If one of the best (affordable) video processors out there had a feature to allow RGB misconvergence adjustments for 1, maybe 2 pixels in horizontal and vertical directions, how many of you would strongly consider purchasing that VP because of this feature?

I posted in the other thread as well, but I will be deleting my post there since it's redudant.

In short, yes, I am very interested in this. It would also be cool if the panels could be digitally "rotated".

Any idea of the MSRP of the processor you mentioned in the other thread if it includes this feature?

Any idea of the MSRP of the processor you mentioned in the other thread if it includes this feature?
MSRP is slated at $3500. AVS is a dealer for this product line.

sounds like a great feature, which product will support this?

The problem is that often apparent misalignment is really just chromatic aberration created by a less-than-perfect lens. When this occurs the misalignment is non-linear and an across-the-board adjustment wouldn't help. You'd end up with the pixels aligned at one part of the screen but with the misalignment worse on another area. About the best you could hope for would be to have the pixels aligned at the center of the screen.

I'd be interested in cheap digital only (HDMI, DVI inputs/outputs only) videoprocessor for better ITVC to 24/48fps and with MC correction. I defenitely wouldn't pay almost price of the 2nd projector for it: it not that bad with my Pearl. So answer: for this price - no

Lumagen or DVDO? What have you heard? :)

Lumagen or DVDO? What have you heard? :)
I made the request and they said they would consider doing it. If there is enough popular demand I'm pretty sure they'll do it.

At first I was going to mention the manufacturer but then decided against it since I don't want to be the one who puts them in an uncomfortable position. ;) I think it's less disruptive if one person says X people want this feature vs. X people emailing them for it.

The problem is that often apparent misalignment is really just chromatic aberration created by a less-than-perfect lens. When this occurs the misalignment is non-linear and an across-the-board adjustment wouldn't help. You'd end up with the pixels aligned at one part of the screen but with the misalignment worse on another area. About the best you could hope for would be to have the pixels aligned at the center of the screen.
My MC is uniform across the screen. And I guess most people who answer "Yes" would fall into that category. But being able to get the RGB pixels aligned in the center is better than no solution at all, IMO.

I would rather see the problem fixed at the source rather than try to kludge it via a processor.

I would rather see the problem fixed at the source rather than try to kludge it via a processor.
That's a nice thought....

If they can offer full digital panel rotation, and maybe even some sort of partial pixel adjustment (if such a thing is even possible), I think this product would be in high demand, especially since it's one feature of an otherwise excellent VP. The ability to perfectly converge an LCD or LCOS projector that does not have digital convergence built in would be a fantastic thing, in my opinion. I would probably buy the processor for this feature alone.

If they can offer full digital panel rotation, and maybe even some sort of partial pixel adjustment (if such a thing is even possible), I think this product would be in high demand, especially since it's one feature of an otherwise excellent VP. The ability to perfectly converge an LCD or LCOS projector that does not have digital convergence built in would be a fantastic thing, in my opinion. I would probably buy the processor for this feature alone.

Check out the e-warp feature on the Realta HQV. It's pretty powerful, but I don't think it allows separate RGB adjustments. I would think that it would be easy for them to add though.

No electronic system is going to be able to compensate or adjust in less than one pixel increments...period. But, being able to do a full pixel shift by color in zones would be a benefit where the global, full pixel shift capabilities now offered don't solve the problem.

Physical convergence/panel alignment capability in HT projectors is just not going to happen, for a variety of technical reasons, some of which I've touched on in previous posts. MC is less a function of panel alignment and more a function of aberrations in the optical system that cause MC in specific areas. There is no physical way to adjust this out.

Cheers,

Yeah, a way to adjust for convergence in say 25 zones like on a crt front projector would be da bomb. This should allow you do dial out any chromatic aberration caused by the lens as well as make up for any physical panel mis-alignment that might exists, right?

pclausen,

No, not right...you can only shift image data electronically a full pixel...so you can really only compensate for pixel errors greater than 1 pixel. Pixel MC less than one pixel cannot be improved, only moved to another location, but with the same error.

CRT projectors had no fixed pixel structure and the scan lines could be adjusted to fall anywhere that was necessary to get accurate registration between colors.

Cheers

Tom,

"Pixel MC less than one pixel cannot be improved"

Even less then one pixel might be able to be improved somewhat though. Say red was 3/4 of a pixel off to the right. You electronically shift red a full pixel to the left. Now your red convergence is off 1/4 of a pixel to the left. Not perfect but improved.

