Yes, your display will be capable to go up to 10'000 nits with this.^^^ Blank DPL? Is that a cheat code?
For my interest - what is your DPL?
150 nits.
Yes.
Yes, any preset can lead to a better or worse result depending on the scene. Generally, the strongest presets will widen this gap even more.
I'm constantly updating the expiration date in new builds. The latest builds expire at the end of March, with a small warning message showing up at the end of February.
Good post. But let's talk about an example here:So one thing I've thought about, and now having "don't add peak nits" makes it very easy to test apples-to-apples, is the conventional wisdom about DPL, and no compression limit, and not lying to madVR about DPL. And I know this has already been discussed in the past, but please be open-minded, and hear me out.
So I understand the reasoning behind not lying to madVR about DPL, and how the no compression limit is intended to address concerns about "protecting" a critical range while allowing frames with peak < DPL to map 1:1. But I think there are some important considerations:
1. madVR essentially lies to itself all the time, in the form of DTN and any resulting target > DPL, and also when NCL is used.
2. When using SDR output, due to the relative nature, it actually does not make any difference what DPL is actually set to, the target is what dictates what the frame will look like.
And to prove this, I did some testing.
Profile 1: DPL 65, DTN 60, NCL 103 - Resulting FALL multiplier of 2.175
Profile 2: DPL 103, DTN 54, NCL 103 (I don't believe this NCL matters in this case) - Resulting FALL multiplier of 2.174
I intentionally used the formulas I posted recently to find a DPL/DTN combo that resulted in the same multiplier and thus the same targets, and this is what I meant above when I said "don't add peak nits" helps test this apples-to-apples. I've theorized about this previously, but finally took the time and was able to actually test it.
Here are some examples of what happens with these two profiles with different peaks/FALLs...
Peak 400, FALL 75 - Target of 163 nits in both cases, end result is identical
Peak 300, FALL 20 - Target of 103 nits in both cases, end result is identical
Peak 120, FALL 10 - Target of 103 nits in both cases, end result in identical
Peak 80, FALL 7 - Target of 80 nits for Profile 1, Target of 103 nits for Profile 2
Peak 50, FALL 5 - Target of 50 nits for Profile 1, Target of 103 nits for Profile 2
So the only difference between these two different approaches is in the case where peak is less than the 103 NCL/altDPL. In this scenario, Profile 1 maps 1:1 (or to peak if in between DPL and NCL), while Profile 2 maps to 103 nits. And in a vacuum, mapping 1:1 when possible seems like a reasonable goal. But let's consider how this is really working in practice.
So I ran through a scene in Valerian which has a mix of frames with peaks <65, between 65 and 103, and >103. And within this same scene, Profile 1 is jumping around between 1:1 mapping for the <65 frames, mapping to 103 for the >103 frames, and in between on the frames between 65 and 103. Profile 2 is consistently mapping to 103 for the whole scene. FALL is never high enough to kick in a higher target via DTN.
Keep in mind this is all within the same scene. The only major difference between the <65 frames and the higher frames is whether something semi-bright is in the frame or not. And so in the case of Profile 1, the scene is mapped inconsistently, while with Profile 2, it is mapped consistently. So I think we need to take a step back and ask ourselves, do we only want to "protect" the critical range (however one wants to define that, 75 nits, 100 nits, 106 nits, 200 nits?) which NCL does, or do we also want to ensure consistent TMing within that critical range as well (which NCL does not do)? Do we really want frames within the same scene to be mapped inconsistently through this critical range? This is importantly different from inconsistent mapping above the critical range, which is a necessary evil of TMing and not too detrimental with good target smoothing. As you ask yourself these questions and answer them, keep in mind that this is the only difference in the end result of these two profiles due to the relative nature of SDR and that it really does not matter whatsoever what DPL is set to otherwise.
This is easy to test now with "don't add peak nits". You just have to create a second profile with DPL equal to your current NCL and then use the formulas I posted to find a DTN that results in the same multiplier as your current DPL/DTN combo. I'm happy to help anyone with this if they need it. Cheers!
Edit to add: This really only applies to those with DPL which is less than the "critical range", i.e. those who are (or maybe should have considered) using NCL to protect this critical range. But that does seem to be most PJ users here. See, we high-nit users do care about low-nit users!
