or Connect
AVS › AVS Forum › Display Devices › Digital Hi-End Projectors - $3,000+ USD MSRP › RS1 Color Correction - JVC Solution?
New Posts  All Forums:Forum Nav:

RS1 Color Correction - JVC Solution?  

post #1 of 1177
Thread Starter 
I noticed in another thread that tstites from JVC has stated that JVC has a solution for the extended primary issue the RS1 / HD1 obviously has with it's colors. I didn't want to derail the other thread, but I'd certainly like to hear more about this external processor Mr. Stites is referring to in the quote below...


Quote:
Originally Posted by tstites View Post

Before anyone jumps on the extended primary issue, the colors are being corrected with an external processor with 3D LUT.


Can Mr. Stites or anyone else comment in more detail about 3D LUT processors or the specific processor being mentioned? How practical are these devices for addressing the extended primaries and more importantly, is this 'correction' going to cost thousands of dollars and take a rocket scientist to operate?

Lastly, could Tom please explain why the extended primaries cant simply be addressed via a FW upgrade???
post #2 of 1177
Check later in the post that you read that. Tom give some examples.

Basically pro-post production equipment -
post #3 of 1177
Wet,

Providing an appropriate CMS in the projector itself requires more than a FW upgrade, the hardware/memory space to do the required LUT does not exist in the projector.

As I have said before, the RS1 design is fixed and will not be updated. We and others will likely show some affordable external solutions at CEDIA.
post #4 of 1177
which thread was it?
post #5 of 1177
Thread Starter 
Quote:
Originally Posted by tstites View Post

Wet,

Providing an appropriate CMS in the projector itself requires more than a FW upgrade, the hardware/memory space to do the required LUT does not exist in the projector.

As I have said before, the RS1 design is fixed and will not be updated. We and others will likely show some affordable external solutions at CEDIA.

Obviously it doesn't sound like JVC is going to do anything with the RS1, but would it not be possible to add another optional primary setting or even change the existing colors via a FW update if JVC wanted to? Or is the only way to address this with a complete CMS?

Please forgive my elementary understanding of PJ design, but It just seems to me like it shouldn't be that hard to add an additional setting via FW that would get the extended primaries much closer without having to implement a full CMS... Granted, this obviously wouldn't give the user full adjustability, but I don't think that's what's needed to satisfy the bulk of the unsatisfied consumers such as myself.

BTW, thank you Tom for talking about this issue!!!
post #6 of 1177
Quote:


We and others will likely show some affordable external solutions at CEDIA.

This is VERY good news.
post #7 of 1177
Thread Starter 
Quote:
Originally Posted by TomHuffman View Post

This is VERY good news.

If affordable doesn't mean thousands and thousands of dollars, I completely agree!
post #8 of 1177
Quote:
Originally Posted by tstites View Post

Providing an appropriate CMS in the projector itself requires more than a FW upgrade, the hardware/memory space to do the required LUT does not exist in the projector.

That's some good planning there... Didn't anyone realize the colors were off before the design was finalized?
post #9 of 1177
Quote:
Originally Posted by Wet1 View Post

If affordable doesn't mean thousands and thousands of dollars, I completely agree!

...and I agree with that.

Quote:
Originally Posted by Wet1 View Post

but would it not be possible to add another optional primary setting or even change the existing colors via a FW update if JVC wanted to? Or is the only way to address this with a complete CMS?

...as well as that. I've mentioned it at least a dozen times on other threads, but I'd like to know why a firmware upgrade couldn't add a picture mode with a reasonably accurate color gamut a la the Panny AE1000.

I don't have a problem with overlooking an occasionally oversaturated green due to the overall excellence of the rest of the RS-1's PQ, but with an accurate color gamut, I'd have no need to upgrade for a very long time.
post #10 of 1177
Quote:
Originally Posted by MikeSRC View Post

...but with an accurate color gamut, I'd have no need to upgrade for a very long time.

Hmmm ... could this be one reason for a manufacturer to *not* provide one?
post #11 of 1177
Quote:
Originally Posted by Wet1 View Post

Obviously it doesn't sound like JVC is going to do anything with the RS1, but would it not be possible to add another optional primary setting or even change the existing colors via a FW update if JVC wanted to? Or is the only way to address this with a complete CMS?

