[Rolls-Royce]

I calibrated this evening the 65C8 of my good friend with a custom OFPS patch set optimised for rec709 colorspace. I aimed 100 nits so before AutoCal I set OLED light via DDC to 24 which results in ~125 nits. So the peak output drop after AutoCal is about 20% this is important to keep in mind.

After AutoCal I had to touch up some spots here and there manually to flatten things out.

I discovered the exact same problem I had with my C8 at 15 IRE with the little bump. I can't fix this manually because the 42p DDC control let me only adjust 13 and 16 IRE.

I also think that the TV doesn't load the LUT correctly because at 100% saturation the colors are off because of the wide gamut. It doesn't map rec709 it's wider. As you can see on my scans I ended up at 106% gamut coverage.

Another strange thing which I discovered is that CalMAN mapped 4-5 times the points to the 3D LUT during the measurements is that normal?

I attached also the cube after loading the 3D LUT into Lightspace and I discovered that 2 yellow points are way off - same with mine C8 maybe Tyler can say something in addition to that.

I know that pictures from the phone aren't really reliable but they show pretty impressive how good this "small" patch set works instead of 3000 or even more iRP LUTs.

Hope that portrait displays is going to improve the algorithm further.

BJ

Honestly, it looks pretty darned good already. I wouldn't worry about that bump at 15%. As you said, the DDC adjustment points are to either side of it, so not much can be done. Besides, can you really see a difference between 0.3 and 1.3 dE at that color and luminance level? Possible but doubtful. As to the program mapping more points to the 3D LUT than are measured, unless I misunderstand your question, this is entirely normal. With a 33x33x33 LUT, doing 1:1 readings would take a whole lot of time. On the other hand, sometimes the program will make more or less measurements than set by the user. I think this would depend on how many valid vs. invalid measurements the software is seeing during the display profiling. What I mean is that the display must fall into a certain performance envelope to be correctable by a 3D LUT. If some readings are outside that range, for whatever reason, I think the program might be taking additional ones in order to get a more accurate picture of the display's condition.

 

[zielin]

I will be making video tutorials this week starting with SDR.

Tyler

I'm looking forward to those videos. I'm just doing basic SDR and the 1D greyscale LUT is crap around 25%. (over -20). I am using PC (0-255) because that is what most of my sources are putting out. Other than that i'm just following the template. ABSL has been disabled.

I'm also using a 77" - which from Vincent's video would suggest that it is a 2017 screen.

 

[WiFi-Spy]

I'm looking forward to those videos. I'm just doing basic SDR and the 1D greyscale LUT is crap around 25%. (over -20). I am using PC (0-255) because that is what most of my sources are putting out. Other than that i'm just following the template. ABSL has been disabled.



I'm also using a 77" - which from Vincent's video would suggest that it is a 2017 screen.

We will be adding full range AutoCAL support soon. It was one of the things that wasn’t able to be completed in time. I can do some testing in the next day or two to see if there’s a workaround.

 

[zielin]

We will be adding full range AutoCAL support soon. It was one of the things that wasn’t able to be completed in time. I can do some testing in the next day or two to see if there’s a workaround.

Thanks for the update! So i shouldn't be using 0-255? (at least for now) Its just real odd. It is only in the mid 20s.

I'd be interest if some of the pros here could try a 77" and see if it's "different" screen has and impact on calibration (or expectations)

EDIT: oh yeah. Much better with limited dynamic range (16-235), but i'm looking forward to Full.

 

[natjencks]

Hi folks, I have a c8 which I'm attempting to autocal for SDR using CalMan Ultimate and a Klien K10A (with offsets built with a PR670), pattern generator is DaVinci Resolve. I'm using the ISF Dark slot.

Is there any advantage to using one slot over another, for example using the ISF dark vs the Technicolor Expert slots?

My main question is regarding grayscale point selection.

Assuming time is not a limiting factor, which Grayscale pattern set is recommended?

