters($("divBottomMarker").innerHTML);

//Parse Ft/in from txtMinCeilingHeight
m = convert_ftin_TO_m($("txtMinCeilingHeight").value);
$("txtMinCeilingHeight").value = m + " m";

//After conversion to meters finishes, we need to correct the Screen Markers and the Req Ceiling Height
forMeters_IfMismatch_CorrectMarkersAndReqHeight();
}


In some cases, yes you could have very simply converted the numbers, but in most cases with math coding there are additional conditions you have to deal with.


Point being, it's never as easy as it sounds...



To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
for both one projector or dual stacks

Web Calculator v023 & v025
-
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

**Current Projector Calculator** --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


Coder's Top Projector Picks of 2012 --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

coderguy is offline  
post #1088 of 4136 Old 10-21-2013, 05:46 AM
AVS Special Member
 
coderguy's Avatar
 
Join Date: Dec 2006
Posts: 9,984
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 147
Quote:
Originally Posted by Highjinx View Post

It would be impossible for any manufacture to be precise for every frame out there in movie land. There are very few if any successive frames where bright would go to dark and then bright again.

Building an IRIS is really just about maintaining a consistent response curve that does not go "out of bounds" to the eye. The end goal is to be as aggressive as possible while maintaining an acceptable level of invisibility.

Although it is not common for the entire scene to go dark to light and back to dark, it is very common for a portion of the screen to do so. An explosion in space that fades slowly is a perfect example of what makes an IRIS hard to program. Another example is the headlights of a car or space ship coming into view and then going back out of view.



To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
for both one projector or dual stacks

Web Calculator v023 & v025
-
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

**Current Projector Calculator** --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


Coder's Top Projector Picks of 2012 --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

coderguy is offline  
post #1089 of 4136 Old 10-21-2013, 06:06 AM
AVS Special Member
 
coderguy's Avatar
 
Join Date: Dec 2006
Posts: 9,984
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 147
Quote:
Originally Posted by noah katz View Post

Unless they totally overengineered it (not likely), their current iris mechanism is likely totally unsuitable in response time and durability.
Why would they struggle to develop formulae for something as complex as psychovisual requirements when they could just use lookup tables?

Go watch a Benq Iris and you'll see why not...

You do use lookup tables of course as you do with almost any code, but you have to do a lot more than "just use a lookup table", or a single hash or array... That said there are different ways to attempt this problem, just like in any program there are 20 ways to skin a cat.

Without me having the ability to question the programmers from Sony that built a good Iris, I would have to assume the following if I were attempting it myself:

The assumption would be that a proper Iris has to respond based on a pattern as it changes as a function over time (the derivative curve), using a combination of average APL, peak levels, minimum levels, and transitional levels (previous scene to current). At 24fps, a frame changes every 41.7 milliseconds and most devices have lag just above that or less, so accounting for future frames is unlikely to be possible with current standards, unless they introduced a pre-load / buffered standard so that our devices were always a few seconds behind (wouldn't work in gaming though anyhow).

Those are the main variables I know of just thinking off the top of my head, there may be a few more.



To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
for both one projector or dual stacks

Web Calculator v023 & v025
-
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

**Current Projector Calculator** --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


Coder's Top Projector Picks of 2012 --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

coderguy is offline  
post #1090 of 4136 Old 10-21-2013, 07:01 AM
AVS Addicted Member
 
mark haflich's Avatar
 
Join Date: Dec 2000
Location: brookeville, maryland, usa
Posts: 19,933
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 443 Post(s)
Liked: 488
Quote:
Originally Posted by Highjinx View Post

The more I think about this the simpler it seems to implement, lens irises have been around for decades, adding a motorised control is fairly new, but nothing that cannot be done fairly simply without patent issues.

How difficuilt would it be to have say a 100 slides of captures of different scenes with dissimilar 'Picture Levels', the engineers have a group of golden eyes doing the observing, while aperture size and gama manupilation is performed for each of the slides to acheive best image quality.

This data becomes the foundation for aperture size control/gama modification for the various picture levels via the software.


Most lens irises are now set at the camera and not by twisting a ring on the lens. Edge wound iris motors are very common. How do you think a projector that has a lens iris but not one that it operates dynamically changes the iris opening when you go into the menu and change the setting? It sends a signal to the iris motor. With a DI, the projector is doing this potentially scene dependent very frequently. Nothing need have changed, the projector is telling the iris to move and it moves very quickly. Just like it does in a camera. You view the scene with the iris wide open, when you press the shutter button, the iris goes to where you set it or the auto program set it, the shutter opens and closes and the iris reopens. This technology has been around for yers and patent terms of 17 years have long run. The trick is telling the iris what to do and adjusting the gamma dynamically. Its science but also part art. And the art part is hard to grasp by the code guys.

