CurtPalme.com Home Theater Forum

[cbe317]

[deronmoped]

Guys, it's not that difficult. Right click on your screen and bring up NVKeystone Adjustment, this is a example of what I'm talking about. It's a software driven geometry adjustment. All that is needed is it to be done to three different colors separately. I would like to have more then just Keystone adjustment though.

Deron.

[macgyver655]

Since you seem persistant on this subject, here is some reading material for you, BUT, let me point out that these are not CRT projectors, which is a totally different concept. Non CRT based devices dont not require convergence. This is the critical issue for what you are proposeing. This is just a primer for your theory. If after you read it you want more reasons why this would NOT work on CRT based devices I would be glad to inform you. Happy Reading!!!!!


http://www.cs.princeton.edu/omnimedia/papers/DeskAlign.pdf

[deronmoped]

OK

Here is something very interesting, it goes to show you what can be done with software.

http://www.aurigaimaging.com/registar_details.html


"Small images can be stitched together into a mosaic, photos can be stacked to improve image quality, or high-resolution tricolor images can be formed".

Deron.

[kebowers]

Possible with true digital signals and complex digital circuitry and measuring systems, not feasible post manufacture.

[deronmoped]

I wonder what our spy satellites use to take pictures. I can imagine they have like 10 super high resolution cameras that will take pictures of the same object at different angles. They then just stack all the images together to form one extremely high resolution 3D image.

Deron.

[kebowers]

I am afraid the public has been 'brainwashed' by science fiction movies portraying exotic and fantastic capability of US military satellite imaging. Truth is the workhorse is a 25 year old design that relies on high precision optics. BUT, the resolution is in the order of meters, not CM.

Low altitude 'spy satellites' have very short lifetimes and are not very adaptable in coverage--very difficult to change the orbit to cover other territory.

The US program to invent and implement 'super spy' capability disintegrated into babble when the prime contract was awarded to Boeing, who had never designed a satellite. The real 'experts' in the field were not included at all. After several years and many BILLIONS of wasted $, the program was killed recently.

Result: Cold war era capability, failing old satellites, limited ability to respond to new areas, very limited imaging analytical ability. In fact, the US military has relied very heavily on COMMERCIAL imaging companies for much of the middle east, Iraq, and Iran.

This debacle in capability is what happens when arrogant managers with no true knowledge of the subject impose 'academic exercise' approaches on the real scientists and engineers. Rumsfield (GWB and his team of hacks) has severely damaged the US capability to see what is happening.

[kebowers]

For more in depth discussion of 'convergence through software', one should peruse US patents on the subject. There is a fairly large body of work already out there (owned by the big manufacturers and a few specialized firms).

The critical fact is what can be implemented post manufacture (in the field so to speak). Near perfect convergence in the field CAN be achieved automatically, but only at very high cost and complexity.

[macgyver655]

[kebowers]

When the electron beam leaves the emitter, it goes straight unless influenced by magnetic fields. The horizontal and vertical sweep circuits do that to cause the beam to 'paint' the entire frame. Coincident with the beam sweeping, the intensity is modulated to result in phosphor emissions (brightness) for each portion of the picture frame.
The frequency with which the beam intensity can be modulated determines the maximum number of point on each horizontal scanning line that can be resolved. The capability of the vertical sweep circuits determines maximum number of frames per second, while horizontal sweep circuit capability determines maximum number of horizontal lines that can be swept each frame.

Achieving 'convergence' by additional modulation of the vertical and horizontal sweep circuits is problematic. The 'convergence correction' modulation frequency would have to be high enough to correctly 'place' each point (or group of points) on each horizontal scan line (which also has to be correctly placed vertically). This is a non-trivial requirement and substantially beyond the frequency response of the horizontal and vertical sweep coils.

[deronmoped]

What does all that gobbledygook have to do with what is in the source material before it reaches the PJ. The PJ just paints what it gets from the source. If the source says there is a red, green and blue dot on the screen, the PJ will paint a red, green and blue dots on the screen. Now if the source says there is a white dot the PJ will output a red, green and blue dot to make the white dot. The PJ does not care if the convergence is done before it gets to it.

Here is another way to look at it. Set up a PJ with the cross pattern so the vertical lines of the red, green and blue are right next to each other. Now bring up a image from you computer with something like Photo Shop running. What you do next is paint three vertical lines in your Photo Shop program, only this time you swap the red and blue, so you have blue green and red. What you now see on the screen is a converged white line, even thought the convergence of the PJ is off. Convergence done through software.

Deron.

[David_Web]

Yeah but the accuracy sucks!
Why do you thing people spend countless hours tweaking the convergence? Because mediocre is not good enough.
It is fully possible to do and is not that hard but not really useful for a HT or other critical viewing.

@kebowers: You could only converge the image horizontally by changing the timings. If you would like to converge it vertically you would have to move the whole line and that would look bad. Still it's hard to see why you would want to.

[deronmoped]

Problem is you can tweak the PJ for hours and still be unable to get a perfectly converged PJ. Not all PJ's are totally correctable. It would be really cool if there was a drift function built into the program, perfect convergence on startup. This might even be a solution for digital PJ's that come from the factory off by a smigin.

If I knew how to write software, I would look into it myself, with the billions of displays out there I bet I would have no problem getting rich

Deron.

[kebowers]

The complexity of manipulating the 'signal' or 'source' BEFORE routing it to the PJ requires a complete 'map' of the corrections required and then applying those to each frame , line, pixel transmitted. Doing so with analogue signal is not achievable in my opinion.

A DIGITAL signal is another ball of wax. Changing the digital location values for a pixel is much more straightforward than trying to figure out the equivalent location data(and then modify them on the fly before it reaches the yoke)in an analogue signal