What are the differences between the normal and "LG" grayscale point sets? What about "video" ranges vs SMPTE (non-extended). In the pro video world we usalls simply have VIDEO range which is the same as SMPTE and extended (FULL) range.

When I select and of the "LG" Grayscale series it goes haywire with huge errors and us unable to hit target deltaE. Seems to do fine with either 28- SMPTE or 21 - VIDEO settings.

Thanks!
-N

 

[jrref]

I've seen this when the new DV config file doesn't update properly. I usually try to re-load the same file and if I don't get the prompt, I change the date on the file and then re-load it.

 

[BlackJoker]

Tyler maybe you can answer some of my questions I posted above. I also want to know if there is any noticable difference if we using a different cube formula for calculation for example dE calibration max oder JNDab can we expect better results?

 

[robbeh81]

I've seen this when the new DV config file doesn't update properly. I usually try to re-load the same file and if I don't get the prompt, I change the date on the file and then re-load it.

I‘m pretty sure it loaded well.

 

[Anger.miki]

I'm also using a 77" - which from Vincent's video would suggest that it is a 2017 screen.

Vincent, in his review 77C8, not only said that the 77" subpixel structure is the same as the 7 series (B / C / E / G / W7) but also clearly stated that he would not use the new features (not even 3D Lut autocal.) Therefore, after rtings.com, even HDTVtest (Vincent Teoh) came to the same conclusion, suggesting (more or less implicitly) to the enthusiasts (read: visual perfectionists) to buy a 2017 model and saving a lot of "lilies". Gamers apart, what do you think?

 

[jrref]

There are people who follow what the professional reviewers say to the letter and there are those who consider the information they present along with everything else and then come to their own conclusions. As the years pass and this technology matures, the delta between the sets year to year gets smaller and that's what you are seeing. I believe this is why Sony didn't release a new set for 2018. (The A8F is really only a stand change).


I would get a 2018 LG or Sony just for the screen uniformity alone. Virtually no banding and tinting on the 2018 panels. This is a huge improvement!

 

[gadgtfreek]

Thats for sure, uniformity has been a issue for awhile, nice to know they made a step forward.

 

[Anger.miki]

As the years pass and this technology matures, the delta between the sets year to year gets smaller and that's what you are seeing. I believe this is why Sony didn't release a new set for 2018. (The A8F is really only a stand change).


I would get a 2018 LG or Sony just for the screen uniformity alone. Virtually no banding and tinting on the 2018 panels. This is a huge improvement!

Hi John, I see a bit of contradiction but I got your point. So you think that next year the delta will be even smaller, don't you?

 

[jrref]

For next year, my personal guess is that we may see sets capable of "native" 240Hz or higher refresh rates vs the current 60Hz/120Hz and if that happens, it will be a major improvement from what we have now as the baseline is raised to the next level.

As for my comment above, I meant PQ differences become smaller as the technology matures. Banding and tinting falls into another category lol.

 

[zielin]

Vincent, in his review 77C8, not only said that the 77" subpixel structure is the same as the 7 series (B / C / E / G / W7) but also clearly stated that it would not use the new features (not even 3D Lut autocal.) Therefore, after rtings.com, even HDTVtest (Vincent Teoh) came to the same conclusion, suggesting (more or less implicitly) to the enthusiasts (read: visual perfectionists) to buy a 2017 model and saving a lot of "lilies". Gamers apart, what do you think?

That isn't why they suggested 2017 over 2018. They suggested it because for little extra, for most people, it is better to save money and get last years model. That said, tone mapping for game mode and autocal aren't on 2017. Also the 77" c8 is cheaper than the g7/w7.

