AVS Forum banner

1 - 20 of 33 Posts

·
Registered
Joined
·
45 Posts
Discussion Starter #1
I've calibrated 2 point white balance using this app and a grayscale gradient (30% and 80%). Any reason this won't work for all 10 points? I'm using a reference white alongside the grayscale gradient to calibrate the app and trying to get close to the numbers below.

Understand this won't match the accuracy of a dedicated monitor, but better than nothing?

https://play.google.com/store/apps/details?id=com.loomatix.colorgrab&hl=en_US
In settings, change left to RGB, right to Grayscale

https://www.mulita.com/images/11-step-gray-scale-with-brightness.jpg
Cast using chromecast

0% gray = RGB 0-0-0
10% gray = RGB 25-25-25
20% gray = RGB 51-51-51
30% gray = RGB 76-76-76
40% gray = RGB 102-102-102
50% gray = RGB 127-127-127
60% gray = RGB 153-153-153
70% gray = RGB 178-178-178
80% gray = RGB 204-204-204
90% gray = RGB 229-229-229
100% gray = RGB 255-255-255
 

·
Registered
Joined
·
45 Posts
Discussion Starter #2
I should mention for anyone that wants to try this themselves, 30% gray for offset, 80% for gain. Adjust red and blue first to get the numbers close then tweak green after. Final result should have all three values in the app read the same (R G B).

Have to hold the phone a couple inches from the screen and make sure there isn't weird glare. I ended up moving it different angles until it all read the same.
 

·
Registered
Joined
·
45 Posts
Discussion Starter #3
While I'm waiting on people to chime in a call me an idiot for posting this thread, the 10 point calibration I completed looks great (subjective obviously). I was able to get all 10 points RGB values equal, and everything looks balanced and lifelike. I'm lukewarm on the ks8000 for its abysmal motion handling (smeared panning or microstutter with AMP enabled), but the color representation and HDR are fantastic. Bias lighting and smart LED solved the corner light bloom.
 

·
Registered
Joined
·
5,325 Posts
This method is totally dependent upon the accuracy of your device's camera and its settings. Since this is essentially unknown, it's really not much more accurate than eyeballing it.
 

·
Registered
Joined
·
45 Posts
Discussion Starter #5
That's a good point, so my panel is essentially tuned to my cameras output. Can't say I'm too disappointed, it's markedly improved from out of box, and the pixel2 is considered one of the better smartphone sensors.

I'll get my DSLR and compare accuracy to what my smartphone read, see how close it got. But the premise of balanced rgb values at each gray step is conceptually accurate?
 

·
Registered
Joined
·
5,325 Posts
Maybe conceptually, but almost certainly not realistically. If your camera is balanced for D54, for instance, the balanced RGB values would likely give you a white point of D54. Problem? Video content is color corrected on monitors calibrated to D65. Watching with any other color temperature will give you wrong colors, particularly in the secondaries.
 

·
Registered
Joined
·
45 Posts
Discussion Starter #7
Makes sense, I planned to custom white balance the slr (18% gray card) with my 6500k bias lighting to get it close to 'spec' first. Will see what the rgb metering thinks
 

·
Registered
Joined
·
45 Posts
Discussion Starter #8
So what did we learn with the DSLR? Well first, setting the white balance on camera to 6500k (canon 80d btw) resulted in very strange colors. The TV looked like neutral grays to my eye, but the camera showed heavy red, or blue depending on the white balance selected.

Tried a different approach, set custom white balance to the 80% gray on the screen and checked my phones calibration settings. Viola, everything lined up pretty spot on! But only down to 40-50% white. Setting the camera to custom white balance at 30% white resulted in the same on the opposite side, 10-40% were pretty darn close, with some color shift up the scale. Using 50% white as the custom white balance was pretty spot on all around.

So what does that mean... maybe I should have just stuck to 2 point and 30% and 80%? I feel pretty good with what i'm seeing on the 10 point, but some values just can't get adjusted enough to be perfect (20% white is maxed 50, -50, -50). This is probably operator error, should I have reset my 2 point before doing my 10 point? Maybe I'll try when my eyes recover from spending this much time 8 inches from the TV.

