AVS Forum banner

3361 - 3380 of 3408 Posts

·
aka jfinnie
Joined
·
4,448 Posts
It's very encouraging to know the Envy has a calibration process that is simple, that's exactly what the average enthusiast requires
I know this is going to seem like nitpicking, but you have to be careful with words here. I don't think there is any process difference of note between the calibration workflows on Envy relative to Lumagen (if there are I'm really interested to understand them as I'm very interested in all things calibration! :) ). The calibration side is the adjustment of the video output through to the display to a given standard to render an image to that standard, or provide a reference point to tone map within, which requires external sensors and software and an understanding of the colour science, and really has almost nothing to do with the unit itself. It might well be that initial setup out of box is different or simpler (I guess particularly if there is any sort of wizard on the Envy, is there? - maybe someone could do an unboxing and setup so we can see what it is all about?). But that initial setup is decidedly not calibration. :)

The two differences of note that I see relative to calibration process would be that the Envy supports IP patch generation and LUT upload (whereas the Lumagen does require serial or USB connect) and the Envy is supported by displayCAL / ArgyllCMS which is a free option, whereas the Radiance is only supported by commercial software.
I think at a technical implementation level there are some differences in the LUT sizes available in the two products, but that isn't related to ease of calibration.
 

·
Registered
Joined
·
395 Posts
It seems you know well the Madvr platform: do you know what the “assume 4:2:0” setting in the chroma format under incoming video override does? I have a Panny UB9000 and struggling to find the best chroma option for both HD and UHD discs, you know the Panny doesn’t feature a source direct option and res+chroma must be setup manually...
[/QUOTE] Yes, its unfortunate that the Panasonic player don't have a source direct option because of a potential interference of the player with the original source. E.g. if chroma upscaling by the player is not optimal the Envy faces then a signal and situation that might be potentially spoilt by an inferior upscaling. Also the upscaling from 1080p to 2160p by the Panasonic is not that bad (still poorer than that of madVR / Envy). My friend has an [B]Zappiti Pro 4K HDR [/B]that I understand is similar to an Oppo. He uses there 4:2:0 (source direct) and that is what his Envy gets then. Besides the Zappiti he has also a Panasonic 9004 but he prefers now to play his movies from a NAS and not from the physical disc anymore. Not because of poor results but for convenience reasons. So unfortunatelly I can't tell you the correct answer straight away. I would suggest that you take one of chroma upscaling testpatterns (e.g. Spears and Munsil UHD HDR Benchmark disc or R. Masciolas testpattern or the "Quick Brown Fox 4K Testpattern") and see what setting (combination) gives the best results for your setup. I think that is the safest way for you to get the best possible results with your Panny.
 

·
Registered
Joined
·
395 Posts
@WGenzo
Yes, its unfortunate that the Panasonic player don't have a source direct option because of a potential interference of the player with the original source. E.g. if chroma upscaling by the player is not optimal the Envy faces then a signal and situation that might be potentially spoilt by an inferior upscaling. Also the upscaling from 1080p to 2160p by the Panasonic is not that bad (still poorer than that of madVR / Envy). My friend has an Zappiti Pro 4K HDR that I understand is similar to an Oppo. He uses there 4:2:0 (source direct) and that is what his Envy gets then. Besides the Zappiti he has also a Panasonic 9004 but he prefers now to play his movies from a NAS and not from the physical disc anymore. Not because of poor results but for convenience reasons. So unfortunatelly I can't tell you the correct answer straight away. I would suggest that you take one of chroma upscaling testpatterns (e.g. Spears and Munsil UHD HDR Benchmark disc or R. Masciolas testpattern or the "Quick Brown Fox 4K Testpattern") and see what setting (combination) gives the best results for your setup. I think that is the safest way for you to get the best possible results with your Panny.
 

·
Registered
Joined
·
4 Posts
@WGenzo
...would suggest that you take one of chroma upscaling testpatterns (e.g. Spears and Munsil UHD HDR Benchmark disc or R. Masciolas testpattern or the "Quick Brown Fox 4K Testpattern") and see what setting (combination) gives the best results for your setup. I think that is the safest way for you to get the best possible results with your Panny.
Thanks Mori. I’ll give it a try with Masciola’s test patterns.
 

