1 hour ago, efis007 said:

... which is flying elegantly in a beautiful clear day sky, and casting pitch black shadows inside.

Black shadows? 😳
On earth?
In broad daylight?
It's impossible!

Stop with the camera examples - we all have eyes and see natural light every day!  We all know what the real eye can see.  It is a simple problem to understand - the dynamic range of the real world is is FAR higher than what our monitors can display. There are two ways to deal with this.

1.  Maximise the dynamic range on the rendered screen and use a technique like 'eye adaption' to change the brightness to match what you are looking at..
2.  Use a software technique to compress the dynamic range, to make the whole scene a similar brightness

You prefer 2, but along with LR I much prefer 1.  The more that is done in software to achieve 2, the more option 1 suffers.

In the photo above, the pilot is clearly looking outside only.  In the same situation in the sim, to maximise dynamic range, the panel would be darker than reality.  From the angle of the pilots head, he will not be able to move his eyes only to glance at the panel, he would have to move his head.  In the sim, this head movement towards the panel will brighten it, Sure the outside may be a bit brighter than in reality, but is this really a problem, because we are currently looking at the panel?

Whatever way we view the simulator, we haven't got the dynamic range available to us to see reality, there has to be a compromise - flattening the image and reducing the contrast isn't it imho.

29 minutes ago, Biology said:

So I assume you see the world like -2.0 EV or the +2.0 EV options and nothing like the combined exposure option? This couldn't be further from my experience.

 

26 minutes ago, Murmur said:

What I see in RL is much closer to the left image than the right ones:

No, you don't see the world as in any of those images.  Your eyes will not see the left image so flat and lacking in contrast, you will see a very bright sky, but be able to see detail in the shadows.  I would hate to see that left image on my monitor, I would much prefer to move my head around the scene and have the exposure change to match what I'm looking at.  If I looking mostly towards the top, I would expect the bottom to be almost black.  If I look mostly in the shadows, I would expect the sky to be at least partly blown out. 

The sim never looks like those two shots on the right in any case, but maximising contrast and using an 'eye adaption' technique is surely a better way to go than making the overal image flat like the image on the left?

9 minutes ago, mSparks said:

the vast majority of users use head tracking of some form or other.

15 minutes ago, mSparks said:

I'm intrigued. How can you make the cockpit brighter without making it more visible?

That's the thing, it should be more visible. The current implementation is overdone. I'm not asking it to be completely flat like X-Plane 11, but the effect shouldn't be this drastic either. By mixing the results of both methods, you can have the same eye adaptation effect in a more subtle and life-like way.

33 minutes ago, Biology said:

I don't claim that a simulator can look exactly like reality, there indeed always will be a compromise. But this doesn't mean there can't be better compromises. There are certainly much better compromises than the current implementation, without sacrificing much contrast. As I said before, a good local tonemapper would not affect VR or head/eye tracking users while making the experience much more realistic for people who use neither. So yeah, the point is subtlety.

How can software discern a shadow in the cockpit, to one outside in the scenery?  I agree some level of tonemapping is going to be needed, but an increase in tone mapping will not only effect the cockpit - the overal scene will be effected. Far better to maximise dynamic range and reduce tonemapping for overall quality of the image.

25 minutes ago, Biology said:

Why would a solution which would have literally no impact on VR and eye/head tracking users but vastly improve the experience of people who don't use them be a worse solution? I don't think this is about "moving with the times", the truth is that a majority of users don't use either and they all come with their drawbacks.

VR is a head tracker so no issue.  Outside VR a head tracker does the same job for a monitor.  I have use a TrackIR for years and have never perceived a drawback - what possible issue is there when such a device helps developers maximise the dynamic range of a display? 

9 minutes ago, Biology said:

it should be more visible.

Based on what? 

bright daylight is 100,000 lux

a 600lux caution bulb when displaying bright daylight at 400lux would be 0,0,0 rgb - it's already more visible than it should be:

On 3/9/2023 at 7:55 PM, Biology said:

 

Edited March 11, 20233 yr by mSparks

16 minutes ago, MrBitstFlyer said:

How can software discern a shadow in the cockpit, to one outside in the scenery?