Please forgive my elementary understanding of PJ design, but It just seems to me like it shouldn't be that hard to add an additional setting via FW that would get the extended primaries much closer without having to implement a full CMS... Granted, this obviously wouldn't give the user full adjustability, but I don't think that's what's needed to satisfy the bulk of the unsatisfied consumers such as myself.

BTW, thank you Tom for talking about this issue!!!

Actually it IS hard - which is why few PJ under 10K have real CMS. A color control is simply gain manipulation of the color components - an amplifier in analog - a multiplier in digital. A gamut change is a 3x3 matrix transform - which requires a lot more hardware to implement. Once you add the extra hardware to do the gamut warping - then you can simplify the user interface to just choose video display standard presets fixing the transform (i.e. REC709) and thus make the hardware simpler LUT - or you can go all out with fine tune of each color to make whatever gamut you want which requires more flexible variable transform (i.e. CMS). Many under 10K do have the presets without having the full CMS - but if the hardware does not exist to do the transform - you cannot upgrade the firmware menus to enable this missing feature.

An external solution will need more than 8b video to avoid creating banding/dithering/clipping - certainly available in PRO hardware - but HDMI1.3 is just now taking off in consumer.

Of course JVC realized the colors were off - while in the PRO world it is great for the Digital Cinema space - it is wrong for REC709 work - and they know an external processor would be required. But they had to outdo Sony - which sold the promise of MORE colors and MORE contrast. And it worked with the result they had to listen to only a few of us noisy calibrators complaining about it.
post #12 of 1177
Mike,

Remapping color coordinates is more complicated than just making a new picture mode or preset...if that were the case you could do it with the existing controls and save it as a user defined setting. As I said in the previous post, additional hardware is needed in the projector.

As for the colors being "off", that was a design decision the factory made for specific reasons that are unknown to me. While I would be the first to admit that they are off with respect to absolute accuracy, I know full well that the average consumer finds "production grade reference monitors" flat looking and prefers the more "vibrant" colors of consumer level TV's/Displays...one possible explanation for the choice.

Stay tuned, I'm sure we'll eventually satisfy the most demanding among you.

Cheers,
post #13 of 1177
Quote:
Originally Posted by krasmuzik View Post

.

An external solution will need more than 8b video to avoid creating banding/dithering/clipping - certainly available in PRO hardware - but HDMI1.3 is just now taking off in consumer.

.


As long as the processing itself takes place at 10bit or greater then you can deflate the output back to 8bit with realistically minimal problem.
post #14 of 1177
Quote:
Originally Posted by tstites View Post

Stay tuned, I'm sure we'll eventually satisfy the most demanding among you.

Cheers,

By making this projector's colors better or a new model?
post #15 of 1177
Quote:
Originally Posted by tstites View Post

Providing an appropriate CMS in the projector itself requires more than a FW upgrade, the hardware/memory space to do the required LUT does not exist in the projector.

Thank you for (finally) clarifying this. I wish however you had posted this earlier. A definitive statement such as yours may well have influenced my purchase decision in the direction of BenQ. I was holding out hope that JVC would ultimately do the right thing by their customers, but it now appears that is not going to happen.

Quote:
Originally Posted by MikeSRC View Post

I'd like to know why a firmware upgrade couldn't add a picture mode with a reasonably accurate color gamut a la the Panny AE1000.

Here is my guess: The correction needs to be done in hardware. As a specific example, an oversaturated green is tamed by mixing in a little red and blue into the green channel. It is not just a question of altering the firmware to reduce the gain on the green channel. Rather, a full (3*3) matrix multiply is required. While the RS1/HD1 contains a matrix multiplier (for example in the component -> RGB color decoder), based on some posts by Greg Rogers, that piece of circuitry is on the wrong side of the gamma circuit. (Caveat - I have not been through the math to confirm the above is correct).

That said, I would be somewhat surprised if the digital video path is not implemented in an FPGA, (where the FP' stands for Field Programable') which means that, although the circuitry may not currently be capable of desaturating the primaries, it could be reprogrammed to do so (hopefully by the end user or at worst by the factory). If this is the case then stating that hardware/memory space to do the required LUT does not exist in the projector is a little sophistic. No worse though than the "Reference Series" moniker when, in fact, it is incapable of rendering reference colors.

