AVS Forum banner

2661 - 2676 of 2676 Posts

·
Registered
Joined
·
584 Posts
Sure but I thought we were talking about how if we use DTM then it cannot be considered HDR.

Clearly not the case.
I agree. (And yes, this is a sidetrack but interesting!)
Even if you told DTM you had 10,000 nits it would still not be linear like you think it might. The specification doesn't dictate that, so putting it inside an SDR container won't do that either..
This I disagree. Isn't tonemapped "SDR container" signal linear though? I mean you are or should be using a straight gamma in projector so it has to be linear. And for that reason maybe there is a limit with display brightness where "tonemapped hdr in sdr container" cant be used anymore.
If you want to see what linear looks like, open an HDR film but put your display into SDR mode, it will be very washed out, the image is just not designed to look like that, what is also still clear is the 0-100nit range of the image is very much still visibly half the signal bandwidth.
Thats a different situation. That is coded as a HDR. I was talking about a HDR signal "tonemapped" to a SDR gamma for a 10 000 nits display. Imo it has to be linear. With 100 nits projector it would look super dim, but with a 10 000 nits display it would look right.
 

·
Registered
Joined
·
8,949 Posts
This I disagree. Isn't tonemapped "SDR container" signal linear though? I mean you are or should be using a straight gamma in projector so it has to be linear..
Yes and no, your display 'sees' linear gamma but, it's using an SDR gamma container because it's completely predictable and we know that almost all displays can replicate it properly and you can generally trust its somewhat right, but every displays eotf rendition is totally unpredictable, the DTM processor has absolutely no idea what it would look like on your display, so the only solution is to map it to a linear gamma, but as said it's not linear reqlly., Your display this is it's linear, but the processed image is not and it does not suffer from any of the issues you mention.

There is still a curve internally done by the tone mapper, it's an s curve. This means the highlights are suitably compressed and the low range expanded as expected, a good tone mapper will still have the correct bt2390 rolloff etc.

First thing we did with madvr was to get it to look absolutely identical to the Arve curves, so much so we actually fed madvr the Arve curve itself and it output that in SDR. It was completely identical to the pq Arve curves on my JVC. Once we did that we had a good foundation to work from and could finally ditch the HDR mode on the projector and start working on making it dynamic.

Like was said before, you REALLY need to stop thinking about it as SDR.
 

·
Registered
Joined
·
8,949 Posts
Thats a different situation. That is coded as a HDR. I was talking about a HDR signal "tonemapped" to a SDR gamma for a 10 000 nits display. Imo it has to be linear. With 100 nits projector it would look super dim, but with a 10 000 nits display it would look right.
Hmm, truly linear is what it looks like when you look at the raw HDR without processing, if you tell DTM you have 10,000 nits it will spit out the eotf version of that but in an SDR container so yes it would look quite dark, however the 0-100nits portion should not be 1% it should still be 50% of the encoded signal.

@madshi may need to jump in here and correct me if I am wrong.
 

·
Registered
Joined
·
584 Posts
Hmm, truly ljnear is what it looks like when you look at the raw HDR without processing, if you tell DTM you have 10,000 nits it will spit out the eotf version of that but in an SDR container so yes it would look quite dark, however the 0-100nits portion will not be 1% it should still be 50% of the encoded signal.

@madshi may need to jump in here and correct me if I am wrong.
If you shoot a 50% white linear gamma signal to the display, shouldn't it be half of the available brightness?

Raw HDR is not linear imo, because there are absolute brightness values linked to the pixel values. So it has to be always displayed with a curve or tonemapped with a curve.
 

·
Registered
Joined
·
8,949 Posts
If you shoot a 50% white linear gamma signal to the display, shouldn't it be half of the available brightness?

Raw HDR is not linear imo, because there are absolute brightness values linked to the pixel values. So it has to be always displayed with a curve or tonemapped with a curve.
Right so I picked a scene from Blade Runner, this shot is meant to be 90 nits. Pretty Dark :)

Its not 1% though, my scopes are showing 10% ?

3048995


Here I put my cursor over 1000 nits, and it seems to be somewhere around 30-40%... Shouldn't this be 10%?

3048997
 

·
Registered
Joined
·
584 Posts
Right so I picked a scene from Blade Runner, this shot is meant to be 90 nits. Pretty Dark :)

Its not 1% though, my scopes are showing 10% ?
Did you tonemap these to a linear gamma with some display brightness, or are these the original hdr?

I don't know if it's possible for you(or do you want to :) ) to tonemap some test scene to a different display brightness levels (for example 50, 100, 500, 1000 nits) and check the waveforms then?
 

·
Registered
Joined
·
8,949 Posts
Did you tonemap these to a linear gamma with some display brightness, or are these the original hdr?

I don't know if it's possible for you(or do you want to :) ) to tonemap some test scene to a different display brightness levels (for example 50, 100, 500, 1000 nits) and check the waveforms then?
These are tonemapped to 10,000 nits using gamma 2.4 or 2.2 can't remember. So it's the max possible bandwidth.

But I think it shows we are both somewhat wrong. If it was truly linear wouldn't the 100 nit shot be 1% in my scopes. Likewise the 1000nit point on the clipping test should be 10% but it's more like 30-40%...