Shawn

In case Lumagen is paying attention to this, I want to be on record saying that I would be interested even if we only had 1 pixel vertical and horizontal movement. That would fix about 90% of the problem on my Pearl.

2 pixel is better than 1 pixel. 3 is better still.

Of course, if I'm going to pay 3K for a VP, it's other features had better be downright impressive as well.

I'd be willing to pay several hundred bucks for a little black box that just fixes convergence. There's your killer app.

Shawn,

What you say is correct, tho at some point you have to revert to not shifting to the left, if you're not doing a global shift...two different source pixels will end up being mapped to the same pixel at that point...the result of which is certainly an artifact of some sort. It's something to consider.

Shawn,

What you say is correct, tho at some point you have to revert to not shifting to the left, if you're not doing a global shift...two different source pixels will end up being mapped to the same pixel at that point...the result of which is certainly an artifact of some sort. It's something to consider.

But what he says is accurate as long as you're sliding the entire panel to the left, correct?

Is there currently software available on a HTPC to do this shifting? Sounds like it should be possible?

The problem with "mis-convergence" is it's caused by two things: linear panel shift and non-linear CA at the edges caused by the lens. The first is easy to fix, the second, less easy - even though the latter can easily cause as much or more shift than the former.

Even simple digital linear RGB would be a big plus, and is so easy to implement that I don't know why projector manufacturers haven't implemented it - perhaps they're afraid of admitting to or drawing attention to the fact of panel misconvergence.

Anyone else interested in weighing in? :)

Am I reading this correctly, that this feature would compensate for the softness found in 3 chip projectors, without adding artifacts, essentially bringing the sharpness to the level of single chip DLP? If so, I am definately interested.

Tom,

"Pixel MC less than one pixel cannot be improved"

Even less then one pixel might be able to be improved somewhat though. Say red was 3/4 of a pixel off to the right. You electronically shift red a full pixel to the left. Now your red convergence is off 1/4 of a pixel to the left. Not perfect but improved.

Shawn

This is perposturous... I cant live with a 1919 x 1080 display ;)

seriously though maximum 1/2 pixel error is the best that can be gaurenteed. But some could have greater benefits like in the scenario above.

Am I reading this correctly, that this feature would compensate for the softness found in 3 chip projectors, without adding artifacts, essentially bringing the sharpness to the level of single chip DLP? If so, I am definately interested.

The best measure I've seen of sharpenss is MTF, which is effected by lots of different parameters. Better convergence will improve sharpness, yes. But even a perfectly converged LCOS device is likely to have a lower MTF than a single-chip DLP.

Anyone else interested in weighing in? :)

I've already weighed in. But, as the baby in Dinosaurs says, "I WANT DAT!"

Beyond MC correction there is potentially a huge market for electronic CA correction because the optics can be made less expensive. The question is whether the price of the electronic correction can be lower than the cost of color corrected optics themselves as well as "preaberrated" display devices.

Just FYI - I have been waiting since reading this thread in november to be able to announce this. :)

We announced our new 'Geo' chip today. It can do independant image warping for each of the 3 RGB color channels, effectively giving you digital convergence correction and digital chromatic aberration correction to a surprisingly accurate degree.

Our image warping technology is based on cubic (curved) line equations allowing any type of distortion (barrel, pincushion, spherical, etc) to be accurately corrected.


Here is the press release:

http://www.siliconoptix.com/chipProducts/Realta/pressRelease/SilcionOptixGEOChip.cfm?CFID=&CFTOKEN=7b7a427637ef81d3-F3091923-7E90-E2A3-B943DC5EE5BEC3D4

Let me know if you have any questions!

Thanks,

Andy K.
ASIC Design Engineer
Silicon Optix, Inc.

Andy,

Will you be demoing this at CES and, if so, where?

JVC seems to have gotten this right with the HD1/RS1. They provide adjustments for the red, green and blue channel to shift multiple pixels and in any direction. This should enable one to get 1/2 pixel or less alignment with all 3 colors, provided any MC is uniform. That's about my only concern. This is a great step forward over what is sorely lacking for the Ruby/Pearl. Ideally though they would have provided a few different zones so that different parts of the image could be shifted differently to account for any non-uniform MC. Still, a great step in the right direction.

Shawn, let me double check and get back to you.

I agree the JVC pixel shift was and is a great feature (and kudos to them to have put it in every DILA projector as far as I know, all the way back to the G1000 which I own) but it wont handle non-uniform convergence issues as you suggest. Non-uniform convergence is one of the issues our new chip was designed to correct.