Right, and that's the reasoning one might use for not using NCL at all, if they decide that is most important to them, accepting the drawbacks of doing so. But NCL also exists for some compelling reasons, and using it has its own benefits and drawbacks. And so I feel that especially for someone who decides to use NCL for its benefits, we should also consider this third approach, which retains those benefits and also keeps consistency through that critical range, and of course has it own drawbacks.
So one thing I've thought about, and now having "don't add peak nits" makes it very easy to test apples-to-apples, is the conventional wisdom about DPL, and no compression limit, and not lying to madVR about DPL. And I know this has already been discussed in the past, but please be open-minded, and hear me out.
So I understand the reasoning behind not lying to madVR about DPL, and how the no compression limit is intended to address concerns about "protecting" a critical range while allowing frames with peak < DPL to map 1:1. But I think there are some important considerations:
1. madVR essentially lies to itself all the time, in the form of DTN and any resulting target > DPL, and also when NCL is used.
2. When using SDR output, due to the relative nature, it actually does not make any difference what DPL is actually set to, the target is what dictates what the frame will look like.
And to prove this, I did some testing.
Profile 1: DPL 65, DTN 60, NCL 103 - Resulting FALL multiplier of 2.175
Profile 2: DPL 103, DTN 54, NCL 103 (I don't believe this NCL matters in this case) - Resulting FALL multiplier of 2.174
I intentionally used the formulas I posted recently to find a DPL/DTN combo that resulted in the same multiplier and thus the same targets, and this is what I meant above when I said "don't add peak nits" helps test this apples-to-apples. I've theorized about this previously, but finally took the time and was able to actually test it.
Here are some examples of what happens with these two profiles with different peaks/FALLs...
Peak 400, FALL 75 - Target of 163 nits in both cases, end result is identical
Peak 300, FALL 20 - Target of 103 nits in both cases, end result is identical
Peak 120, FALL 10 - Target of 103 nits in both cases, end result in identical
Peak 80, FALL 7 - Target of 80 nits for Profile 1, Target of 103 nits for Profile 2
Peak 50, FALL 5 - Target of 50 nits for Profile 1, Target of 103 nits for Profile 2
So the only difference between these two different approaches is in the case where peak is less than the 103 NCL/altDPL. In this scenario, Profile 1 maps 1:1 (or to peak if in between DPL and NCL), while Profile 2 maps to 103 nits. And in a vacuum, mapping 1:1 when possible seems like a reasonable goal. But let's consider how this is really working in practice.
So I ran through a scene in Valerian which has a mix of frames with peaks <65, between 65 and 103, and >103. And within this same scene, Profile 1 is jumping around between 1:1 mapping for the <65 frames, mapping to 103 for the >103 frames, and in between on the frames between 65 and 103. Profile 2 is consistently mapping to 103 for the whole scene. FALL is never high enough to kick in a higher target via DTN.
Keep in mind this is all within the same scene. The only major difference between the <65 frames and the higher frames is whether something semi-bright is in the frame or not. And so in the case of Profile 1, the scene is mapped inconsistently, while with Profile 2, it is mapped consistently. So I think we need to take a step back and ask ourselves, do we only want to "protect" the critical range (however one wants to define that, 75 nits, 100 nits, 106 nits, 200 nits?) which NCL does, or do we also want to ensure consistent TMing within that critical range as well (which NCL does not do)? Do we really want frames within the same scene to be mapped inconsistently through this critical range? This is importantly different from inconsistent mapping above the critical range, which is a necessary evil of TMing and not too detrimental with good target smoothing. As you ask yourself these questions and answer them, keep in mind that this is the only difference in the end result of these two profiles due to the relative nature of SDR and that it really does not matter whatsoever what DPL is set to otherwise.
This is easy to test now with "don't add peak nits". You just have to create a second profile with DPL equal to your current NCL and then use the formulas I posted to find a DTN that results in the same multiplier as your current DPL/DTN combo. I'm happy to help anyone with this if they need it. Cheers!