Mark Haflich

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

call me at: 240 876 2536
mark haflich is offline  
post #1091 of 4136 Old 10-21-2013, 07:18 AM
AVS Special Member
 
coderguy's Avatar
 
Join Date: Dec 2006
Posts: 9,984
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 147
Quote:
Originally Posted by mark haflich View Post

The trick is telling the iris what to do and adjusting the gamma dynamically. Its science but also part art. And the art part is hard to grasp by the code guys.

Right, the basic curve is pure science, I mean you know you have to stay within a certain boundary limit and that the curve will generally start here and end here over time. The problem becomes the conditional polynomial equations that you use to "correct" or "intensify" the IRIS response, this is primarily where the art comes into play. Stacking complex conditions as a math solution is absolutely a messy nightmare for something like Iris code, which is one reason I assume many of these companies just give up entirely and say forget it, this is good enough. The fewer the conditions you end up with in the end, the better, you want your initial curves to handle most of the changes with as few corrective conditional adjustments as possible, though there will be some.



To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
for both one projector or dual stacks

Web Calculator v023 & v025
-
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

**Current Projector Calculator** --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


Coder's Top Projector Picks of 2012 --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

coderguy is offline  
post #1092 of 4136 Old 10-21-2013, 08:49 AM
AVS Addicted Member
 
mark haflich's Avatar
 
Join Date: Dec 2000
Location: brookeville, maryland, usa
Posts: 19,933
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 443 Post(s)
Liked: 488
And that is why any good DI Implementation requires intelligence.

Mark Haflich

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

call me at: 240 876 2536
mark haflich is offline  
post #1093 of 4136 Old 10-21-2013, 09:01 AM
AVS Special Member
 
coderguy's Avatar
 
Join Date: Dec 2006
Posts: 9,984
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 147
It is an AI problem to where the trick is to use the least amount of AI as possible (hah it's a contradiction), but that's almost always the trick of programming good AI (don't use more AI than you need).

That said, most of the Irises reaction timings would probably be be based upon some simple data of gamma and APL, but it's conditioning the Iris to react in ways that are not within the norm of a single deterministic system that cause it to be an AI problem, it's getting it to work in that last 20% of the time.



To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
for both one projector or dual stacks

Web Calculator v023 & v025
-
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

**Current Projector Calculator** --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


Coder's Top Projector Picks of 2012 --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

coderguy is offline  
post #1094 of 4136 Old 10-21-2013, 09:15 AM - Thread Starter
Senior Member
 
Thebes's Avatar
 
Join Date: Jan 2007
Posts: 447
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 22
Thebes is offline  
post #1095 of 4136 Old 10-21-2013, 09:26 AM
AVS Addicted Member
 
mark haflich's Avatar
 
Join Date: Dec 2000
Location: brookeville, maryland, usa
Posts: 19,933
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 443 Post(s)
Liked: 488
Quote:
Originally Posted by coderguy View Post

It is an AI problem to where the trick is to use the least amount of AI as possible (hah it's a contradiction), but that's almost always the trick of programming good AI (don't use more AI than you need).

That said, most of the Irises reaction timings would probably be be based upon some simple data of gamma and APL, but it's conditioning the Iris to react in ways that are not within the norm of a single deterministic system that cause it to be an AI problem, it's getting it to work in that last 20% of the time.

Conditioning is a strange word to use here. Its not like toilet training a dog to go outside. Its all programing in response to a given stimulus input. Nikon used fuzzy logic years ago to determine what expose would work best for a certain image. It stored about 10K images in a memory bank with each coded to a certain iris setting which then varied by shutter speed chosen. The camera matched the image being shot to the closest one in its memory.

Mark Haflich

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

call me at: 240 876 2536
mark haflich is offline  
post #1096 of 4136 Old 10-21-2013, 09:34 AM
AVS Special Member
 
coderguy's Avatar
 
Join Date: Dec 2006
Posts: 9,984
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 147
Conditioning in AI is often used as a general term which means to adjust / fix / add / subtract variables and then normalize the test before checking the (in this case) perception again.