Whatever the case, watching normal cable everything looks "right" where I previously struggled with reds that were too hot and blues that looked muted. I wish I could display that fancy calman graph everyone has, but for the low low cost of $0 I feel pretty OK with things.

For those without a DSLR that can do fine tuned white balance and RGB histogram I found the camera sensor got very close, so any fine tuning with the SLR was just extra.
 

·
Registered
Joined
·
5,325 Posts
Part of the differences you're seeing may be due to nonlinearity of the camera sensor. Or choosing 6500K instead of D65. You see, 6500K can range from a cyan- to a magenta-tinted "white", while D65 is a discrete point on the black body curve. Michael Chen @MichaelTLV posted a series of photographs of the same subject illustrating different color tints - all 6500K - a number of years ago.
 
  • Like
Reactions: agd

·
Registered
Joined
·
713 Posts
Okay, so let me see if I got this straight, because this is rather convoluted. You used your phone camera that isn't calibrated, at any particular known reference, or with any linearity or other sensor issues (like noise) taken into account, to adjust the white balance settings on your TV so that your phone camera (not your eyes) read grayscale patches as neutral. Then to confirm if you did it right you took your DSLR, which isn't calibrated to any particular known reference, and haven't accounted for any sensor issues like noise or linearity, set it to 6500K and when it didn't give you the results you wanted you custom white balanced your camera to the gray scale patches on the TV itself (the whole point of white balancing is to remove any colour cast that might be present in a white object) and are now happy that the white balanced photos you took with your DSLR are showing neutral grays.......












I'm not sure I know where to begin! :p
 

·
Registered
Joined
·
45 Posts
Discussion Starter #11
Okay, so let me see if I got this straight, because this is rather convoluted. You used your phone camera that isn't calibrated, at any particular known reference, or with any linearity or other sensor issues (like noise) taken into account, to adjust the white balance settings on your TV so that your phone camera (not your eyes) read grayscale patches as neutral. Then to confirm if you did it right you took your DSLR, which isn't calibrated to any particular known reference, and haven't accounted for any sensor issues like noise or linearity, set it to 6500K and when it didn't give you the results you wanted you custom white balanced your camera to the gray scale patches on the TV itself (the whole point of white balancing is to remove any colour cast that might be present in a white object) and are now happy that the white balanced photos you took with your DSLR are showing neutral grays.......












I'm not sure I know where to begin! :p
You got it! Use the phone sensor on gray patches and got them all consistent. So it's calibrated to something, which I think we all agree is undetermined and maybe not linear, but something. And then confirmed with my camera that it is indeed calibrated to something consistently.

Is it calibrated to d65? Doubt it, would be neat to find a way to set white balance on my slr to set this. Noise isn't an issue, it's a high end SLR so I have control of that-- noise was low. But it is calibrated to something and is more balanced all around. I think 10 point was probably too much because of linearity (will try to work around this), but 2 point balanced to a 30% and 80% reference gray would result in a "flatter" calibration correct?

I think my idea here is that without cal equipment I can't reliably get d65 (6500k didn't work as mentioned). But with my equipment I can get everything aligned to a common reference point so colors are more accurate. Which I did, and looks subjectively better to my eyes. Thoughts?
 

·
Registered
Joined
·
5,325 Posts
You got it! Use the phone sensor on gray patches and got them all consistent.

But with my equipment I can get everything aligned to a common reference point so colors are more accurate. Which I did, and looks subjectively better to my eyes. Thoughts?
You mean "so colors are more" consistent, not necessarily accurate...
 

·
Registered
Joined
·
45 Posts
Discussion Starter #13
You mean "so colors are more" consistent, not necessarily accurate...
Right, so my hypothetically my phone/camera confirmed it was (close to) consistent with say d54 instead of d65, there aren't crazy reds or muted blues, it's just isn't accurate to the director-defined/studio-produced 65?


