If torus is so good, why dont we all at least curve our screens?

Kam,


Designing a torus really isn't all that difficult. Only thing that I don't know are the gain curves (i.e. how much light is reflected to a spot at 7 degrees off center, 15 degrees, etc.). If I knew those or knew of references with formulae, designing one is simply nasty iterative trigonometry, which is what computers are for, ;-))


I guess Cineramax decided to use a vacuum to do his. If I were to tackle the project I would machine it. $100 worth of wood for the rig would be sufficient to do it. In fact machining one has a couple of major advantages over using a vacuum setup.


Anyway, its an interesting mathematical challenge. The process is sitting in my head, with no way in the near future to see it implemented. LOL


Oh well, take care.


Sand

Sand,


I agree completely, I've thought through the calculations a couple of time myself. I don't even know if iterative trig is necessary, perhaps finding the equation relating curvature and angle and taking the derivative will give good results (which is really what you are doing by performing iterative calculations). One set of equations for the vertical and horizontal curvatures should suffice.


In practice, though, there is a lot of actual designing involved. In most cases there are multiple cases to consider for all the seats in the theatre, trying to get a good image for all/most seats could present some challenges (even with the computational power of a computer). Having the gain curve is not enough, one also needs to know the acceptable light fall-off before hotspotting becomes apparent.


Certainly possible for the determined user, but no one has really done it yet!


Regards,


Kam Fung

Kam,

I know some computer people at Carleton who are pretty determined!

Its only a matter of time.
http://www.avsforum.com/ubb/smile.gif Besides didnt mike2 and others come close?

Regards Ron


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

Ron

"Your priorities will be different-its the weighting that counts!"

Kam,


The iterative process is what would take care of optimizing for all seats. If you figure out how to do one calculation, doing 100 and optimizing from those is easy enough. I am a little confused by your statement:


Having the gain curve is not enough, one also needs to know the acceptable light fall-off before hotspotting becomes apparent.


Isn't one of the inherent advantages of the torus that it evens out the light fall-off for the outside positions on the screen? If you do optimize for all seats, than this is taken care of implicitly, no? Perhaps I am not getting exactly what you are trying to say (wouldn't be the first time ;-).


If I had a good formula estimating the gain curve falloff with angle, programming it would be of medium difficulty. Machining would be trial and error, but it could certainly be done, and without terrible difficulty. I think getting a screen on the machined backing would be the more difficult part of the whole thing.


Take care.


Sand


[This message has been edited by Sand (edited 09-13-2001).]

Mike2,


Great web site - I had looked at Stewart before but missed that page.


As far as the machining goes, with the angles involved nothing more complicated than a 1/2" router bit can be used. Using that I figured the machining error would be on the order of .0003" or so. Not much. Whatever error is there can be cleaned up with a very light sanding.


Only question is what material to use to machine into. A lightweight stiff foam type material would be ideal.


This complex shape can be done with the right setup.


Take care.


Sand

Guys,


I have designed a nice app to do the geometrical calculations for torus and flat screens and present the data in an understandable way. The light is ray-traced (10 000 rays at present) from projector to screen (where the gain curve is applied) and back to the viewer. The illumination from the various segments is calculated and the data is presented as a grey scale map. I also extended the algorithm to calculate the color shift for CRT's, where the results are presented as a color map.


If there is enough interest, I will build a GUI and create a compiled version of the Matlab script I am running. I have to warn you I am not very sharp in doing Matlab GUI - yet http://www.avsforum.com/ubb/smile.gif . Still, I need to learn it anyway.


I may also add an algo to calculate the optimal screen shape (not quite a torus in most instances) if I get the time.


All suggestions are welcome!


Oh, and I have to tell you, with the results the app has produced, I have a hard time understanding people with flat high gain screens (FPS is ON!).

Sand,


Stewart has published the angle vs gain graphs for their materials here -
http://www.stewartfilm.com/business_...ssional_2.html


What kind of machining are you referring to?


Thanks,

Mike

Iceman,


that sounds like the application i always wanted to write! Especially the colorshift analysis for CRTs. I won't have time to do anything like that for at least another year, so i would VERY much appreciate you building a gui for it. If you need ideas for additional features or feedback on the GUI, let me know.


Best regards

Bjoern


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

"Never underestimate the predictability of stupidity" (Bullet Tooth Tony in 'Snatch')
My HT in action (Screenshot Page) | My Ultimate 'Edge Enhancement' Guide | My DVD/LD SPL page

Mike,


how exactly do you use your lightmeter to meassure the reflective characteristics of different screen materials.


Thanks


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

"Never underestimate the predictability of stupidity" (Bullet Tooth Tony in 'Snatch')
My HT in action (Screenshot Page) | My Ultimate 'Edge Enhancement' Guide | My DVD/LD SPL page

Sand,

Isn't one of the inherent advantages of the torus that it evens out the light fall-off for the outside positions on the screen? If you do optimize for all seats, than this is taken care of implicitly, no? Perhaps I am not getting exactly what you are trying to say (wouldn't be the first time ;-).


Not quite so simple as that. I believe it is mathematically impossible to have perfect uniformity for *every* seat in the house, one can optimize one seat completely, but not all of them. When you account for the gain curve you will find that you will get less uniformity in the corners because one edge of the screen is at an smaller angle to the viewer than the other edge (i.e. different parts of the gain curve). It them becomes an exercise in minimizing non-uniformity at the extremes of the seating area while still maintaining it in the primary seating areas. Probably possible with iterative calculations, but perhaps faster if with user intervention and a GUI to position the "light cone" for different parts of the screen.


Of course, the question is still: how much non-uniformity is too much? (10%, 20%, more, less?)


Regards,


Kam Fung