Gradient handling is worst when set to 10 or 12bit
Hey RTINGS team!
Today i received my unit and i’ve been doing some testing. Especially with gradient handling there is some weird behavior.
When HDR is enabled the gradient handling is significantly better when running in 8bit mode compared to 10bit or 12bit. When checking my display settings in Windows it’s actually running at “8bit with dithering” (FRC) when set to 8bit. Is it actually possible that 8bit+FRC looks better than native 10bit or even 12bit?
When HDR is disabled there is absolutely no difference in gradient handling between 8bit, 10bit and 12bit with my unit and test pattern. When checking the windows display settings it says either 8bit or 10/12bit, but not “with dithering”.
Basically by far the best gradient handling is in HDR using 8bit. The worst gradient handling is in HDR using 10bit or 12bit. Without HDR it sits about in the middle of the spectrum and there is no difference between different bit depths.
To test this i’m using the “7. Bit-Depth/Precision” test pattern in the official VESA DisplayHDR app.
What settings did you use (both on the TV and on the PC) to evaluate your gradient score?
Can you maybe try to recreate this and confirm that i’m not crazy, blind or both?
Another matter? Are you able to confirm if Calman’s Autocal already works with this generation of TV’s?
Thanks in advance!
- 21010
Given the fact that CSD’s can be misleading given the window type, time vs frequency resolution what bother showing them? What additional information do they reveal that is related to perception and sound quality that a high resolution frequency response curve provides? A high/medium/low Q resonance is visible in the frequency response as long as the FFT length is sufficient and you haven’t applied too much frequency smoothing. Other than providing colorful pictures of waterfalls that look pretty I have never seen the value of these displays.
Hello Daniel, You bring good points. As mentioned in the article, we have had many requests for these CSD Graph. We, like you don’t think they show useful information, but for now it is data that we make available to consult. Note that this is not scored and have zero impact on our products evaluations. We want to bring this to the community, humbly maybe there are aspects that our analysis missed and some information displayed on these graph are relevant. As I said. More questions than answers.
- 21010
Given the fact that CSD’s can be misleading given the window type, time vs frequency resolution what bother showing them? What additional information do they reveal that is related to perception and sound quality that a high resolution frequency response curve provides? A high/medium/low Q resonance is visible in the frequency response as long as the FFT length is sufficient and you haven’t applied too much frequency smoothing.
Other than providing colorful pictures of waterfalls that look pretty I have never seen the value of these displays.
- 21010
Hi Fighterboy. We introduced in this Test Bench Update the performance usages. For objective assessment of audio fidelity, you can look at the audio reproduction accuracy scores. We shifted the focus slightly compared to the old neutral sound score, since true neutrality is not yet fully characterized and defined in headphone reproduction, we still of course value the tonality but to a lesser extent, and have put a bit more weight on the purely objective metrics of audio fidelity. https://www.rtings.com/headphones/tools/table/168449
Edited 2 months ago: more precision - 21010
I noticed that the headphones table tool is missing a fidelity score. Is that ever coming back?