Thanks,

Andy K.
ASIC Design Engineer
Silicon Optix, Inc.

Just FYI - I have been waiting since reading this thread in november to be able to announce this. :)

We announced our new 'Geo' chip today. It can do independant image warping for each of the 3 RGB color channels, effectively giving you digital convergence correction and digital chromatic aberration correction to a surprisingly accurate degree.

Our image warping technology is based on cubic (curved) line equations allowing any type of distortion (barrel, pincushion, spherical, etc) to be accurately corrected.


Here is the press release:

http://www.siliconoptix.com/chipProducts/Realta/pressRelease/SilcionOptixGEOChip.cfm?CFID=&CFTOKEN=7b7a427637ef81d3-F3091923-7E90-E2A3-B943DC5EE5BEC3D4

Let me know if you have any questions!

Thanks,

Andy K.
ASIC Design Engineer
Silicon Optix, Inc.


Absolutely fantastic. Time to pull out the checkbook. Do we have MSRP?

Absolutely fantastic. Time to pull out the checkbook. Do we have MSRP?

Is this announcement about a chip that will power other devices and we have to wait for such devices to now be built? Or are such devices already available?

I agree the JVC pixel shift was and is a great feature (and kudos to them to have put it in every DILA projector as far as I know, all the way back to the G1000 which I own) but it wont handle non-uniform convergence issues as you suggest. Non-uniform convergence is one of the issues our new chip was designed to correct.

Thanks,

Andy K.
ASIC Design Engineer
Silicon Optix, Inc.

Excellent indeed. Can you please elaborate on how it addresses convergence and in particular, how does an end user make these adjustments (what types of controls and settings are available than enable you to make adjustments for non-uniform MC)?

Just to clarify, this is a chip and not an end-user product. It would need to be integrated into a projector or a standalone processor by one of our customers.

Our chip works on the principle of a 'warp map' which is a set of regions of the screen, and within each region a parametric equation can be used to distort the pixels per color component in a curved (or linear) fashion. This enables virtually unlimited flexibilty in terms of how the pixels are distorted or shifted, on a per color component basis.

Generation of the warp map (in other words, the interface by means to generate it/tweak it) could be done similar to CRT-type convergence (move a cursor around a grid and shift color components around the cursor) or by using some kind of camera/sensor to take a picture of the convergence and create a 'warp map' that represents the inverse of that convergence.

Thanks!

Andy K.
ASIC Design Engineer
Silicon Optix, Inc.

I have an old JVC G10 type projector, and the Dilard calibration software does in fact have the ability to digitally shift any of the three panels by one or two pixels in any direction. I voted not useful because the projector seems to be aligned well enough internally that all the digital corrections I tried make things worse.

But the kind of spline-based, infinitely adjustable mapping that kromkamp is talking about later in the thread could clearly be very useful. The lenses and optical systems will soon be the real bottlenecks in these systems.

I would rather see the problem fixed at the source rather than try to kludge it via a processor.
What we need is something like piseo electric devices that can move the DLP chips mechanically. Piseo electric devices can make movement down in the tenth of an atom range for alignment. Probably accurate enough convergence even for AVS members.

What we need is something like piseo electric devices that can move the DLP chips mechanically. Piseo electric devices can make movement down in the tenth of an atom range for alignment. Probably accurate enough convergence even for AVS members.

Piseo?!?! what a Fancy name. Whats the premium to pay for such tech? Always consider economic viabilty with our <5K PJs.

piezo

Piseo?!?! what a Fancy name. Whats the premium to pay for such tech? Always consider economic viabilty with our <5K PJs.
Piezo electric crystals bend when a voltage is applied to them. They could be used to accurately align the three color panels. Other techniques, such as electronically movable mirrors outside the chips could be used for accurate alignment.

The question is why can't the panels be accurately aligned at the factory. Is there really any need for the user to be able to align the panels in their home?

I agree the JVC pixel shift was and is a great feature (and kudos to them to have put it in every DILA projector as far as I know, all the way back to the G1000 which I own) but it wont handle non-uniform convergence issues as you suggest. Non-uniform convergence is one of the issues our new chip was designed to correct.Andy,

I saw a report about one of the new JVC's that was nonuniform (closer in the center than out towards the edges). It could have been caused by the lens, but regardless of the cause, I think this is a great idea to allow adjustments with a processor. Any news on the timeframe before consumers might be able to buy a product with this in it?

BTW: I think this technique would be especially powerful with a 4k projector and a processor that could handle this stuff in 4k space for the future, given how small the adjustments could be with that many pixels.

Thanks,
Darin