- This topic has 7 replies, 2 voices, and was last updated 1 year, 5 months ago by
.
-
Posts
-
Hey there, everyone. I would like to ask regarding proper workflows for calibrating for a LUT box. Basically, here’s my signal chain: Macbook M1 Max -> Ultrastudio Monitor 3G (via thunderbolt) -> Blackmagic SDI to HDMI 3G Micro Converter (via SDI) -> BenQ PD2700U (via HDMI). Previously, I didn’t have the converter in the middle of the signal and just calibrated using the guide provided here in DisplayCal to generate a 3D lut for Resolve and put it inside the software. Now, I’m borrowing a converter from work to try out a new workflow so I can store the LUT inside the converter to act as a LUT box.
First of all, is my thinking proper when I say: I use the monitor for both color grading and normal usage, so sometimes i use the ultrastudio and sometimes not. I’ve calibrated for both. But when I use the grading signal chain, the reference monitor is in its natural state correct and the only correction i CAN apply is via the resolve settings? But as far as I’m aware, this compared to proper hardware calibration is still lacking?
Anyways, here’s what’s been my settings (based on the 3D lut for resolve base settings):
- On the calibration tab, i changed the tone curve to Custom – 2.4 – relative, but black output offset at 0% (Can someone help to explain what black output offset actually does and at what % it should be most of the time? + I turned off black point hue correction
- On the 3D lut creation tab, changed the tone curve as well to Custom – 2.4 – relative, output offset at 0%. Here i’m still applying calibration (vcgt), should i be doing that? Then i set the output encoding to video levels (TV 16-235) and 3D lut resolution is 33.
On that note, I’d like to ask video vs full levels. When to use which? 3D lut res is 33 because the LUT box can only support 33×33 max. My main outputs are for broadcast and web, so around 2.4 or 2.2. But mainly i focus on calibrating to rec709 gamma 2.4 at 100-110 nits.
What am I doing wrong? Since when the calibration is finished and I load the LUT into the box, the image just went nuts and lost like 50% of the contrast. Before when i set it to 2.4 absolute, the contrast only decreased slightly. When i took a measurement report with devicelink profile, i believe it showed it produced a 2.2 gamma instead. I’m trying to do things manually and just try different settings out but if anyone can point me in the right direction, that would be of much help so I can really understand what’s going on in each of the settings.
Attached are my full settings and results from displaycal
2.4 gamma relative is gamma 2.4 with black compensation. 0% output offset means it put the black grey at 0 in the gamma curve at 0%. Black is raised and is good on a low contrast monitor. 2.4 gamma on a low contrast monitor is not good. It will look dim at low nits. 100 nits is sdr nits. 80nits is srgb. Before with absolute and 1000 contrast ratio 2.4 absolute with 0 output offset was bt1886. 1886 on 1000 contrast screen is 2.2 gamma with a correction at the low end to make blacks visible. It goes up to 2.3 almost. It does change with contrast of the screen. What is your contrast ratio? Correcting black levels will lower contrast ratio unless black levels are calibrated hardware. Hardware is 2.4 or is 1886 ?
Hey, Ben.
First of all thanks for replying. My display is a BenQ PD2700U, not a reference broadcast monitor by any sorts, but I do want to calibrate it to the broadcast standard since my output is either that or 2.2 web delivery. The contrast ratio is around 800-1200 : 1 in my previous calibrations.
Thanks for explainig some of the settings. Does that mean i should be sticking with my choice of ‘2.4 gamma relative with 0% output offset’ since it will map the black on my output display to be the 0% of the curve? I’m still a bit confused on how different 2.4 and 1886 is. You said that my previous attempt “2.4 absolute, 0% output offset” on a 1000:1 contrast ratio is bt1886, which is 2.2 gamma with a correction in the low end (Thanks for this explanation, it explains the measurement report results!). So what is right? What should I be aiming for if I want to achieve the broadcast standard (rec709 gamma 2.4 at 100 nits)? I know pure power curve for an LCD display like this is not possible.
P.S.: Just before calibration when looking through this signal chain, my macbook m1 max and benq look similar (if you disregard the difference in black level capabilities). But when i import the LUT via this LUT box, so far they’ve made the image lose contrast, in any of the settings i’ve tried. With the official guide for 3D Lut creation for resolve, the LUT i’ve inputted inside the resolve software is already pretty good and mainly just adjusts the hue of the colors, contrast is good and still pretty similar to my macbook m1 max for what it’s able to. I just wanted to see if using a LUT box that’s more proper/similar to a hardware calibration can make the image look more accurate as opposed to doing it internally via the software settings.