Vincent also said he didn't think autocal was worth it because of the time needed and he'd have to stay over at someone's house. He also didn't like the fact that he couldn't save it and it locked the settings. All, other than maybe the last one, are non issues for someone that owns the equipment and TV.
If you want to watch his video again, he talks about autocal and 3d lut @ 6m30s

 

[jrref]

BTW you can autocal the CMS using the Matrix LUT for HDR and it will do just as good of a job as you can do manually without any artifacts. What Vincent was saying was if you want to get better results than a manual calibration, you need to do at least a 1000 pt CMS calibration, ( 1100- 1200pts is really what you need ). With a fast meter like a Klein K10A, I was able to do this CMS calibration in about 45 minutes but realize this is only for one PM. When you multiply this over many picture modes, this is where it gets impractical to do in the field.

I still believe you get better overall PQ calibrating with Autocal because you are able to do the 42 pt gray scale which really "nails" the lower IRE blacks and even the Matrix and or Lightening LUT seems to resolve any "crazy" CMS issues on some sets vs spending a lot of time manually. If you know how to use Autocal, you can get a slightly better PQ result in less time then calibrating manually. But as Vincent says, you still can achieve a fantastic PQ result calibrating the C8 manually.
In talking with Spectracal, i'm convinced we will see an autocal backup at some point.

 

[gadgtfreek]

BTW you can autocal the CMS using the Matrix LUT for HDR and it will do just as good of a job as you can do manually without any artifacts. What Vincent was saying was if you want to get better results than a manual calibration, you need to do at least a 1000 pt CMS calibration, ( 1100- 1200pts is really what you need ). With a fast meter like a Klein K10A, I was able to do this CMS calibration in about 45 minutes but realize this is only for one PM. When you multiply this over many picture modes, this is where it gets impractical to do in the field.

I still believe you get better overall PQ calibrating with Autocal because you are able to do the 42 pt gray scale which really "nails" the lower IRE blacks and even the Matrix and or Lightening LUT seems to resolve any "crazy" CMS issues on some sets vs spending a lot of time manually. If you know how to use Autocal, you can get a slightly better PQ result in less time then calibrating manually. But as Vincent says, you still can achieve a fantastic PQ result calibrating the C8 manually.
In talking with Spectracal, i'm convinced we will see an autocal backup at some point.

If you run the 42pt, and lightning LUT, how long does that take for each cal mode?

 

[jrref]

If you run the 42pt, and lightning LUT, how long does that take for each cal mode?

About 45 minutes with the K10A

 

[zielin]

About 45 minutes with the K10A

wow. the K10A is a monster then
For the C6 is just 45 for the 42 points. I'm doing 3500 point SDR LUT now and yeah... this is going to take some time.

 

[mrtickleuk]

Vincent also said he didn't think autocal was worth it because of the time needed and he'd have to stay over at someone's house. He also didn't like the fact that he couldn't save it and it locked the settings. All, other than maybe the last one, are non issues for someone that owns the equipment and TV.
If you want to watch his video again, he talks about autocal and 3d lut @ 6m30s

Yes. Being able to give the user saved settings in case they ever need to restore their TV to factory defaults and want to be in a position to restore the calibration they had, is critical to his business. At the moment, a traditional calibration and giving the user the values to enter themselves in a menu which isn't locked out, is the only way he can do this. He also said that he'd fed back to SpectraCal and LG and my impression is that some kind of backup will [have to] be added.

He didn't just say he'd have to stay over! He said he'd be eating your food, and then there was a strange jump-edit in the video

 

[natjencks]

So I ran a 10,000 point profile overnight and in the morning I just got a window saying "ERROR: 300".

So then I ran the profile again this time using 3000 points. This completed in about 3 hours, and all seemed to go OK, but after running the 3DLUT measurements, the monitor is still quite far out when I use the color checker... primaries are way too saturated and deltaE is hitting an average of 3. Is it possible the 3D LUT was not loaded into the TV or was not loaded correctly?

Does one have to load it manually?

What is the "retarget 3D LUT from profile" button, and what is the "Load 3D LUT" button (inside the DDC menu).

Thanks for any advice.
-N