View previous topic :: View next topic |
Author |
Message |
Nashou66
Joined: 12 Jan 2007 Posts: 16171 Location: West Seneca NY
|
Link Posted: Fri Oct 30, 2009 2:01 am Post subject: |
|
|
Thats one thing Bob posted over on avs was about not having contrast modulation and how he never noticed the edges being dimmer than the rest till he decided to blend.
You could pull the one PJ down and then put the other next to it on the floor for testing as its really the only way to find the location for the two later on when your ready to
go top mount. you might have to move a row of seat from your theater pics or they might fit right up front.
Athanasios
_________________ Don't blame your underwear for your crooked ass~ unknown Greek philosopher
"Republicans believe every day is the Fourth of July, but the Democrats believe every day is April 15." --- President Reagan
One Smart Dog!!!
Marquee High Performance Bellows now shipping!!
Marquee Modifications and Performance Enhancement
Marquee C-element and Bellow removal
|
|
Back to top |
|
|
MikeEby
Joined: 24 Jun 2007 Posts: 5238 Location: Osceola, Indiana
|
Link Posted: Fri Oct 30, 2009 4:02 am Post subject: |
|
|
huggy wrote: | Antorsae.........you are a legend
Mike........If you can come up with a GUI for this like eac3to so dumdums like myself can actually have a chance of blending via a p.c,then I will gladly jump on a plane ,fly up to the states and kiss your beloved ass OK maybe not you get what I mean
|
Really not that much too it, Andres did all the work. One thing you do have to realize the blend zone will be very visible on the windows desktop there is no blend to it at all. Dialog boxes in the blend zone will look really messed. The blend is really only for the video playing...I'll pass on the kiss...Thanks for the offer.
Nashou66 wrote: | Thats one thing Bob posted over on avs was about not having contrast modulation and how he never noticed the edges being dimmer than the rest till he decided to blend.
You could pull the one PJ down and then put the other next to it on the floor for testing as its really the only way to find the location for the two later on when your ready to
go top mount. you might have to move a row of seat from your theater pics or they might fit right up front.
Athanasios |
I wonder if you have less problem with dark edges if you don't maximize the raster area quite as much? Better focus too in the corners, I'm only probably going to go with about a 138" wide screen so I wouldn't be pushing the brightness too hard. What projectors have contrast modulation? I don't have to go with NEC they are so fricking loud, I wouldn't mind finding something a tad quieter...I'm probably going to need a new green tube anyway before all is said.
I do wonder about non-scope movies?...For those it seem like it would be better to run 1080p? Andres what's your feeling about that?
Yea I have the lift so taking the projector down is not really a big deal...BTW this is not a project I'm starting tomorrow just looking into what is involved.
Also anyone trying this make sure you have a keyboard with a numeric keypad. All the zoom functions in MPC-HC use those...My smaller HTPC wireless keyboard doesn't have them. A separate USB keypad might be really handy or just program one of the fancy wiz-bang remotes with those same keycodes.
Mike
_________________ Doing HD since the last century!
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Fri Oct 30, 2009 7:47 am Post subject: |
|
|
Hey guys - Thanks for the kind words.
I've always thought that the biggest challenge with the blend is to match the two PJs color-wise.
What really needs to happen is that all 6 colors (R G B, R G B) need to EXACTLY match a gamma curve (L = k V ^ g), where g is the numeric gamma curve), and the k values (i.e. amount of R G B proportionally) need to be matched between the two projectors.
In addition to that, you may want them to be matched at 6500K (a specific set of 3 different k values, i.e. R G B proportions), but -while very nice- is not 100% required for a seamless blend, yet the PERFECT match of the 6 light output curves is...
The blend compensation line:
Code: | if (abs(tex.x-0.5) < blends) c = pow(abs(tex.x-0.5)/blends, 1/2.22);
|
...just multiplies the pixel value at the blend zone in such a way that when you add the conversely compensated value for the other projector, the light output remains that of the original pixel... BUT... ONLY IF THE LIGHT OUTPUT OF THE TWO PROJECTORS FOLLOW THE FORMULA ABOVE!!!!! i.e. with this approach it is only possible to blend projectors that match exactly a power function (a CRT follows that closely, but not 100%). This approach is what the TV-ONE does and a more generic one is what the DVX does (it is still based in multiplying the original pixel values, but with an arbitrary curve), but you have to edit the curves manually...
That's why my current project is to first adjust G2/DRIVE settings automatically with a digital camera and software I'm writing that controls marquees, the camera and a pattern generation (with a server-side program in another PC) to get as close as possible with the PJs alone, and then measure the error vs. a perfect power function (L = k V ^ 2.2) and upload custom LUT tables to the PC to compensate for the error (this can either be done at MPC enhancing the shader code for this -very easy- or at the video-card level -which is what I've already done-).