Tom please don't take this as a personal attack. I very much appreciate your participatation in these forums. I am merely expressing my frustration, apparently shared by a few others here, that this projector could have been so much better (in fact almost perfect) with just a relatively minor amount of extra effort in the initial design and implementation.

Brent
post #16 of 1177
Any ideas on pricing yet Tom?

Gary
post #17 of 1177
Tom, I second the sentiment. Thanks for the input. Sorry to kill the messenger. Your participation is greatly appreciated. I am just deeply frustrated with the colors.
post #18 of 1177
The issues i've read regarding the colours on the jvc are the single reason I don't own one right now. I'm still close to ordering one, but the closer Cedia comes the less tempting it is.
post #19 of 1177
Warbie
Make the decision from what you read AND what you see with your owns eyes.
post #20 of 1177
Quote:
Originally Posted by Ohlson View Post

Warbie
Make the decision from what you read AND what you see with your owns eyes.

I agree with this approach. Here is an image from a Pearl and an RS1 of roughly the same frame. Neither projector was calibrated as I just spent a few minutes with each before I took the shots. However from what I have read I expected the green plant to be glowing in the RS1 image... I'll let you be the judge. Please don't pick apart the other differences as again neither projector was really tweaked.



post #21 of 1177
Quote:
Originally Posted by Mr.D View Post

As long as the processing itself takes place at 10bit or greater then you can deflate the output back to 8bit with realistically minimal problem.

Only if you dither...which may get smoothed out by the LCOS panel anyways so it might work - but warping a color space simply cannot preserve the original RGB space - warping means you are skipping or merging colors.
post #22 of 1177
Quote:
Originally Posted by kiwishred View Post

That said, I would be somewhat surprised if the digital video path is not implemented in an FPGA, (where the FP' stands for Field Programable') which means that, although the circuitry may not currently be capable of desaturating the primaries, it could be reprogrammed to do so (hopefully by the end user or at worst by the factory).


Brent

Even if it was - FPGA's are often capacity limited - a 3x3 matrix multipliier simply takes more space - it is an RGB2RGB transform that is addition to the video2RGB transform existing in the decoder..and best done with even more bits after the RGB gamma. You cannot reprogram the originally FPGA unless you had the design code the FPGA implements anyways. Bit far-fatched to think an FPGA is user programmable just because it is field programmable....
post #23 of 1177
Quote:
Originally Posted by Warbie View Post

The issues i've read regarding the colours on the jvc are the single reason I don't own one right now. I'm still close to ordering one, but the closer Cedia comes the less tempting it is.

If you have never seen a calibrated image and colorspace - then do not worry about it. You will not know what you are missing. The PJ was designed to make the average joe happy who thinks MORE color means better color.
post #24 of 1177
Quote:
Originally Posted by krasmuzik View Post

Only if you dither...which may get smoothed out by the LCOS panel anyways so it might work - but warping a color space simply cannot preserve the original RGB space - warping means you are skipping or merging colors.


What do you think a LUT does anyway?

You go to 10bit thats 1023 code values compared with 255 for 8bit processing.
You do your processing with 1023 code values of precision to describe your colour remap. This is then deflated back to 255. Your rounding error in 10bit is greater than 4 code values before it feasibly produces a potential contouring threshold back in 8bit . Your not going to introduce additional visible banding with any reasonable colour remap. Certainly you are not going to introduce banding that wasn't already present in the original 8bit material.

Bear in mind that most chipsets these days work in 12bit or higher anyway for the processing.

Conclusion , you ain't going to introduce banding and you don't need to stay inflated at a higher bit depth after the processing.
post #25 of 1177
Video processing is using increased bits internally specifically for rounding error minimize through the processing steps - you would lose data if you did math with 8b for an 8b result.

Rounding means that internally your 8.0 bit pixel gets represented as 8.2 bits internally - which helps it round up or round down and not lose precision to truncation that happens as it goes thru the chain. Hopefully you have no rounding errors in the end so 256 values in means 256 values out. Internal fractions with internal rounding means that you can have fractional values so that 8.25 + 8.75 = 17.00 rather than 16.00 that you get with truncation - only to later in the chain add with 8.5 getting now 24 rather than 25.5.

