Theres an italian in the house! - Page 15

I'll be using mine with the zoom method on two electric screens, just like I do now with my dVision 1080p. And with the short throw T1 lens. I don't expect it to work any different than with the dVision - which works great.

Craig or anyone,

Can you tell if the T3 lens has enough zoom range, when utilizing the zoom method, to go from 16:9 to 2.35 on a 110 inch wide 2.35 AR screen at a 22ft throw?

Did we get when these will start shipping ?

Art

I expect that Lumis units in general will have high MTFs also. I think those units Alan and coldmachine got likely have high MTF at full resolution and close to full resolution. But basically, if somebody got a unit with a focus problem that reduced those I wouldn't conclude that it must also have poor ANSI CR and if somebody happened to get a Lumis that only had 100:1-200:1 ANSI CR I wouldn't conclude that the MTFs at full resolution or close to full resolution must be poor on that unit.

BTW: I got some measurements for the center 4 squares of an ANSI CR checkerboard from a 2nd Samsung A800B and this one got about 350:1. I don't know if I'll get a chance to compare this one to the A800B that measured closer to 130:1 for those, but hope that I will as I think that could be interesting.

--Darin

Couldn't one use electronic vertical shift to raise the 2.35 image so that it is top aligned with the screen, assuming the Lumis has one?

I have a Lumagen XD which I could use if the Lumis does not electronic vertical shift. Are there drawbacks to using electronic vertical shift?

CM. I'd welcome a summary of your views on zooming vs. anamorphic lens.

Also regarding your comments about the difficulty of proper implementation of motorized vertical lens shift, does that apply to a manual lens shift if you put it in one position and leave it there?

Currently my throw is 25' onto an 11'6" 1.3 Studiotek. I zoom between 2.35 (11'6") and 1.78 (about 9'6"). I have sealed soffit with fan. How deep is the Lumis? Are its connections on the side or rear?

Agree on stretch being an option, especially if you have to "stretch" to get a Lumis. I have a Panasonic 3000 in one of my setups, and it's fantastically convenient to have one button poor man's cinemascope. I'm really surprised this is not showing up on more higher end units when such an affordable projector as the 3000 can implement this nifty feature. Is it perfect? no but it gets focus pretty close on scope, close enough to not bother with having to refocus and on switching back to native is dead to nuts focused on 16:9.

Sounds like the Lumis has manual shift so wouldn't be possible to have an automated lens memory system anyways. Guess if you can afford one of these Lumins you also can afford an Isco or Schneider, especially now with the newer info that ANSI does not suffer as much as we originally believed when using an anamorphic lens.

The cinem4home work indicted that ii is in fact negligible.

Art

i give up to tell about the lost of ansi cr. when use such a lens in use.
seams it is senseless.

"No, thankfully, it [focus/zoom memories] doesn't offer this facility...The cost of doing this correctly is not inconsiderable and requires a very sophisticated focusing arrangement."

I haven't been following it, but early reports on the Pan 3000, including from Cine4home, said that it worked fine.

For me, that's the problem. It must be perfect or I wouldn't want it. The units Ive seen with servo zooms do it incredibly well.


See above. Personally I feel its an all or nothing deal. A $38k machine cannot implement a servo zoom without compromises elsewhere.

Rather than turn the thread into a monkey $hit fight, I think I will just say that I very much prefer the CIH route.

Having said that, if someone throws the first turd, I'm in.

You asked for it CM

Lon, that ANSI is just off the charts!

I previously owned a DPI Highlite and did the "zoom" thing with it. The memories were dead perfect going from 16:9 to 2.35:1. I never had to tweak focus or shift or zoom. This was an expensive projector so I was surprised to see this feature in an affordable unit like the 3000.

who would calibate a lumis, my local central wisconsin guy?

"I previously owned a DPI Highlite and did the "zoom" thing with it. The memories were dead perfect going from 16:9 to 2.35:1. I never had to tweak focus or shift or zoom. This was an expensive projector so I was surprised to see this feature in an affordable unit like the 3000."

It needn't be that expensive, stepper motors and encoders are relatively cheap, especially when you're in the realm of $38k.

Sim2 REQUIRES that their projectors be installed by the dealer (or contracted out and supervised by the dealer). This should include some tweaking, but a full calibration is not required by Sim2 which leaves the option of calibration up to the buyer. (In other words, find someone you trust to calibrate.) There is certainly no shortage of qualified folks here on Forum. AVS will calibrate as well.

(If the question was "sarcasm," my apologies.)

Jim

What a monster projector for the price point and beyond, thanks for filling us in Alan and CM

Darin, yes I believe you are correct. I just ran through some numbers using the MTF = (Imax-Imin)/(Imax+Imin) formula. This is what it looks like:

So basically what this says is if everything were ignored (edit: i.e. perfect) but ANSI, the MTF would be over 98% for even a device with 100:1 ANSI.

From what I've read a good 9" CRT can do maybe 30% MTF @ the 1080p limit (rough ballpark estimate). I think JVC has published ~70% on one of their QXGA1 white papers. I have no idea what to expect a quality DLP to deliver but let's say it's 80%. An 80% MTF would mean that the measured contrast between pixels is going to be 10:1 which is much less than ANSI. In other words ANSI becomes a second order effect because it's so much higher than the low contrasts that would yield typical MTF readings of less than 50:1.

That's a great paper and one of the few I've seen on MTF on the source itself. I read it once before but it's definitely worth re-reading.

Very interesting. Thanks for posting that.