I already had a version of the above software operational to get the LUT curves alone and now I am improving it to get the G2/DRIVE values.
I hope this helps anyone!!!
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Fri Oct 30, 2009 7:52 am Post subject: |
|
|
Quote: | I do wonder about non-scope movies?...For those it seem like it would be better to run 1080p? Andres what's your feeling about that?
|
Mike - you are absolutely correct. What I want to do is to make sure there's no resampling to get 1:1 pixel mapping, both horizontally and vertically. For 16:9 movies you want to have whatever the vertical resolution of the source is, either 1080p or 720p.
Ciao.
|
|
Back to top |
|
|
Nashou66
Joined: 12 Jan 2007 Posts: 16171 Location: West Seneca NY
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Fri Oct 30, 2009 3:36 pm Post subject: |
|
|
Athanasio
Re: blend zone -> the TV-ONES already provide exactly the kind of blend formula I implemented in software with the pixel shader. No need to tweak further (unless you want individual gamma values for R G B, but that's another discussion that opens it own can of worms).
Re: gamma LUT correction -> YES! I talked to Robb @ TV-ONE and he said he'd be interested in adding this control to their units, however it never happened (and I don't have TV-ONE units to test).
The stable version of my software can calculate the gamma correction LUTs required for a perfect blend zone (no matter which blending devide you use: PC, DVX or TV-ONE). You can deploy that LUT correction tioday in NVIDIA cards (at 10-bits depth)... but to be honest in an ideal world it would be the TV-ONEs doing it.
|
|
Back to top |
|
|
Nashou66
Joined: 12 Jan 2007 Posts: 16171 Location: West Seneca NY
|
Link Posted: Fri Oct 30, 2009 4:01 pm Post subject: |
|
|
Tv One is suppose to add the controls to their units for a 256 point gamma curve that you can grab with your mouse pointer and adjust to where you want it. Now I am not sure if those will be for the blend zone alone and for each color individually but i do think that is what they were planing. I wonder If Andy Haliday in london has any units you could try. I would call Tv-One Andre and see if they can send you two demo units for your research and they might even let you have the codes for the gamma thats already in the units, they just have not written the PC based software yet. maybe since they are busy they can freelance it out to you since your already set up for testing it. BI et there would be some type of contract to sign but id rather have you working on it for us CRT blenders than the TV-One guys who mostly have this for Digital's. Maybe I can have them contact you. its worth a shot if your up to it?
Athanasios
_________________ Don't blame your underwear for your crooked ass~ unknown Greek philosopher
"Republicans believe every day is the Fourth of July, but the Democrats believe every day is April 15." --- President Reagan
One Smart Dog!!!
Marquee High Performance Bellows now shipping!!
Marquee Modifications and Performance Enhancement
Marquee C-element and Bellow removal
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Fri Oct 30, 2009 4:49 pm Post subject: |
|
|
YES!
|
|
Back to top |
|
|
Nashou66
Joined: 12 Jan 2007 Posts: 16171 Location: West Seneca NY
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Wed Nov 04, 2009 10:18 am Post subject: |
|
|
Here's a video explaining how this looks like:
http://www.youtube.com/watch?v=3Tw9fC8ecjU
This is the work-in-progress version of the 2nd application (to tune G2/DRIVE parameters interactively). I also have another one (not shown in the video) that measures IRE windows to calculate the gamma curves and then synthesize a compensation curve LUTs accordingly (which right now can be uploaded to an HTPC and ideally could be uploaded to the TV-ONE).
I am going really slow with this as I can only devote 30 mins or 1hr per day (max!) to this project late at night....
|
|
Back to top |
|
|
Nashou66
Joined: 12 Jan 2007 Posts: 16171 Location: West Seneca NY
|
Link Posted: Wed Nov 04, 2009 7:35 pm Post subject: |
|
|
antorsae wrote: | Here's a video explaining how this looks like:
http://www.youtube.com/watch?v=3Tw9fC8ecjU
This is the work-in-progress version of the 2nd application (to tune G2/DRIVE parameters interactively). I also have another one (not shown in the video) that measures IRE windows to calculate the gamma curves and then synthesize a compensation curve LUTs accordingly (which right now can be uploaded to an HTPC and ideally could be uploaded to the TV-ONE).
I am going really slow with this as I can only devote 30 mins or 1hr per day (max!) to this project late at night.... |
Ok I understand a bit more. I like how you put the patters to each PJ at a different location, or I assume this is what you have done. so one of the checkerboard patterns is from one PJ and the other is from the Second PJ. this way your able to match the colors in the area where the blend will be more closely as the paterns are right ontop in the same area of the phosphor.