It doesn't necessarily mean to condition a learning AI system, the Iris is not learning but the conditions are too many to be completely deterministic or deciphered by pure code logic from the beginning. What that means is that even the brightest minds are unlikely to be able to know the exact results of how their code affects the image at first, they have to train themselves on how to continue to fix the code by visually observing the results of their current code. This is done by conditioning.

If I were to guess, it is likely about an 80/20 deterministic to non-deterministic problem. Any problem that has initial unknown or unexpected results due to there being too many random possibilities (in this case APL and scene transitions), means the problem will usually require some conditioning.



To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
for both one projector or dual stacks

Web Calculator v023 & v025
-
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

**Current Projector Calculator** --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


Coder's Top Projector Picks of 2012 --
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

coderguy is offline  
post #1097 of 4136 Old 10-21-2013, 09:36 AM
AVS Addicted Member
 
R Harkness's Avatar
 
Join Date: May 2001
Location: Toronto, Ontario, Canada
Posts: 11,995
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 110 Post(s)
Liked: 329
Quote:
Originally Posted by Thebes View Post

Hi,

two previews after a demonstration in U.K. :

http://www.trustedreviews.com/jvc-dla-x700r_Projector_review

http://www.hdtvtest.co.uk/news/jvc-x700-201310213386.htm

Thank you.

Looks quite promising (for those not concerned with native 4K content as of yet).

Rich H


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.



To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
R Harkness is offline  
post #1098 of 4136 Old 10-21-2013, 10:01 AM
AVS Addicted Member
 
mark haflich's Avatar
 
Join Date: Dec 2000
Location: brookeville, maryland, usa
Posts: 19,933
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 443 Post(s)
Liked: 488
Show reports but a lot higher quality than the reporting done here from Cedia. Interesting that JVC now provides for xyYcc color space. That's my prediction for the 4K color space we are going to get. I just wish JVC adds HDCP 2.2 to this year's projectors. If not, they may become boat anchors with regard to showing 4K content from other than a Redray and then provided that the content provided the Redray from Odemax is not HDCP 2.2 protected. Most indy stuff won't be so protected relying instead on the inherent Red protection.

Mark Haflich

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

call me at: 240 876 2536
mark haflich is offline  
post #1099 of 4136 Old 10-21-2013, 10:53 AM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,183
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 57
Quote:
Originally Posted by Seegs108 View Post

It needs to look at several frames in advanced to get an average picture level of the "scene" thats about to be projected.
Although I'm sure that would help, I don't know of any projector actually doing that. One of the challenges is making a good DI without looking ahead several frames. A system could be designed to look ahead and delay the video while doing so, but would need something for audio sync.

I realize that commercial cinemas can have more problems with other lighting and so be limited for system on/off CR, but I've thought in the past that if they wanted to do a dynamic iris system they could look ahead either in real time, pre-look ahead for the whole movie they are going to show, or even have the information encoded with the movie to indicate the best current iris position.
Quote:
Originally Posted by Highjinx View Post

There are very few if any successive frames where bright would go to dark and then bright again.
There may not be a lot, but if a DI design team decides to not care about whether their's works well in some of those cases they will likely end up with an inferior implementation.

A friend showed me a scene where a Sony from a few years ago did a very poor job. I believe it was from Sex and the City. There were 2 characters in a room talking to each other with the camera going back and forth. One of the characters had a light behind them off to the side and one did not. The Sony iris would delay and then react to the changing peak level in the frames and this gave an annoying pumping effect to our eyes.

Changing iris position quickly and changing slowly each have their challenges.

If the human eye was as limited as many accepted as experts in this industry claim then I think creating a DI system that is good for normal human vision would be a lot easier than it actually is.

On another note, I think that the fact that DIs have negative side effects compared to native on/off CR and yet most people prefer images with a good one enabled shows how much on/off CR matters, contrary to what many of these experts have professed for years.

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is online now  
post #1100 of 4136 Old 10-21-2013, 12:25 PM
AVS Addicted Member
 
noah katz's Avatar
 
Join Date: Apr 1999
Location: Mountain View, CA USA
Posts: 20,532
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 229 Post(s)
Liked: 156
Thanks, Thebes.

