Calibrating the LG AU810PB projector with calman

I’ll give this a shot with caveat that I’m not an expert, but just following instructions on AVS Forum. Forgive me, if its a little verbose as I’m just going through the details in my mind.

When we use Calman Home for LG to calibrate SDR BT2020, we select the SDR workflow in Calman, but change the Colorimetry to BT2020. The Gamma curve is set at 2.2, 2.4 or BT1886, depending on room conditions, reflecting a SDR calibration. MadVR is not involved and disconnected at this point.

We chose a mode eg. Cinema and set the pattern generator, either LG or PGenerator to output BT2020 patterns. During Autocal’s colorimetry calibration, Calman uses the colorimeter to read the BT2020 patterns in order to map each RGB triplet sent out by the PG to the correct color output by the projector. It then builds a 3D LUT and writes it to the projector’s memory. So the 3D LUT is not HDR or SDR specific, its just a mapping for that video mode, which happens to be SDR Cinema. All it does is re-map received RGB triplets into others so that the colors are correct and knows nothing about HDR or SDR. Each HDR or SDR mode can have one LUT. If you output Rec. 709 content directly through a LUT calibrated to BT2020, you will get oversaturated colors, particularly red. This is because a RGB triplets for 709 and BT2020 simply reference different colors on the CIE chart. Vincent Teoh of HDTVTest has made a video on this.

So, with our new SDR BT2020 calibration, the source to the projector must be BT2020. We achieve that, by connecting MadVR to the projector set to BT2020. If you tell MadVR that you have a BT2020 display, and send it a SDR stream, it will convert SDR 709 to SDR BT2020 and you will not get the over saturation because MadVR processes the colorimetry from one color space to the other in real time. If you send a HDR stream to MadVR, it will not change the colorimetry, buy will tone map the high nit content to whatever your projector’s calibrated max lumens is (measured by Calman and in my case 47 nits). MadVR thus converts HDR video to SDR BT2020 and mapped exactly to the projectors maximum calibrated lumens.

So MadVR is not involved in the 3D LUT created by Calman that was calibrated to BT2020. The LUT is stored in the projector’s SDR mode. MadVR simply outputs exactly what the projector wants to see, regardless of input to it.

It is possible to put a 3D LUT into MadVR and not use the 3D LUT of the projectors, but you have to use the Displaycal software rather than Calman, as Calman for Home won’t create a 3D LUT file.

Hope this makes sense

1 Like

yes let me try 2.4 next time.
I only watch movie at night so it is complete darkness except the projector and some reflections.
I didn’t do any reflection treatment to my living room at all.

1 Like

Hi Sammy, thanks for the detailed explanation. It all makes sense now . So I was missing this point, on the LG’s and pgenerator’s ability to output SDR2020 patterns at the time of calibration. Also it seems that this PJ is able to store 3D LUT tables for SDR2020 colorimetry. That was the 2 concerns I had. It is all clear now! Thank you again for the detailed explanation


Oh, then u shouldn’t use gamma 2.2. It can be a lot more inky with gamma 2.4 in a dark room. If you have light control with the surroundings perceived contrast will improve to another level

See how much difference it makes from the video below with light control… see how the gray checkered box turns to black …

1 Like

wow, that special curtain seems like a magic
wonder how can install such thing

Thanks for bringing this up. Just got me rethinking what gamma i should use again for dark room. Bright room is clear, G2.2.

My dark environment is all lights off, all curtains drawn (no ambient light shining in), curtains are full block out type. My front wall is dark color, curtains are ash black. But my rear wall and ceiling are white and reflects some light. I’ve been calibrating SDR Dark with G2.4 on my LG C1 OLED.

Can someone recap what are the advantages / disadvantages would we expect between G2.4 and BT1886 ah? I rem the main diff is the the lower end of the Gamma curve where BT1886 is sloping down VS G2.4 is straight flat.

In a pitch black room, where you struggle to even see your own hands… that type of level of darkness I’m talking about, then gamma 2.4 is the way to go… confirmed drool

But it can also look very good with gamma 2.4 if the reflected light is absorbed

I had the privilege to compare BT1886 vs gamma 2.4 in the living room.

When u have light control ( all ceiling, floor, sides) absorbed, gamma 2.4 looks stunning

Without light control, BT1886 looked nicer … I think BT1886 tracks more closely to gamma 2.3, it’s just in between 2.4 vs 2.2

I am now trying SDR 2020 gamma 2.4 with 21points.
going to sleep now as it may take 6.5hours lol

If you wake up after sunrise, you won’t be able to do a post

I wake up at 6, perfect time finished. (It took 6:23 to finish FYI)

Filmmaker mode
SDR rec2020
Iris 7
Brightness 55
Contrast 79

However the result is not satisfy.

Also calman had some issue, after it finished the color calibration, there was a popup saying websocket connection error… I was afraid the 3D LUT cannot be uploaded. But luckily it seems it did.

But I don’t know why the gray scale has shifted up…

Another observation is,
I choosed gamma 2.4 in calibration, but the projector still shows it as BT.1886. the option is dimmed.

Do you know if it is normal? Or we cannot change the gamma during calibration?

I tried to switch to different mode and all are dimmed, and some are switched to 2.2, some are 1886, but none of them are 2.4

1 more thing to share, when using Pgenerator, if we switch between 709 and 2020, make sure to check the projector information (using the … Button) if it is receiving the correct signal. If not, we have to toggle the pgenerator display resolution or HDR/SDR so that the 2020 can be displayed correct. Not sure if it is a bug in pgenerator.
I suspect last time when I did 2020, it was still sending 709 so everything was off.

