View Full Version : VP for Color Uniformity Correction
bsntn99 06-27-07, 10:57 AM What would really sell me on getting a VP would be one that could provide rgb offsets for different areas of the screen to correct for color uniformity issues. I have seen this problem with many LCD projectors and this has been reported by some of the RS1/HD1 owners. Sony has a built in correction feature that allows offsets at different IRE steps for different areas of the screen in the service menu for this but most pjs don't offer or have a way to access this.
So, I put it to DVDO, Lumagen, or whoever, can this feature be implemented on one of your current units or maybe the next gen product. This to me would be a huge selling point and I bet a lot of pj owners would be on board with this. I for one would be first in line for one of these units.
Cheers.
TomHuffman 06-28-07, 02:48 PM This is doable with digital technology but as of now no one has brought a commercial product to market.
bsntn99 06-28-07, 09:28 PM Thanks for the response. I know you appreciate accurate color as do I and I think this would be great feature to add to the current processors and probably easy to implement. I hope someone takes this up and offers it in one of their products.
oferlaor 06-29-07, 06:26 PM this is actually what william Phelps has been doing for LCOS and SXRD projectors for quite a while now (particularly JVC variants). He addresses each pixel individually and calibrates it.
the same can be done here, but only with very sensitive cameras and the capability to address each pixel directly (native rate).
The procedure would involve a video camera or sensor of some kind calibrating the gamma curve and color temperature of each individual pixel and then creating a LUT for that correction.
Quite feasible, if somewhat complex.
each LCOS unit has to calibrated in the factory, the reason why there is still room for improvement is simple. If you do it right you need to look for the darkest area and use this as a reference. You would probably get a projector that is quite dark and maybe even an unit that will not cover the entire IRE range any longer.
And even if a VP would provide a function like this, the adjustment is quite a science and I am pretty sure the results would be questionable.
Glimmie 06-30-07, 11:42 AM Interesting. We have actually looked into this at work where we use Panasonic AE 1000s for on set screenings.
What you are asking for is a "shading corrector". This porcess is as old as color TV its self. Camera pickup tubes (before CCDs) always required shading correction.
The process is as follows: You need to generate a waveform which is the inverse of the non uniformity of each panel. You then need to multiply this waveform with the panel video signal. This will null out the imbalance. Now in cameras the signals were a mix of sawtooth and parabola signals at H & V scan rates as that was the only significant distortions to correct. But a flat panel may require more complex correction waveforms. Nothing however a modern DSP chip can't generate. The trick is mapping the error which could be done on coarse blocks with a light meter. A computer program could take that data and generate a correction profile data set for the three panels at H & V rates. Then download this to the correctors waveform generator. With enough horse power you might be able to do the corrections in real time through a PC. The actual multiplication to the video data is realitivily simple and therefore not very CPU intensive.
Will we build it? No! By the time we invest the R&D into this there will be better projectors out there within the next six months. The only reason we use the AE1000 is it can do P12 color space within a $4000 unit which we can afford to loan out to productions. Both Christie and Barco have plans for small P12 capable projectors at a <$20K price point for small screening rooms.
bsntn99 07-01-07, 11:33 AM Actually, what I am looking for is much simpler. Just divide the screen into say a 5 x 5 matrix and allow different rgb offsets for each block. You could even break it down into smaller blocks if you want. This is akin to having separate white balance adjustment for each block rather than the whole screen as it is currently implemented in all devices. To me this is extremely simple to implement in any current video processor given they can already deal with the complexity of motion detection and scaling to 1080p.
TomHuffman 07-01-07, 04:16 PM Actually, what I am looking for is much simpler. Just divide the screen into say a 5 x 5 matrix and allow different rgb offsets for each block. You could even break it down into smaller blocks if you want. This is akin to having separate white balance adjustment for each block rather than the whole screen as it is currently implemented in all devices. To me this is extremely simple to implement in any current video processor given they can already deal with the complexity of motion detection and scaling to 1080p.Some displays (Sony front projectors) have this, but no external processor I know of offers this. It would be nice if one did.
Glimmie 07-01-07, 11:32 PM Actually, what I am looking for is much simpler. Just divide the screen into say a 5 x 5 matrix and allow different rgb offsets for each block. You could even break it down into smaller blocks if you want. This is akin to having separate white balance adjustment for each block rather than the whole screen as it is currently implemented in all devices. To me this is extremely simple to implement in any current video processor given they can already deal with the complexity of motion detection and scaling to 1080p.
Well that would provide a very coarse correction and may in fact be good enough for most viewing. But if you want to completely eliminate the shading or more realistically come very close then you must alter the RGB gains in an exact inverse of the error. In order to do that you must have an accurate model of the error and keep in mind it probably changes, i.e. drifts with temperature. So the detection and correction will have to be dynamic.
oferlaor 07-02-07, 04:19 AM 5x5 would be course and useless for all intents and purposes.
It has to be per pixel or nothing at all...
mhafner 07-02-07, 04:53 AM I doubt you can have good results with 8 bit correction data. You would need the panel to accept 10 bit or more precision from the video processor. This is best done internally using the full precision of the panel via service menu.
bsntn99 07-02-07, 03:43 PM 5x5 was just an example. Of course higher granularity would be better. I don't think it needs to be per pixel. My current pj has about 20 units shift in red across the screen in measuring 8 bit rgb levels. If I could get adjustments for maybe 10 blocks across, I could get all pixels within 2-3 rgb units which I doubt anyone could see any difference. Sony's have rgb offset adjustments at around 5x5 at different IRE step's and I have been able to dial this in on my older pj to where I could not distinguish any difference. Just hoping someone would see value in this and pursue this as an added feature to an outboard processor. Of course, I would be willing to beta test. :)
Cheers.
|