I'm encouraged from these reports that JVC has made real strides in PQ from both the new chips and the II.
Quote:
Originally Posted by Thebes View Post

Hi,

two previews after a demonstration in U.K. :

http://www.trustedreviews.com/jvc-dla-x700r_Projector_review

http://www.hdtvtest.co.uk/news/jvc-x700-201310213386.htm

Noah
noah katz is offline  
post #1101 of 4136 Old 10-21-2013, 02:59 PM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,183
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 57
Quote:
Originally Posted by Thebes View Post

two previews after a demonstration in U.K. :

http://www.trustedreviews.com/jvc-dla-x700r_Projector_review
I hope somebody can get clarification about the part of this article that implies JVC is taking 4K frames and downscaling them to single 1080p frames before applying e-shift. I'm talking about his part:
Quote:
However, e-Shift can currently only work with 1920 x 1080 signals. So even though JVC has enabled the HDMI inputs of all three of its new projectors to receive native 4K sources – even up to 60p in 8-bit, with 4:2:0 colour sampling – these 4K source images have to be converted into 1,920 x 1,080 images before then going through the e-shift 3 process to have their 4K levels of pixel density restored.

...

In fact, so far as we can tell from our questioning, JVC isn’t even claiming any really significant picture benefit from using a 4K source versus a 1080p one; rather the support for 4K inputs in the new projectors seems to be there as a futureproof device, enabling the projectors to play future types of 4K content when they arrive.
We know that they don't deliver pixel-for-pixel 4K, but there is a difference between extracting two 1080p images per 4K frame and extracting only one 1080p image per 4K frame, which is what I think this article implies (otherwise the reason for supporting 4K input should be clear).

How significant the difference is can be argued and tested, but I thought that JVC was clear that they would extract two 1080p images per 4K frame and so would not be the same as downscaling the 4K source to 1080p outside the projector.

I wonder if the writer confused extracting two 1080p images per frame with downscaling to just one 1080p image per frame and applying e-shift, or if JVC really is only extracting one 1080p image per frame (which is basic downscaling).

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is online now  
post #1102 of 4136 Old 10-21-2013, 03:04 PM
AVS Addicted Member
 
mark haflich's Avatar
 
Join Date: Dec 2000
Location: brookeville, maryland, usa
Posts: 19,933
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 443 Post(s)
Liked: 488
Its two different 1080p frames. I suspect these are processed rather than just 2 x 1080p pixels out of the 4 times 1080p pixels..

Mark Haflich

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

call me at: 240 876 2536
mark haflich is offline  
post #1103 of 4136 Old 10-21-2013, 03:10 PM
AVS Special Member
 
Geof's Avatar
 
Join Date: Feb 1999
Location: Eden NY
Posts: 6,012
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 5 Post(s)
Liked: 33
Quote:
Originally Posted by darinp2 View Post

I hope somebody can get clarification about the part of this article that implies JVC is taking 4K frames and downscaling them to single 1080p frames before applying e-shift. I'm talking about his part:
We know that they don't deliver pixel-for-pixel 4K, but there is a difference between extracting two 1080p images per 4K frame and extracting only one 1080p image per 4K frame, which is what I think this article implies (otherwise the reason for supporting 4K input should be clear).

How significant the difference is can be argued and tested, but I thought that JVC was clear that they would extract two 1080p images per 4K frame and so would not be the same as downscaling the 4K source to 1080p outside the projector.

I wonder if the writer confused extracting two 1080p images per frame with downscaling to just one 1080p image per frame and applying e-shift, or if JVC really is only extracting one 1080p image per frame (which is basic downscaling).

--Darin
I did a double take on that wording as well. In the end I think he's correct but it is confusing.

The eShift process takes 2K and upscales it to 4K (or on this years models one can input 4K). The next step is converting a 4K frame into two 1080 sub-frames. The process that JVC uses to convert a single 4K frame into two 1080p frames is unknown (Madshi objected to my using the term "down-sampled" tho) but somehow JVC is making two 1080 sub-frames from one 4K frame.

Geof
Geof is offline  
post #1104 of 4136 Old 10-21-2013, 03:22 PM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,183
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 57
Quote:
Originally Posted by Geof View Post

In the end I think he's correct but it is confusing.

... but somehow JVC is making two 1080 sub-frames from one 4K frame.
If they are making two 1080p sub-frames from one 4K frame without first making a single 1080p sub-frame that the second one is built from, then it seems to me he is not correct. At least not in his implication that this isn't different than inputting 1080p.

There is definitely a difference between going to a single 1080p frame and then doing all the work from that (no longer using any of the original 4K information) and using the 4K to directly make two single 1080p frames (or sub-frames if you want to call them that). I believe the article implies the former and if the JVC is doing the latter, then at the very least the implication of the article is wrong.

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is online now  
post #1105 of 4136 Old 10-21-2013, 03:30 PM
AVS Special Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 7,108
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 219 Post(s)
Liked: 356
Interesting that in the UK the new Sony 500 is considerably more competitively priced than it is here in the US. Even the top of the line JVC is a few thousand less here at MSRP (and probably even more at street).

