Banding with streaming sources

[charles d]

Hello,

I’ve been learning to create 3d luts with Argyll/DisplayCAL for an LG B6 oled using an eeColor box.   I’ve been getting very good results, especially with high quality sources like blu rays.  My issue is with some streaming content  that is inherently noisy, like “House MD”  on Amazon instant video .  After applying a lut, mostly uniform background noise in out-of-focus areas (like walls) will sometimes show more noticeable banding.

I’ve been experimenting, trying to reduce the banding, but I’m wondering if this might be a limitation of the TV’s stability during profiling, and the banding is just more obvious  with the noise/artifacts of a more compressed source.    I’ve been using 1500-2000 patches for profiling.   Excluding errors on my part, the profile self checks have had consistent results with the average dE around .5, max between 5 and 6, RMS around .5.   The max is down from around 9 dE after I disabled the TV’s static ABL.  My extended verifications are good, with a .5 average dE and max under 2.

I’m using a Colormunki Display with a correction matrix for a LG B6.  I’m running Displaycal/MadTPG on a Windows 10 laptop.  My source device is a Sony BDP-S5100 blu ray player.

Any ideas on if this sounds like a display stability limitation or something else that can be changed to reduce or eliminate the banding? I appreciate any suggestions.

Thanks!

Calibrite Display SL on Amazon  
Disclosure: As an Amazon Associate I earn from qualifying purchases.

[Florian Höch]

Hi,

I’ve been getting very good results, especially with high quality sources like blu rays. My issue is with some streaming content that is inherently noisy, like “House MD” on Amazon instant video . After applying a lut, mostly uniform background noise in out-of-focus areas (like walls) will sometimes show more noticeable banding.

If the banding/noise is already part of the source material, there’s little you can do. That the results are good with high quality sources seems to confirm this.

Excluding errors on my part, the profile self checks have had consistent results with the average dE around .5, max between 5 and 6, RMS around .5.

Good but not exceptional, there may be an issue with long-term display stability like you said – this probably won’t affect apparent banding with the LUT applied, but it has an effect on accuracy. Have you tried white level drift compensation? The next version of DisplayCAL will also have some additional patch sequence options (currently, the sequence of patches is always optimized to minimize the cumulated display response delay and thus measurement time), which may help with displays with unstable light output. I’ll probably make a beta available later this week in case you want to try it out.

[charles d]

Hello,

Thanks for your response.

I’ll probably make a beta available later this week in case you want to try it out.

Yes, that would be great.

Have you tried white level drift compensation?

Yes, I started using white level drift compensation earlier, which helped accuracy.  I do wish I could get the self check max dE down a bit, but since my max dE on the large verification test chart seems to stay under 2, I figured it was alright.  I imagine larger test charts might show larger errors.

I attached some jpg’s of the banding with and without the lut applied where you see the rougher color shifts.   A Granger rainbow looks good, but I do see a bit more banding on a grayscale ramp, but not terrible.

Thanks again!

Attachments:

You must be logged in to view attached files.

[charles d]

Also, here’s my latest profile.

Attachments:

You must be logged in to view attached files.

[Author]

Posts