Is your monitor in limited range 16 to 235? The 3d lut is set for output in limited.
They do not have a 2.4 gamma preset. 2.4 gamma relative with 100 black output offset is correct I think. Make sure there is no lut in the blackbox when running a calibration with it in the chain.
Finding the clues on how to do this. I think even native gamma would work and the 3d lut would fix it to 2.4 since its a unconventional video card. https://hub.displaycal.net/forums/topic/how-do-i-calibrate-my-display-to-rec-709-gamma-2-4-what-are-the-exact-settings/
I was right about 2.4 relative 100% output offset. https://hub.displaycal.net/forums/topic/gamma-2-4/ . Hope you keep your room light dark. I do not like that gamma for watching. 5% is only 0.21 nits and 0 is 0.119 at my 120nit brightnest. 1886 is 5% at 0.521 nits. Both are still visible but light really mess up the picture and I need to see everything in dark in some games.
I think you can do 2.4 gamma on a 1000 contrast display. Yours is better. I have a 2.4 gamma setting and it actually hits some 2.4 power law nits at the low end in its native gamma with 2 point white balance. I watched some streaming shows really good on that when the peak brightness was high.
Does the blackmagic sdi have a video signal out range adjustment? It should be the same as the video card.
Thanks for clarifying, Ben. Really appreciate this! As for the signal, since i’ve not been encountering any issues as of now I think i’m pretty good. In my few tests today, i’ve done it consistently so the LUT is full level and also the lutbox itself, the monitor can handle up to full if i saw the tech specs correctly, and then clamp everything to video level via resolve. Mainly to make sure i’m not ruining anything else with this new ‘lut box’ setup, so full just to be sure.
As for the calibration, yes visually the 2.4 gamma relative 100% offset did give me the best results. But when verified, they gave a delta e of around 1.5 (average) and 6.5 (maximum), which was kind of good but discouraging for a color grading monitor (mainly the maximum). That’s why currently i’m trying out several settings to compare between absolute and relative and several output offsets. So far:
– 2.4 gamma, relative, 100% offset have given me the best results visually, not so much on the report
– 2.4 gamma, absolute, 20% offset gave me an image with decreased contrast (maybe to a gamma 2.2 ish), so the image looked brighter and flatter on the shadow areas. But the report was good (both cases: 1) no lut on lutbox + devicelink profile 2) lut on lutbox, no devicelink profile).– 2.4 gamma, absolute, 0 and 10% offset gave me similar results and measurement reports, varying slightly on the maximum and average, but haven’t gotten around to checking how different/similar exactly.
– later today, I’m going to try out 2.4 gamma, absolute, 50% and 80% offset. Then compare those to the 2.4 gamma, RELATIVE, in 0% 20% 50% 80%.So yeah, i’m going the trial and error route since I currently have the time for it. But seeing as you’ve also supported the 2.4 gamma relative 100% offset method, maybe I’ll revisit it first and see if i did anything wrong in the process. I’m mainly going to be comparing via the measurement report results (for every calibration, i exported the 2 measurement reports like i said before).
If you have any more tips, please do let me know. It could really save some time haha!
It would take some time to get it right. I am not sure of any tips. Relative should work same as absolute. It could be raised blacks which is a levels problem. I would keep the video card levels the same in the chain. That is extra conversions of full to limited. The conversion messes up one value. One device is not reading 0,0,0 right.
Thanks for the tips, Ben. I’m proud to say i’ve done all of my iterations (after a long day of full-on calibration!) and the results came out like you said. Relative gave me similar results to absolute when using the same black output offset. I did comparison measurement reports with 0%, 20%, 50%, andd 80%. While all showed good average delta e values < 1 and maximum values < 3 (which has made me really proud to be able to do so!), only the contrast ratio varied (which checks out with what you explained the black output offset does) It ranged from 600:1 all the way up to 1000:1.
I then did some perceptual matching to see which contrast fit my case the most (one that resembled closely to what i see on my calibrated macbook). Went with the 2.4 relative with 50% offset (similar result to 2.4 absolute 50% offset).