Senior Video Editor
Secrets of Home Theater and High Fidelity


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Kris Deering is offline  
post #1106 of 4136 Old 10-21-2013, 03:34 PM
AVS Special Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 7,108
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 219 Post(s)
Liked: 356
Another interesting bit is the part about the noise being cut down by 2 dB in the low lamp mode. From what I heard from JVC at the show the noise was cut down in high lamp mode, which is where it was needed most. Guess we continue to wait and see!

Senior Video Editor
Secrets of Home Theater and High Fidelity


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Kris Deering is offline  
post #1107 of 4136 Old 10-21-2013, 03:34 PM
AVS Special Member
 
Geof's Avatar
 
Join Date: Feb 1999
Location: Eden NY
Posts: 6,012
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 5 Post(s)
Liked: 33
Quote:
Originally Posted by darinp2 View Post

If they are making two 1080p sub-frames from one 4K frame without first making a single 1080p sub-frame that the second one is built from, then it seems to me he is not correct. At least not in his implication that this isn't different than inputting 1080p.

There is definitely a difference between going to a single 1080p frame and then doing all the work from that (no longer using any of the original 4K information) and using the 4K to directly make two single 1080p frames (or sub-frames if you want to call them that). I believe the article implies the former and if the JVC is doing the latter, then at the very least the implication of the article is wrong.

--Darin

My interpretation of the part you quoted agrees with my understanding of how shift functions....
Quote:
However, e-Shift can currently only work with 1920 x 1080 signals. [this is true because it is the light output from the 1920x1080 panel that is being eShifted] So even though JVC has enabled the HDMI inputs of all three of its new projectors to receive native 4K sources – even up to 60p in 8-bit, with 4:2:0 colour sampling – these 4K source images have to be converted into 1,920 x 1,080 images before then going through the e-shift 3 process to have their 4K levels of pixel density restored. [This is true also because eShift only operates on 1920x1080 images....a 4K image cannot be eShifted]. The tricky part is how those 4K images are converted into two 1920x1080 sub-frames (I believe sub-frames is the correct usage because it takes two to make a pseudo 4K image).

From Cine4Home:

Geof
Geof is offline  
post #1108 of 4136 Old 10-21-2013, 03:38 PM
AVS Addicted Member
 
mark haflich's Avatar
 
Join Date: Dec 2000
Location: brookeville, maryland, usa
Posts: 19,933
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 443 Post(s)
Liked: 488
OK Chris. I can't resist. smile.gif What projector doesn't need noise reduction the most in high lamp rather than low lamp smile.gif Please identify one that is quieter in high lamp mode than in low lamp mode.

Mark Haflich

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

call me at: 240 876 2536
mark haflich is offline  
post #1109 of 4136 Old 10-21-2013, 03:43 PM
AVS Special Member
 
Geof's Avatar
 
Join Date: Feb 1999
Location: Eden NY
Posts: 6,012
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 5 Post(s)
Liked: 33
Quote:
Originally Posted by Kris Deering View Post

Another interesting bit is the part about the noise being cut down by 2 dB in the low lamp mode. From what I heard from JVC at the show the noise was cut down in high lamp mode, which is where it was needed most. Guess we continue to wait and see!
Hopefully noise was lowered in both lamps modes.

Geof
Geof is offline  
post #1110 of 4136 Old 10-21-2013, 03:46 PM
AVS Club Gold
 
AV Science Sales 5's Avatar
 
Join Date: Sep 2011
Location: A beautiful view of a lake
Posts: 7,719
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 362 Post(s)
Liked: 418
Quote:
Originally Posted by Geof View Post

Hopefully noise was lowered in both lamps modes.

Reported earlier that it was cut down in low and high lamp mode. Noise level was very low.
Low 21db
High 28db

Mike Garrett, AV Science Sales Call Me: 585-671-2968
Email Me:
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

Brands we sell: 
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
 
Call for B-stock projectors
Stewart, Seymour, SE, SI, Falcon & many more.
RBH, Martin Logan, Triad, Atlantic Technology, MK Sound, BG Radia, SVS & Def Tech.
AV Science Sales 5 is offline  
Reply Digital Hi-End Projectors - $3,000+ USD MSRP

Tags
Sony Vpl Vw1000es Projector , Jvc Dla X900r , Jvc Dla X700r
Gear in this thread - X900r by PriceGrabber.com

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