·
Premium Member
Joined
·
460 Posts
The two differences of note that I see relative to calibration process would be that the Envy supports IP patch generation and LUT upload (whereas the Lumagen does require serial or USB connect) and the Envy is supported by displayCAL / ArgyllCMS which is a free option, whereas the Radiance is only supported by commercial software.
Thanks @bobof for clarifying this, i was about to jump on this piece of @roxydog13´s post.
It´s exactly as you´ve described: it´s the whole end-to-end process of getting the Envy calibrated. You start the calibration software, it automatically connects to the Envy to utitlize the built-in pattern generator, running through the process and then automatically uploads the 3D-lut file into the Envy.
Of course you still need a software and a sensor and you need to know what to do to come to a proper calibration.
But the whole process of doing it is really straigt-forward.
 

·
aka jfinnie
Joined
·
4,448 Posts
Thanks @bobof for clarifying this, i was about to jump on this piece of @roxydog13´s post.
It´s exactly as you´ve described: it´s the whole end-to-end process of getting the Envy calibrated. You start the calibration software, it automatically connects to the Envy to utitlize the built-in pattern generator, running through the process and then automatically uploads the 3D-lut file into the Envy.
Of course you still need a software and a sensor and you need to know what to do to come to a proper calibration.
But the whole process of doing it is really straigt-forward.
What calibration SW are you referring to here - is it Calman? I use Lightspace and Colourspace (which I think is generally regarded to have the better LUT engine), and the two steps you refer to as "automatically" aren't automatic for any LUT device in Lightspace / Colourspace. Note I'm not complaining as the user model suits me. I think Chromapure is the only SW I've used that featured an "automatically" uploaded a LUT (and it doesn't really auto upload it though that is the net result - it does an iterative test - tweak - test - tweak cycle of adjusting individual values on the LUT, so the LUT is being put in place during the measurements, there is no profile as such). I don't recall Displaycal doing anything automatically for the EEcolor unit I used with it (had to use external SW for 3DLUT load), though I see there is an upload LUT button for MadVR in displaycal. I've only fleetingly used Calman though, and never for 3DLUT gen.
 

·
Premium Member
Joined
·
460 Posts
What calibration SW are you referring to here - is it Calman?
I used DisplayCAL/Argyll since this is the software most of our customers deal with at first since it´s free.
I have no experience with Lightspace/Colourspace as i´m not a professional calibrator.
It automatically detects the Envy (same procedure as with a madVR PC) and uses it´s built-in pattern generator. At the end of the process, the file is automatically transfered to the Envy.
What really happens behind the scenes could probably be explained by @madshi, but this is how it looks from an end-user perspective (to which i was refering to).
 
  • Like
Reactions: bobof

·
aka jfinnie
Joined
·
4,448 Posts
I used DisplayCAL/Argyll since this is the software most of our customers deal with at first since it´s free.
I have no experience with Lightspace/Colourspace as i´m not a professional calibrator.
It automatically detects the Envy (same procedure as with a madVR PC) and uses it´s built-in pattern generator. At the end of the process, the file is automatically transfered to the Envy.
What really happens behind the scenes could probably be explained by @madshi, but this is how it looks from an end-user perspective (to which i was refering to).
I cut my teeth on displayCAL / ArgyllCMS with the EEcolor boxes, I do like it, though like everything there are annoyances (the UI always bugged me in the way it functioned). Sounds like Envy is well supported by it though. It always struck me as a bit of a shame there is no displayCAL / ArgyllCMS support for Lumagen built in - I keep threatening myself with making an external LUT uploaded for the Lumagen unit but never seem to have the time or the inclination (given I already invested in Colourspace!).
 

·
Registered
Joined
·
976 Posts
I used DisplayCAL/Argyll since this is the software most of our customers deal with at first since it´s free.
I have no experience with Lightspace/Colourspace as i´m not a professional calibrator.
It automatically detects the Envy (same procedure as with a madVR PC) and uses it´s built-in pattern generator. At the end of the process, the file is automatically transfered to the Envy.
What really happens behind the scenes could probably be explained by @madshi, but this is how it looks from an end-user perspective (to which i was refering to).
Is DisplayCAL easy to use in conjunction with the Envy?