If this is not the case then You just gave me an Idea to invert one PJ's image with a colored pattern towards the top and on the other C2-2250 i can flip the image to get the pattern at the bottom therefore putting each image on top of each other. now with the blend zone off I can better color match each color since they are in the same area of the outer edge of the phosphor.
I love how you can use the lap top to control the Marquee, i need to lear how to do that. I need to make a visit to europe!
Athanasios
_________________ Don't blame your underwear for your crooked ass~ unknown Greek philosopher
"Republicans believe every day is the Fourth of July, but the Democrats believe every day is April 15." --- President Reagan
One Smart Dog!!!
Marquee High Performance Bellows now shipping!!
Marquee Modifications and Performance Enhancement
Marquee C-element and Bellow removal
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Thu Nov 05, 2009 6:43 am Post subject: |
|
|
Hi Athanasios - yes, each projector displays its own pattern near the blend zone - which is where you really want the colors to match.
I had to write a pattern generator piece of software on the computer which drives the projectors (which is not the laptop, it is a regular HTPC connected to the two marquees via RGBHV) that displays test patterns as instructed by the laptop (color ramps, checkerboards, etc.).
|
|
Back to top |
|
|
bbfarmht
Joined: 27 May 2006 Posts: 1273 Location: Where the Mississippi runs east to west!!
|
|
Back to top |
|
|
AnalogRocks Forum Moderator
Joined: 08 Mar 2006 Posts: 26690 Location: Toronto, Ontario, Canada
TV/Projector: Sony 1252Q, AMPRO 4000G
|
Link Posted: Sat Nov 07, 2009 3:55 am Post subject: |
|
|
That's an ATi card. This is for Nvidia only isn't it?
_________________ Tech support for nothing
CRT.
HD done right!
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Sun Nov 08, 2009 8:50 am Post subject: |
|
|
I only have NVIDIA cards, so I've only tested it w/ Nvidia and XP. This doesn't mean it doesn't work with ATI cards, only that it hasn't been tested.
In fact, ATI released something called Eyefinity (see review here http://www.guru3d.com/article/ati-eyfinity-editorial-review/) which SHOULD work (again, not tested) and not only that, should work in Windows 7 (NVIDIA Horizontal Span modes are reported NOT to work under Win 7).
|
|
Back to top |
|
|
bbfarmht
Joined: 27 May 2006 Posts: 1273 Location: Where the Mississippi runs east to west!!
|
Link Posted: Sun Nov 08, 2009 12:53 pm Post subject: |
|
|
AnalogRocks wrote: | That's an ATi card. This is for Nvidia only isn't it? |
Yeah, remembered that just after I posted. With the eyefinity I may be able to use ATI cards, but not with that card. Back to the drawing board.
_________________ Adam
"Any society that would give up a little liberty to gain a little security will deserve neither and lose both"
Benjamin Franklin
|
|
Back to top |
|
|
antorsae
Joined: 16 Jul 2006 Posts: 297
|
Link Posted: Thu Nov 12, 2009 7:25 am Post subject: |
|
|
Update: I stayed up last night working on the G2/DRIVE auto-match (computer-driven color matching, no manual interaction needed). I am happy to report that I have G and R matching perfectly. I need to clean up the code to match B (piece of cake, it's just that I don't want have the same code copied everywhere).
Setting this up is a bit complex and frankly I am not sure if ready for general consumption unless you can tweak the code if something doesn't work right, but to me is great.
If there is anybody out there with a:
* Canon DSLR supported by the EOS SDK,
* 2 Marquees setup in blend conf.,
* an HTPC w/ Nvidia card (or ATI w/ equivalent span mode) and
* another computer to run the program connected to the Marquees via RS232 and the camera w/ a USB cable (pretty steep requirements, I know!).... and
* Visual Studio and some knowledge of C++ (ability to compile and debug)
Let me know!!!!
|
|
Back to top |
|
|
Nashou66
Joined: 12 Jan 2007 Posts: 16171 Location: West Seneca NY
|
|
Back to top |
|
|
digitalayon
Joined: 02 Mar 2009 Posts: 921
|
Link Posted: Sun Nov 22, 2009 12:09 pm Post subject: |
|
|
why are the marquees the only ones having this be written for? Why not a base program code for all and then have individual drivers for each projector created? This could also open the door for not limiting the projectors to be the same units but possibly two different CRT's in a blend.
|
|
Back to top |
|
|
digitalayon
Joined: 02 Mar 2009 Posts: 921
|
Link Posted: Sat Nov 28, 2009 6:14 am Post subject: |
|
|
you guys think i can do a side by side stack with two barco 1208's using this method?
|
|
Back to top |
|
|
|
|