Edit to add: This really only applies to those with DPL which is less than the "critical range", i.e. those who are (or maybe should have considered) using NCL to protect this critical range. But that does seem to be most PJ users here. See, we high-nit users do care about low-nit users!
ya in practice it didnt work, maybe for movies like Starship Troopers, but for everything else it was much too dark. to the point where im trying a number below whats measured at the screen off the lens now xD the idea was simply sdr is graded at 100-120nits gamma 2.4, and reference/diffuse white is sometimes defined for HDR grading at 100 nits, or now more recently at 200 nits, if the standard for sdr on projection is lower match that. throwing all that out and setting dpl at a brightness you're seeing works better...
ill try that too thanks i hadnt tried using ncl with the lower dpl yet i dont think.
It certainly is interesting!
And this setting DPL < real nits (lying to madVR in the other direction) is a 4th approach. So let's think about the differences there...So I think first we have to always consider a given DPL/DTN combo together, since that's what determines targets. And so if you use a different combo that results in different targets, that's really apples-to-oranges, and is the difference you're seeing due to the different DPL, or different DTN, or the combo? Do you use "don't add peak nits"?
Okay, so ignoring DTN for the moment due to what I mention above, with 75 real nits, and just comparing (DPL 75) vs (DPL 50 + NCL 75), this is very similar to approaches 2 and 3 I listed, except the 1:1 mapping instances in that DPL 50 scenario actually become 1:1.5 mapping, or brighter than even originally coded! Sounds like what you describe as raising your near blacks, except it only happens sometimes, even in the second scenario. I won't judge you or try to tell you what you should do, but that sounds kind of wonky to me!
I feel strongly that what you subjectively prefer should certainly trump everything else.
Just as a side question that may be somewhat related, do you use gamma processing to change 2.2 gamma to 2.4? That would potentially make the side effects of this approach, the "expansion" of near blacks, potentially less offensive. Just a random thought, I just thought I may have remembered that you used gamma processing.
Right, so with my preferred setting I am getting 1:1 up to my DPL. Above 75 I am mapping to brighter than coded, but then I am using a higher DTN which brings the brightness down somewhat (sat least compared to 50 DTN which is what I prefer when trying to use 75 DPL).It certainly is interesting!
So I think first we have to always consider a given DPL/DTN combo together, since that's what determines targets. And so if you use a different combo that results in different targets, that's really apples-to-oranges, and is the difference you're seeing due to the different DPL, or different DTN, or the combo? Do you use "don't add peak nits"?
Okay, so ignoring DTN for the moment due to what I mention above, with 75 real nits, and comparing (DPL 75) vs (DPL 50 + NCL 75), this is very similar to approaches 2 and 3 I listed, except the 1:1 mapping instances in that DPL 50 scenario actually become 1:1.5 mapping, or brighter than even originally coded! Sounds like what you describe as raising your near blacks, except it only happens sometimes. I won't judge you or try to tell you what you should do, but that sounds kind of wonky!
I feel strongly that what you subjectively prefer should certainly trump everything else.
Just to be clear, what I posted about today is not the same thing as what I posted to you about a few days ago or whenever that was. Or at least, it wasn't intended to be, I hadn't even tested this stuff yet. I was under the impression then that your real nits was ~50 and that you were setting DPL to 25. I even mentioned an assumed real nits of 50 in my reply to you:
What I had suggested you try then (assuming your real nits was 50) was comparing the 25/75 combo you were running with specifically a 49/62 combo while leaving everything else the same. This should have resulted in the same exact targets whenever DTN kicked in. And so the only difference you should have seen should have been on frames with peaks <49 nits. But I was thinking at the time that the change you would see is instead of an "expansion" mapping of 1:2 you'd be getting 1:1 mapping, while keeping the targets you loved. Now I'm not so sure you actually have 50 real nits.
I actually think you are getting 1:1.5 expansion mapping <= 50 (where madVR thinks it is mapping 1:1 but then your PJ outputs it 50% brighter?), then that shifts more and more toward 1:1 as we go up in target toward your real nits of 75, at which point madVR thinks it is compressing 1.5:1 but then your PJ outputs it brighter and results in 1:1? Or something like that, my brain is tired!
no i dont think so either;] i re-measured 70nits at the screen(meter facing the projector), and my screens gain i measured at ~0.65 compared to a st100 sample. which would be about 45nits. if you factor in half gain i think 40 nits is more appropriate.
OLED - HDR "pass-through" added to above post
So your OLED image is an actual photograph, but the other two are just screencaps in Windows?
Thankyou..