The whole point of RGB warping is instead of stepping from 16 to 17 in the original RGB space - we step from 15.8 to 16.4 so that after the warping - we still have the same number of colors. The colors are just shifted around to different fractional values so they all still fit in the native colorspace. If you round back to 8b - then you step from 16 to 16 - and that is the very definition of banding - or for other fractions you might go 16 to 18. The steps become too close or too large or not existant - which does not happen with more bits to communicate the result. You could dither by flipping between 16 and 17 to make it look like 16.4. The whole point of the RGB transform is it shifts the range of values that each of RGB can occupy - you cannot stuff 256 values back into the same 256 values - then you have not done a transform.

We are not using fractional values in a RGB transform to avoid accumulation of error during the processing - rather the fractional values are the intended ->result<- of the transform. So a video processor maybe OK rounding off 12b to 8b - but if you are doing a transform - you just defeated the purpose of what you are trying to do. By having a larger 10-12b color space that goes out to the native gamut - you can remap the subset of colors that are the translated video to the same amount of colors just a different subset to fit a different gamut.

I think designer that is planning to put a CMS into an external processor that communicates the results over 8b knows that banding/dithering/clipping will be the side impact. HDMI1.3 with higher bit support would be required. Though if you think you can market a processor with banding/dithering/clipping - maybe enough people think getting proper colors is worth it.
post #26 of 1177
Thread Starter 
Quote:
Originally Posted by Bulldogger View Post

Tom, I second the sentiment. Thanks for the input. Sorry to kill the messenger. Your participation is greatly appreciated. I am just deeply frustrated with the colors.

I feel the same way Tom. I apologize if you find my posts offensive and redundant (and I'm sure you do), but I can't tell you how disappointed I was with my RS1. If the colors were addressable I would never have sold my PJ, but the often unrealistic colors were so distracting to us that I HAD to get rid of it. Even my wife commented about the colors the instant she saw an image, and this says A LOT! I hope JVC does come out with a reasonable solution to the extended primaries (and that it addresses CIH capabilities as well), but I fear this solution will be a several thousand dollar band-aid or a trade-off for something else. Let's hope JVC proves me wrong this time.

Tom,
Again, thank you for finally addressing this issue and the realities of fixing it... it's been a looonnnnng time coming.
post #27 of 1177
There is a lot of doom and gloom on this forum. Many people, including myself, have been complaining about the RS1 colors (and an industry trend towards blown out colors in general) and now we learn that JVC will definitely address this very soon with an external processor.

Gee, isn't this a good thing, rather than a bad thing? Let's not allow the perfect be the enemy of the good. This is great news, especially if the device is marketed by JVC as a completely separate product that can be used with any display. But even if it is RS1-specific it should make a lot of owners and potential owners happy.
post #28 of 1177
Quote:
Originally Posted by TomHuffman View Post

There is a lot of doom and gloom on this forum. Many people, including myself, have been complaining about the RS1 colors (and an industry trend towards blown out colors in general) and now we learn that JVC will definitely address this very soon with an external processor.

Gee, isn't this a good thing, rather than a bad thing?

So, you have to buy an external $1k box because JVC wanted to save $10 in parts in the projector? And this is good news?
post #29 of 1177
Cant look at it like that. You have what you have right now, which is a not accurately color correctable RS1.

If an external box will take care of that problem for many of these guys, thats a good thing for them.
post #30 of 1177
Quote:
Originally Posted by Charles R View Post

I agree with this approach. Here is an image from a Pearl and an RS1 of roughly the same frame. Neither projector was calibrated as I just spent a few minutes with each before I took the shots. However from what I have read I expected the green plant to be glowing in the RS1 image... I'll let you be the judge. Please don't pick apart the other differences as again neither projector was really tweaked.




Ugh.... look how neon that plant is and the horrible sun burnt skin tones!!!

Which ones which by the way????
JVC #1, Sony #2?
New Posts  All Forums:Forum Nav:
  Return Home
This thread is locked  
AVS › AVS Forum › Display Devices › Digital Hi-End Projectors - $3,000+ USD MSRP › RS1 Color Correction - JVC Solution?