What number of swatches do you resort to?
 

·
Premium Member
Joined
·
460 Posts
Is DisplayCAL easy to use in conjunction with the Envy?
Yes, as descibed above. Only flaw i found is that it seems to have difficulties initially finding the Envy via wifi. So i recommend to connect your measuring PC/laptop via LAN the first time, then you can switch over to wifi.
What number of swatches do you resort to?
I´m just using the default - around 400 AFAIR.
 

·
Registered
Joined
·
976 Posts
Yes, as descibed above. Only flaw i found is that it seems to have difficulties initially finding the Envy via wifi. So i recommend to connect your measuring PC/laptop via LAN the first time, then you can switch over to wifi.

I´m just using the default - around 400 AFAIR.
Thanks for that useful information.

Is there a threshold for the number of swatches beyond which little or no visible gain would be realised?

What is the maximum number of swatches?
 

·
Premium Member
Joined
·
460 Posts
Is there a threshold for the number of swatches beyond which little or no visible gain would be realised?

What is the maximum number of swatches?
I think that´s better discussed in a calibration thread. As i said, i´m not an expert on calibration. :)
Technically, DisplayCAL is capable of doing >10.000 swatches.
 

·
Registered
Joined
·
2,577 Posts
Thanks for that useful information.

Is there a threshold for the number of swatches beyond which little or no visible gain would be realised?

What is the maximum number of swatches?
If you like you can do 10000 or more but diminishing returns usually set in earlier and I would not expect significant improvements going beyond 5000 patches in the two cases below. It depends on how linear the display is and how much the native color space diverges from the color space you want to profile for but generally it would have to be a very hard case that needs more than 5000 patches when you can work with the native color space.

Below is a Sony VW590ES that has less than 90% coverage of DCI P3. As you can see average dE is quite low but that took 2681 patches. Average dE was still over 1.5 with ca. 1553 patches.

3046975


When we checked a VW1100ES it already looked very good with "only" about 1148 patches when we mapped DCI P3 to its native color space while the VW590ES looked significantly worse with the same number of patches. Integration with the Envy is pretty good and the software is sufficiently easy to use that I believe it would be a very good option for the end user who owns an Envy and the necessary equipment.
 

·
Premium Member
Joined
·
11,518 Posts
I think there is a white paper that was developed on LUTs a few years back that basically showed that diminishing returns starts to kick in when you exceed 3000 points. But if you have all the time, no reason not to do as many as your comfortable with. The Envy has a 21x21x21 based 3D LUT if I remember right, so you could load a 9,261 point LUT in. Not sure if it has a 1D LUT or not for grayscale/gamma.
 

·
Premium Member
Joined
·
9,906 Posts
The Envy has a 65x65x65 point LUT, I believe extrapolated internally to 256^3. The current 21pt limitation is only the max for standard patch sets in software solutions. You can create larger custom patches if needed. No 1D LUT yet, but some software solutions already allow you to augment say a 17 or 21pts 3D LUT with a 33pts greyscale (or larger). It is an area where Envy is significantly superior to the Radiance, which is limited to 17^3 if I remember correctly.
 

·
Premium Member
Joined
·
11,518 Posts
The Envy has a 65x65x65 point LUT, I believe extrapolated internally to 256^3. The current 21pt limitation is only the max for standard patch sets in software solutions. You can create larger custom patches if needed. No 1D LUT yet, but some software solutions already allow you to augment say a 17 or 21pts 3D LUT with a 33pts greyscale (or larger). It is an area where Envy is significantly superior to the Radiance, which is limited to 17^3 if I remember correctly.
Interesting. Yes, the Lumagen max 3D LUT is 17^3 and a 21 point 1D LUT.
 