Yes. I was thinking of this in terms of that paper earlier. If you go around 3/4ths of the way down in that paper and look at the first graph that says, "Film Dig. Inter. Sys. Data" and has cycles/mm and horizontal pixels at the bottom, the ANSI CR test is very close to where the graph starts on the left, since I'm pretty sure the horizontal pixels value would be 4 and the cycles/mm would be 0.08. The first mark is cycles per mm of 2, so this is so close to the left side of the graph that it would look like it was on the vertical line for zero cycles/mm. And the Lumis would be around .998 there, while 100:1 ANSI CR would be .98 there, like you said. Graphically those would both look like they started pretty close to where the green lines meets the left axis in the graph. The question is partially how much the difference of .998 from .98 on the left edge of that graph tells us about what happens halfway across at close to 2000 horizontal pixels.

--Darin

Again, you have to keep in mind that MTF is usually normalized to be 1 at the lowest spatial frequency. The MTF of very low spatial frequencies is not all that meaningful because they are all going to be very close to 1. For very low frequencies, the MTF data before normalization is what should be looked at.

Simple descriptions of MTF have been posted before, but I think that it would be useful to do it again. MTF is nothing more than the contrast of detail in the image. It's a curve, starting with the relative contrast of the coarsest detail and going to the contrast of the finest detail that the system can resolve. Why is it relevant to a discussion of the Lumis? It is because it may be the aspect of performance where projectors like the Christie HD6K-M and the Barco DP-1500 have an advantage.

Earlier you said:

Do you consider 100:1 ANSI CR to be poor, good, ...? If not poor, what would you consider poor ANSI CR? If your MTF data for a number of pixels of 4 (the ANSI CR test) is that the 2 white boxes horizontally are each 10 ft-lamberts and the video black boxes are 0.1 ft-lamberts, what does that tell you about MTF at high frequencies? I don't see how this normalization affects the discussion of whether what would be considered poor ANSI CR means that MTF will be poor at all frequencies.

What kind of MTF would you consider poor or good at full resolution (1920 pixels horizontally) or for a 2 pixel by 2 pixel checkerboard pattern?

Nobody is disagreeing that the MTF curve is relevant. The question is whether a point that is at an extremely low frequency (ANSI CR) being at say .98 or .99 tells you that the MTF must be poor at high frequencies. It is not the case with MTF where the room is included, as I mentioned earlier.

--Darin

Alan G,

It looks like you are keeping the Lumis. I see your 11s2 is for sale. Congrats.

How do you plan to run 2.35 to 1 with the Lumis? Are you going to zoom or use an ISCOIII with sled?

You mentioned earlier you liked zooming this pj better than the rs20.

my demo unit was ship out from sim2 and should arrive me tomorror if
nothing went wrong.

this unit have some changes inside compare to the 2 units that alan and cm
had but this unit will be also not the final version but very close.

in the best case you can read tomorror at this time some infos.

Darin,

I consider projector ANSI CR of 100 very poor. I consider 500 as good and anything over 750 as very good. I think that it's better to leave the room out for this and consider only the projector.

Others will differ, but the threshold of acceptable MTF for me at the highest frequencies is 0.5-0.6. The exception would be at the limiting resolution away from the center, where even 1/4 pixel misconvergence can easily bring it lower. I would want it to be at least 0.5-0.6 at the center. I think that the best DLP designs are close to 0.8 MTF at 1920x1080 at the center of the image.

Again the MTF of the lowest frequencies is 1 or close to 1, by definition, because of the normalization of the lowest frequency MTF to 1. (There are some other MTF formulas that can result in an MTF greater than 1, but I can't remember what they are.) Again, if you take away the normalization factor, the measurement of ANSI CR is almost the same as the MTF measurement of the corresponding spatial frequency using a square wave. My best guess is that 100 ANSI CR would correspond to 0.6-0.7 MTF without normalization, which would be poor performance at that frequency. (edit: I am leaving this last sentence, but it's clearly wrong. The MTF is about 0.99, and although that seems good on a scale of 0-1, it's not at that low a frequency.)

In any case, I don't know what we are arguing about now. I have acknowledged that using ANSI CR as a proxy for MTF performance breaks down at some frequencies if you have confounding factors like misconvergence or anything else that will significantly change the MTF at that frequency but not significantly change ANSI CR. I think that the correlation is still good for most projectors at most frequencies.

You lost me there. How would the case like the 10 ftL/0.1 fL (100:1 ANSI CR) correspond to 0.1-0.2 MTF without normalization? I'm not following your math.

I think we are just discussing it.

I think it can also break down in the other direction with things that significantly change ANSI CR, but do not significantly change MTF at high frequency. For instance, when moving a projector so that its lens shift has it centered on the screen lowers ANSI CR (probably because of reflections going straight back into the light engine instead of at a sharper angle) that likely has little affect on MTF at high frequencies. And as I mentioned, it won't surprise me if that Samsung A800B with just ~130:1 for the center 4 boxes still has good MTF at high frequencies. It depends on what the cause of that lower ANSI CR is. If poor ANSI CR will always correlate well with poor MTF at all frequencies then I should be able to say that this A800B must have poor MTF at all frequencies, but the math doesn't support that if MTF of even .7 is considered good at full resolution. And the logic doesn't support that it must have MTF lower than even .7 at full resolution because of this low ANSI CR issue either, IMO.

I think it is also pretty good. But that it has break downs in both directions. I think certain reflections inside the projector that take a long path (like lens shift centered, light reflects back into the engine, then some of it makes it back out through the lens, or internal reflections from one side of the image to the other) are not necessarilly going to have enough local effect to lower the MTF at full resolution to where you would consider it poor even if they did lower ANSI CR to where you would consider it poor, based on the numbers you gave.

--Darin