I can hard code a (6500k) white balance setting in my dslr, so I'm currently trying to research a way to use that + reference cards (my gray card maybe?) to get a more studio-accurate setting
 

·
Registered
Joined
·
5,325 Posts
Consistent times inaccurate equals inaccurate. Close only counts with horseshoes, hand grenades, and nukes. Honestly, you're going to spend a lot of time and effort validating a solution that can't be guaranteed correct and only tells part of the story, when you could have used less time and effort with HCFR or the free LightSpace DPS and an X-Rite I1 Display Pro to get the rest of the story...
 

·
Registered
Joined
·
713 Posts
You got it! Use the phone sensor on gray patches and got them all consistent. So it's calibrated to something, which I think we all agree is undetermined and maybe not linear, but something. And then confirmed with my camera that it is indeed calibrated to something consistently.
It's not calibrated to anything tho. All you've done is try to correct for how your camera sees your screen. A camera which likely has auto white balance so can't be trusted right off the bat. Most of these point and shoot cameras and cell phones have algorithms to take the raw sensor data and make a more pleasing picture out of the box, rather than give a dull, flat but more accurate picture. These algorithms and guesses will throw off any accuracy and consistency. It also likely uses the sRGB colour space and definitely doesn't use rec 709 to take photos, although I don't know how much that would affect gray scale patterns. Regardless, you aren't "measuring" in the same colour space.

As for the part about confirming your TV is "calibrated" with your DSLR, you most definitely did not. You said you white balanced off the TV. So any colour cast or inaccuracies are compensated for by your camera, which also isn't at any reference either. You are compensating for any bad picture by white balancing your DSLR off the bad picture itself. Just the fact that if you white balance off 30% or 50% or 80% (or whichever you did) and you get different results depending on which one you white balance (instead of consistent results every time), and then you choose the one that gives you the results closest to what you wish it to be, tells you that this method is flawed. Every single step is flawed, inaccurate, and later steps don't even confirm that previous steps are consistent unto themselves.

You keep saying that it's now calibrated to "something". You don't know what that something is, or how accurate it might be to anything, but you think it's calibrated to that. It's not, but let's just say it is. How do you know that "something" is more accurate than your TV was out of the box? That you somehow accidentally got it closer to reference than it was. You might have, you might not have. If after all this work you like the picture you ended up with, that's cool. You're the one who has to watch the TV, not me. But I hope no one goes out looking to learn more about how to calibrate their TV, comes across this method, then thinks they can end up with an accurate TV out of it :)


I think 10 point was probably too much because of linearity (will try to work around this), but 2 point balanced to a 30% and 80% reference gray would result in a "flatter" calibration correct?
Maybe. If your camera sensor is completely linear and you are able to take a photo with a completely flat picture profile, and even then there are other variables that this method shouldn't be trusted.



But with my equipment I can get everything aligned to a common reference point so colors are more accurate. Which I did, and looks subjectively better to my eyes. Thoughts?
Without any sort of reference, you have no idea if your colours are more accurate or if you are just blowing smoke up your own butt :D

As for my thoughts, remove all your custom white balance settings, use something like a Disney WOW disc and a blue filter to get your brightness, contrast, color and tint in the ballpark, pour yourself a stiff drink, then sit down and enjoy a good movie and call it a day! :p
 

·
Registered
Joined
·
45 Posts
Discussion Starter #17
It's not calibrated to anything tho. All you've done is try to correct for how your camera sees your screen. A camera which likely has auto white balance so can't be trusted right off the bat. Most of these point and shoot cameras and cell phones have algorithms to take the raw sensor data and make a more pleasing picture out of the box, rather than give a dull, flat but more accurate picture. These algorithms and guesses will throw off any accuracy and consistency. It also likely uses the sRGB colour space and definitely doesn't use rec 709 to take photos, although I don't know how much that would affect gray scale patterns. Regardless, you aren't "measuring" in the same colour space.

As for the part about confirming your TV is "calibrated" with your DSLR, you most definitely did not. You said you white balanced off the TV. So any colour cast or inaccuracies are compensated for by your camera, which also isn't at any reference either. You are compensating for any bad picture by white balancing your DSLR off the bad picture itself. Just the fact that if you white balance off 30% or 50% or 80% (or whichever you did) and you get different results depending on which one you white balance (instead of consistent results every time), and then you choose the one that gives you the results closest to what you wish it to be, tells you that this method is flawed. Every single step is flawed, inaccurate, and later steps don't even confirm that previous steps are consistent unto themselves.