Even though you aren’t satisfied, that’s a much better calibration than your first SDR BT2020. Was there any daylight leaking in when you did the post? That can affect the chart and cause your Gamma on post to look like that. To verify, when its totally dark, put Calman in manual calibration mode and run a series. I calibrated to BT1886 since I don’t have a perfectly dark room during the day, but its pretty dark at night. I got a good result from Expert Dark, Gamma BT 1886, SDR 2020, Iris 3, but YMMV. I suggest you continue to try different combinations, but the 13pt or 17pt should be enough so you don’t lose too much sleep.

Were you using Wifi? The web socket error is a sign that PGenerator and Calman had a problem communicating, which could have led to errors. In the AVS Forum Pgenerator thread, they suggest a direct wired ethernet connection between the PC and the Raspberry Pi. I used that without issue, but didn’t do a 21pt.

You are right about having to set the pattern generator to BT2020. If I recall, even if you use the LG generator, you have to do that through one of the tabs in Calman. In any case, even if you don’t have super low dE, at an average of 2.5, it should be quite watchable as its below the supposed human ability to discriminate. And besides, it would be still much better than the Matrix LUT in HDR, which takes 30 seconds so its still worth doing this and getting MadVR to convert HDR to SDR. Try setting MadVR to a BT2020 display and take a look.

1 Like

Yes it was still same dark when I do post calibration, it was just 6:15am this morning lol. Actually I prefer doing 21points as I can sleep for 6hrs rather than 3 . Lol.

I guess the websocket timeout is from the LG projector rather than the pattern generator.
Because after the calibration I cannot set brightness or contrast via the calman. I need to disconnect and reconnect the projector.

I am using wifi on projector, but LAN for my pgenerator.

probably I should use LAN for projector too
Or may be the 6hrs calibration makes the projector wifi sleep in the middle.

Tonight I may just do BT 1886 rec2020. as my room is not completely dark as I don’t have reflection control.

During the calibration with PGenerator, the network activity is from PC to PGenerator and then PGenerator to projector via HDMI. Calman would be communicating with the projector only if you were using the LG iTPG. However, at the end, Calman must write the 3D LUT to the projector over the network. If the error was during the 3D LUT write, then data may be corrupt.

Are you connecting to PGenerator directly from Calman PC with an ethernet cable, or are you going through a switch? I don’t know if it makes a difference, but the PGenerator people say its more reliable to connect directly and cite this web socket problem.

Just to check, are you running the HDR file off USB stick from Portrait Displays for the full duration of 6 hours? You have to concatenate the 1 hr HDR files 6-7 times to get it long enough. I also had good results feeding a HDR test pattern from my Panny UHD player instead of using the USB stick.

both the PC and Pgenerator are using LAN cable connecting to the same switch.

Since I use PGenerator, I don’t use any USB stick to play any files on the projector.

Although I can’t really see why it should be a problem, recommendation from AVS Forum was to connect the PC and Pi directly without a switch.

On Pgenerator, yes sorry, I forgot… been awhile…

1 Like

SDR 2020
Gamma 2.4
Iris 7
Brightness 58
Contrast 79

Found that I need to do a manual reset in the projector menu in order to unlock the gamma and set to 2.4 manually,before I connect it to Calman and do DDC reset. Otherwise it will be locked as above.

Result is good.

Also, I am wondering if we should actually set the brightness and contrast before the calibration.

As in calibration mode with DDC reset, brightness and contrast is 50/80 by default.
After calibration , if I don’t change anything, the grayscale will remain at 100 when brightness is 50.
But when I adjust it to 58 , the post cal grayscale will all go up a bit.

Below is madVR 113. set the nits to 50 or 100 doesn’t change much, as good as this

Below is using latest madVR (not 113), the nits is set to 100 but some details are gone

Yea that is true, if you had it calibrated before, you will need to reset it to run a new calibration .

For SDR, Normally you will need to adjust brightness and contrast with the correct patterns before you start the autocal process. In mine, I tested with both brightness and contrast patterns and found that settings for SDR with 50 brightness 85 contrast was the best in my room and eyes.

This is something you need to do with eyes and a specific brightness/ , contrast pattern.

The contrast setting is a bit more challenging, you need to balance it to ensure you don’t clip white with too high a setting. Too low and it will look too dim. This is something you need to balance with iris setting on the PJ before you start your autocal process

I see.

Then the calman workflow is very misleading, as they put this adjustment at the end of the calibration.

I will try again , thanks

You will need to get some of the basic settings done first before you attempt the calibration. That means disabling all processing like dynamic contrast, dynamic tone mapping(HDR). super resolution, noise reduction etc etc. all need to be disabled .

You then need to choose the colour temperature mode that is closest to D65 point. I think they have cool, warm, warm 2, warm 3 … something like that. Set the brightness, contrast and sharpness with external pattern with eyes. all these will not be set by calman software.

calman’s algo will focus on 1D LUT grayscale and 3D LUT colour based on the gamma and colour space you choose

Whatever you do on contrast setting. do not clip white. It’s ok to have a dimmer image but don’t sacrifice white. At no point should below 98% white be invisible, if it is; you have clipped white and some of the images like snow, clouds, sun shine will lack the details. With the laser Pj, this is not a problem for 150” screen and below.

So always keep in mind, you can always compromise somewhat on light output and white balance(RGB High Low), but you should never overdrive the display and allow it to clip white.

Once you have gotten the above correct, only then you start your SDR calibration.

I would opine that in this shot, MadVR is simply not working. I haven’t tried it myself, but AFAIK, the latest version of MadVR does not work with Videoprocessor. Perhaps its just using the default renderer instead.