3D LUT Calibration - Page 5

Fist it depends on your HTPC use case. You would have to be running an application on the PC that can avail itself of madVR. Right now that is limited to MP-HC and JRiver MC for the most part. Now JRiver attempts to be an all emcompassing MediaCenter solution and it matches MP-HC for video playback. Sadly its PVR and EPG presentation are far inferior to WMC nor does it currently support Cable Tuners. I own JR MC18 but mostly watch it and wait. If the EPG/PVR evolves to WMC levels and it get reasonable support for the Ceton tuner. I can standardize on it. It is really then when madVR and a first quality 3D LUT puts us in a new ballpark. All the pieces come into place to have accurately calibrated displays at the highest level wherever an HTPC is attached. The idea of a single integrated source solution comes to the table.

This largely gets away for the need for external boxes like the Radiance.. One source, to render all content at the highest level while also correcting display deficiencies acheiving reference or near reference quality display output using a single consistent interface.Of course JR will have to stop adding 10k features a week and start focusing on bug killing. Sadly, they are more focused on porting their app to the Mac then building a robust EPG/PVR enviroment. They do have file playback largely perfected thanks to madVR and LAV.

Very interesting Gary,

I have JRiver but only use it on my laptop.

Certainly an HTPC setup, if it eliminates the need for a Radiance or for that matter an eeColorbox would be a significant step forward (or threat depending on your point of view).

As you know I have a DUO as VP so cannot access 3D LUT but luckily have a Display that is linear. However if I change my display this linearity may not be repeatable.

Thanks for the detail.

It's not so much JRiver, but the MadVR renderer which you can plug into any app that supports user configurable rendering output. MPC-HC for instance supports it as well.

Still a pretty cool piece of tech.

Ah, but it needs to priced completely different than other solutions. Volumes could be tens or even hundreds of thousands of units per year. $2500 solutions are not the answer. I am sure you are not interested in cannabilizing your high end professional botique market business where price is to some degree no object for a mass consumer product. That would take real vision and courage.. entrenched players are not usually the kind of companies that do this. Not to say your $600 eecolor box doesn't point toward that thinking but it is rare.

So how about it SpectraCal and Light Illusion? Any plans to implement a feature to generate MadVR compatible 3DLUTs from your software?

We are always evaluating the needs of the market.

I recently purchased a Hisense Pulse (2nd gen Google TV box).
At the software level, its completely crippled and useless for this purpose, but as a piece of hardware it is all one would ever need. HDMI in and out, ethernet/USB, IR blaster for $99.
Its unclear if a replacement Google TV OS would give enthusiasts greater control of the incoming video stream (I doubt it as it probably involves decrypting HDCP), but the OEMs who build these boxes (and they are all built on Marvell Armada 1500 SoCs) would no doubt have the means to add CMS functionality to the platform and still stay on the right side of the DMCA.

The dev hardware and SDK looks even better.
http://www.marvell.com/digital-entertainment/armada-1500/software-standards/

I'd happily contribute funds to any capable dev who could make something of this.

Hi Joel,

Thanks for info, I'm afraid I know little about most things but am interested in most.

Certainly this does seem to offer a different way to consider calibration.

You can do this now with free software from http://dispcalgui.hoech.net/. This is a front-end to the ArgyllCMS tools which support popular enthusiast probes. I have not seen any proper evaluation of these LUTs+MadVR in action as of yet.

ArgyllCMS + dispcalGUI = steep learning curve :P

Also, MadVR's current implementation of yCMS does not correct points within the color gamut, only the 100% saturation targets. It works well with displays that are fairly linear. My display is NO WHERE close to being linear. Hopefully, the author of yCMS will add the functionality to correct points within the gamut in version 2 of his product and will be integrated into MadVR.

If Calman can add support for MadVR as a pattern source, then it will be a cinch to calibrate HTPC sources.

I have posted instructions here: How to get perfect color

This method certainly does correct points within the color gamut. Look at the results posted. I did not post the charts but I did test with ChromaPure using both Saturation and Amplitude test patterns and both were corrected throughout the color gamut. I usually do not correct black point or white point to keep maximum dynamic range. I'm not too sensitive to the color of 0% Black or 100% White.

Yes, JRiver is the best HTPC solution I have found to playback HD video with lossless multi-channel audio. Their 64 bit audio with optional convolution supports all forms of digital audio processing for room correction, digital cross-overs, etc.

3D LUTs in madVR make the video equally impressive.

yes, the dispcalGUI LUTs do remap inside the gamut.

Interesting....I will try this on my Sharp Elite display which has horrible color tracking for most of its primaries and secondaries.

How many point corrections can DispcalGUI provide? More than the 125 offered by the Radiance sets?

Interesting. My TV controls allow me to get perfect calibration of color at 100% luminance, but it is impossible to get all saturations correct across both 75% luminance and 100% luminance. If I'm understanding correctly, I can use dispcalgui and create a 3DLUT to apply to my display calibrated for perfect 100% luminance across all saturations and the result will be perfect color tracking across all saturations and luminance?

Has anyone here done this with xbmc+madVR? Worth the pain?

You can define any number of LUT points you want in dispcalGUI, the current set of predefined charts include 124 points up to 2386 points.

This is pretty interesting, I tried it out tonight but it looks like dispcalGUI doesn't support doing a calibrate+profile at BT1886 gamma?

Another thing I noticed is that the final result played through madVR seemed to get blue somewhat wrong as "4-Color Clipping" from the avs709 disc showed flashing blue boxes up through 248 or so instead of clipping after 235 like it should. Anyway I'm sure there was some user error along the way as well. Thanks for the guide.

Hi hulkss, I've tried your guide and I've had some problem in the step:
"Click on the Calibrate & Profile button"

When doing it, a series of full field patches in "over the top" mode starts and the rest of windows become disabled, specially the window in background waiting for "next, next,...".
As I have my system previously calibrated with colorhfcr and AVSHD709, can I create only the profile with the "only Profile" button?

If not, what is going wrong with "Calibrate & Profile button"?
Edited by alamagar - 2/24/13 at 5:28am

hulkss,
Can you please post a chart of DeltaL or DeltaE 2000 for your 3dlut corrected gamut saturations? I used your guide and ran calibrations using dispcal and created the 3dluts. When I used them in madvr, although it did lower DeltaH and DeltaC significantly, it produces saturation points that have very large luminance errors. The DeltaE 2000s for my 25% saturation sweep ranges from 5 to 30.

I had this same problem, xy positions were corrected very well (except for the blue channel for some reason) but luminances were far too low. I believe this was because the target gamut is at gamma 2.4 and the patterns used for the saturation sweeps (at least for HCFR) are hardcoded for 2.2 gamma displays. It would be better to be able to do the 3dLUTs for any gamma desired.