You keep saying that it's now calibrated to "something". You don't know what that something is, or how accurate it might be to anything, but you think it's calibrated to that. It's not, but let's just say it is. How do you know that "something" is more accurate than your TV was out of the box? That you somehow accidentally got it closer to reference than it was. You might have, you might not have. If after all this work you like the picture you ended up with, that's cool. You're the one who has to watch the TV, not me. But I hope no one goes out looking to learn more about how to calibrate their TV, comes across this method, then thinks they can end up with an accurate TV out of it :)




Maybe. If your camera sensor is completely linear and you are able to take a photo with a completely flat picture profile, and even then there are other variables that this method shouldn't be trusted.





Without any sort of reference, you have no idea if your colours are more accurate or if you are just blowing smoke up your own butt :D

As for my thoughts, remove all your custom white balance settings, use something like a Disney WOW disc and a blue filter to get your brightness, contrast, color and tint in the ballpark, pour yourself a stiff drink, then sit down and enjoy a good movie and call it a day! :p
Cheers, thanks for your perspective. I'll keep at it using the various "calibrate your desktop monitor" guides out there as reference, I think I'm onto something with the data points I have. I spend quite a bit of time with photography so I've got some exposure (nice) to the flip side of the equation at least.
 

·
Registered
Joined
·
45 Posts
Discussion Starter #18
Final verdict for now: manually set DSLR white balance to 6500k (raw, natural color profile, no noise, etc etc) and redid my 10p after resetting what I set from the phone sensor. Have all 10 steps of gray with equal RGB values at 6500k.

Is this D65 calibrated? Probably nah. If anything, calibrated to my Canon camera sRGB profile. Which works for me, because I personally like Canon colors and the tv looks very right!

Change from my first pass... I think it was too cool initially, probably from not setting Color Grab white balance properly when I started. It's harder than I thought, and I'm fortunate my DSLR let's me select a white balance temperature.

All that said, using the phone fixed my peaky colors and muted colors, making it feel flatter and more natural. After using 6500k it's flat, natural, and as accurate as I've ever seen the TV.

Could the color balance be more accurate? YES! Did it fix my color issues with my panel (peaks and troughs with color temps)? Yes it did, and I subjectively think it looks as lifelike as I've ever seen it.

So for $0 and a couple hours time, I'm very happy. Stiff drink poured and enjoying the cleaner look.

Thanks everyone for sharing the art and science behind color calibration, I've learned a lot, even though my set isn't actually "calibrated."
 

·
Registered
Joined
·
713 Posts
Change from my first pass... I think it was too cool initially, probably from not setting Color Grab white balance properly when I started. It's harder than I thought, and I'm fortunate my DSLR let's me select a white balance temperature.

Could the color balance be more accurate? YES! Did it fix my color issues with my panel (peaks and troughs with color temps)? Yes it did, and I subjectively think it looks as lifelike as I've ever seen it.

So for $0 and a couple hours time, I'm very happy. Stiff drink poured and enjoying the cleaner look.

Thanks everyone for sharing the art and science behind color calibration, I've learned a lot, even though my set isn't actually "calibrated."

lol yes, this stuff IS a lot harder than one would think! And very time consuming. I get your thinking behind all this and where you were trying to go with it, but with so many unknowns there's no way you can zero in on any sort of reference. It's completely up in the air as to whether you got closer or further to a known reference. None of that matters tho as long as you like the results! Some people like Vivid and Sports modes on their TV as opposed to having a calibrated sets, and more power to them! Kudos for trying to use what you have and come up with an innovative method to get what you want, and I never look at time as wasted if you are learning something new!
 
  • Like
Reactions: agd and Rolls-Royce

·
Registered
Joined
·
1,751 Posts
Question for those more knowledgeable than myself: would the SPD of the display also affect the accuracy of any measurement taken by a camera?
 
1 - 20 of 33 Posts
Top