·
Registered
Joined
·
2,577 Posts
I think there is a white paper that was developed on LUTs a few years back that basically showed that diminishing returns starts to kick in when you exceed 3000 points. But if you have all the time, no reason not to do as many as your comfortable with. The Envy has a 21x21x21 based 3D LUT if I remember right, so you could load a 9,261 point LUT in. Not sure if it has a 1D LUT or not for grayscale/gamma.
As Manni has pointed out there is enough granularity to accomodate even very demanding cases or to achieve even higher precision. The 65x65x65 3D LUT as uploaded to the Envy is has almost 100 MB and even without profiles the Envy should be able to address 16 3D Lut each for Rec709, DCI and Rec2020 for a total of 48 for these three color spaces. They can all be renamed and stored on the calibrators' computer and then uploaded to the Envy and attached to a user defined backup.

Unfortunately we have to be without our K10a at the moment so it did not make much sense to do any further testing with larger patch sets but this is the greyscale that was achieved with the standard Displaycal patch set of 2681 and without any special provisions or adjustments having been made after applying the 3D lut. Calman has only been used for verification as we found precision and speed superior with DisplayCal in our limited testing with the latest version of Calman. It would have been interesting to see if and by how much precision would have increased with 3, 4 or 5000 patches but coming from a dE of 0.5 the improvements to be had are clearly not that big anymore:

3047109
 

·
aka jfinnie
Joined
·
4,448 Posts
Wow, how long does it take to run a 65x65x65 LUT? May need to prepare the guest room for the calibrator. :)
Chromapure is the only solution I believe where the LUT size has significant impact on the profile time.

This is because my understanding is that for every LUT point it tries different values and accepts the point once a certain level of error is achieved. It's very inefficient and not a good approach from a quality perspective, either. (Last time I used CP was maybe 3 years ago, please correct me if I'm wrong anyone who knows otherwise).

For all other solutions, the driver for the profile size is the behaviour of the display, and not the size of the LUT in the LUT holder. You want to choose a profile size that gives you good coverage of the idiosyncrasies of a display's response to its input signal, while balancing reading only to your meter's capabilities, the amount of time you have available, and whether the display drifts significantly either due to time or indeed stimulus sequence (which may affect the optimum patch ordering - see my sig).

If you have your display in a very linear mode - such as profile off on some JVCs - a small number of patches can get you a very good result (though usually not quite as good as larger profiles). If you are trying to map a display's complex behaviour due to undefeatable internal LUT / colourspace conversion you may on the other hand need a very large profile. And some display modes and calibration software combinations can require a fair bit of work (outside of the profile time) to get really good results (eg Lightspace / Colourspace vs the JVC internal colour profiles).

The calibration software should use clever math to map the behaviour of the display throughout the input value range, and then from that it should be able to interpolate any number of LUT points.
 

·
Premium Member
Joined
·
9,906 Posts
For me the benefit of a larger patch set is more with gamut (especially when using profile off with the JVCs) than with greyscale.

Custom patch sets of the same size provide a more significant benefit than a standard linear patch set. For example, a well-designed custom set of 5K points or less would give similar or better results to a standard 21^3 10K patch set.

The superior granularity of Envy with 65^3 point simply means that there is no hardware limitation in ordre to get the best results.

Of course, if you get a max dE unde 0.5 in a large verification patch (at least 1K points, up to 5K), there is no need to spend more time as long as there are no other artefacts, especially banding or posterisation that are not measurable with dEs.

You don’t need a K10A to use large patch sets. I use a Discus or an i1d3/C6 trained to an i1pro2 and I get 1,000 points read in about 15 minutes with the C6, so a 21^3 is done in about 3 hours and an optimized set set that gives similar results with under 4K points is done in a bit more than one hour. Of course you need to optimise settings and positioning to get these results, but there is no loss of accuracy.

The results could be improved with a larger patch set by the way, so depending on the display there is a benefit with over 3K points, especially if you try to improve visual artefacts such as posterisation that are frequent with smaller LUTs using the PJ CMS when the native gamut doesn’t fully cover the target, and not just raw dEs results that don’t tell the whole story.

I plan to write about this if I find the time once the new JVC f/w is out and everyone can use profile off with the filter and large LUTs.
 
3361 - 3380 of 3408 Posts
Top