In saying that I bet if I converted these to adl figures it may come out at 1 and 10% adl due to gamma conversion you have to do, luminance*0.42/255*100 for eg, but then that also shows that gamma 2.4.or some such is just a weaker version/similar of the pq eotf thus not really a problem anyway since the lower brightness content is still scaled in a non linear fashion if you study the eotf above.

Going to sleep now but happy to do more later.
 

·
Registered
Joined
·
584 Posts
These are tonemapped to 10,000 nits using gamma 2.4 or 2.2 can't remember. So it's the max possible bandwidth.

But I think it shows we are both somewhat wrong. If it was truly linear wouldn't the 100 nit shot be 1% in my scopes. Likewise the 1000nit point on the clipping test should be 10% but it's more like 30-40%...

In saying that I bet if I converted these to adl figures it may come out at 1 and 10% adl due to gamma conversion you have to do, luminance*0.42/255*100 for eg, but then that also shows that gamma 2.4.or some such is just a weaker version/similar of the pq eotf thus not really a problem anyway since the lower brightness content is still scaled in a non linear fashion if you study the eotf above.

Going to sleep now but happy to do more later.
Cool, thank you for these. Happy to be wrong if we can find the truth in this but lets continue later :) maybe should also jump to another thread from this New Projectors 2020 thread :D
 

·
Registered
Joined
·
5,778 Posts
Well that doesn't sound like its "...not very loud", at least not worth the extra $ for noise fan alone:(
At 80% the fan on my 760 is quieter than my 550 was on low lamp. Even at 100 we find it is not distracting. :)
 

·
Registered
Joined
·
4,429 Posts
No, I don't understand the claim that "SDR gamma is not capable of storing / representing truly high dynamic range scenes. For example, a highlight that's 1000, or even more times brighter than another element in the scene ".

The way I understand it (which is probably incomplete but I expect not completely off the mark): SDR gamma can easily represent a highlight that is 1000 times brighter than another element. Otherwise you couldn't have more than 1000:1 in contrast. There is in fact no limit to the dynamic range in SDR. The problem with it is that in HDR the normal dynamic range (excluding highlights) that includes most of the image information (the SDR part of the image) is squeezed down to 1-10% of the total range of the signal (1000-10,000 nits HDR). That means you only have 1-10% dynamic range left for encoding the SDR part of the image, which I imagine could lead to severe banding and other artifacts (squezing the SDR range down to 1-10% would imply a decrease in bit depth by 3-7 bits). Especially in darker scenes where you might only use 2% or so of the overall range of SDR. Divide that with another factor of 10-100 to save overhead range dedicated for highlights and you can see we are getting into trouble...
If I understand you correctly, a more factually complete statement would be "SDR gamma is not capable of storing / representing truly high dynamic range scenes. For example, a highlight that's 1000, or even more times brighter than another element in the scene... without introducing severe banding and other artifacts".

I believe you are correct. IIRC in the early HDR presentation they said if they had continued with gamma it would have required raising the bit depth to 15 or 18 (can't recall off the top of my head) With PQ they can get away with 10 bits.

From the layperson's perspective I certainly got the gist of what the poster said which is IMO also correct.
 

·
Registered
Joined
·
5,778 Posts
You're making it very hard for me not to consider the 915(790) :D
As always, if you can, try before you buy to ensure you like what you see and hear.
 

·
Registered
Joined
·
5,778 Posts
If I understand you correctly, a more factually complete statement would be "SDR gamma is not capable of storing / representing truly high dynamic range scenes. For example, a highlight that's 1000, or even more times brighter than another element in the scene... without introducing severe banding and other artifacts".

I believe you are correct. IIRC in the early HDR presentation they said if they had continued with gamma it would have required raising the bit depth to 15 or 18 (can't recall off the top of my head) With PQ they can get away with 10 bits.

From the layperson's perspective I certainly got the gist of what the poster said which is IMO also correct.
That was probably partly what was going on in the 550/675 and 260/285 projectors that had a 10 bit 60hz mode that was squeezed into a restricted pipeline and introduced severe banding on HDR content. Switching HDR off cured the banding but washed the picture out.
 
  • Like
Reactions: Reezepees

·
Registered
Joined
·
1,122 Posts
If I understand you correctly, a more factually complete statement would be "SDR gamma is not capable of storing / representing truly high dynamic range scenes. For example, a highlight that's 1000, or even more times brighter than another element in the scene... without introducing severe banding and other artifacts".

I believe you are correct. IIRC in the early HDR presentation they said if they had continued with gamma it would have required raising the bit depth to 15 or 18 (can't recall off the top of my head) With PQ they can get away with 10 bits.

From the layperson's perspective I certainly got the gist of what the poster said which is IMO also correct.
I would put it slightly different. A highlight that is a 1000 times brighter than another element is not necessarily a problem with SDR because that doesnt say anything about how bright the highlight is since the other element can be very dim and far far below 1 nit ( in fact it can be 0 nit). However, a highlight that is actually 1000 nits is another matter that will cause issues.
 

·
Registered
Joined
·
790 Posts
And it will be a gimick and look like over processed fake crap.
I guess this is off topic but no more so than ten year old projectors and TVs.

Just did some tests with iPhone 12 Pro and the HDR is legit. Recorded some side by sides with my iPhone 11 and it’s amazing the difference.

The one downside is I can’t seem to share them. Messages and iCloud photos shared albums don’t work, anyway, you lose the HDR and they look about the same. I’m sure that will come eventually.
 
2661 - 2676 of 2676 Posts
Top