By using a stencil buffer. A stencil buffer can be used like a binary mask which tells whether each pixel belongs to the cockpit and the outside. And indeed this is what LR is considering using for their improved exposure fusion implementation.

16 minutes ago, MrBitstFlyer said:

I agree some level of tonemapping is going to be needed, but an increase in tone mapping will not only effect the cockpit - the overal scene will be effected. Far better to maximise dynamic range and reduce tonemapping for overall quality of the image.

I think you have a misconception about how tonemappers work - the scene is always tonemapped, so the tonemapping shouldn't be reduced or increased to begin with. What should change is the tonemapper curve applied to the scene, instead of using a more filmic curve like Narkowicz ACES (the tonemapper X-Plane 12 is using right now) which tries to replicate a film, a more perceptual tonemapper should be used.

Edited March 11, 20233 yr by Biology

29 minutes ago, Biology said:

This is exactly my point, you and Mark seem to think that this is a binary either-or situation, when it is not. In reality, a combination of both options can be applied together to replicate the real life experience much better than either options alone, all while not affecting people who use VR or eye/head tracking at all. This is what LR is trying to achieve and I'm sure when it's in the simulator you will understand what I mean.

It is not a binary option, but the more tonemapping is used, the more the overall contrast of the image will suffer.  My view is LR have taken the correct view to maintain more dynamic range, and rely less on tonemapping.

Just now, MrBitstFlyer said:

It is not a binary option, but the more tonemapping is used, the more the overall contrast of the image will suffer.  My view is LR have taken the correct view to maintain more dynamic range, and rely less on tonemapping.

See my post above - you appear to have a misconception about how tonemappers work. See this article from Bart Wronski, which goes into a lot of detail: https://bartwronski.com/2016/09/01/dynamic-range-and-evs/

11 minutes ago, MrBitstFlyer said:

No, you don't see the world as in any of those images.

That's because (as you also said) the dynamic range of a monitor is much smaller than RL. But the image on the left is the one I consider the closest to RL in terms of details visible.

From what Biology said, fortunately LR has also the same opinion and will make tone mapping more realistic!

12 minutes ago, mSparks said:

Based on what?

Based on my real-life experience, which the simulator should try to replicate. Again, just go to a real A320 cockpit and stand around the jumpseat, which is where I took that screenshot at. You will see that the cockpit looks much brighter than it currently does X-Plane 12, even when looking at the outside and the cockpit is in your peripheral vision.

By saying that it should be pitch black you are implicitly saying that a linear tonemapper should be used, which is never the case. The tonemapper used in X-Plane 12, Narkowicz ACES, is far from linear, and it certainly can be adjusted or combined with other tonemapping methods like exposure fusion to replicate human experience better.

Edited March 11, 20233 yr by Biology

Just now, Murmur said:

But the image on the left is the one I consider the closest to RL in terms of details visible.

...and to see those details in the sim, you get a flat image lacking in contrast as the photo demonstrates.

2 minutes ago, MrBitstFlyer said:

...and to see those details in the sim, you get a flat image lacking in contrast as the photo demonstrates.

...or you can use a subtle amount of exposure fusion to keep the local (perceived) contrast the same, while making the visibility much more life-like.

Edited March 11, 20233 yr by Biology

5 minutes ago, Biology said:

where I took that screenshot at. You will see that the cockpit looks much brighter than it currently does X-Plane 12,

And why is that the X-Plane lighting and not the A320 albedo I already pointed out needs reworking for the new lighting?

With all the others needing reworking again if they change the X-Plane lighting again.

Edited March 11, 20233 yr by mSparks

Just now, Biology said:

Again, just go to a real A320 cockpit and stand around the jumpseat, which is where I took that screenshot at. You will see that the cockpit looks much brighter than it currently does X-Plane 12

In a real cockpit you are using eyes to view a massive dynamic range. In the sim you are viewing the scene on a monitor, which hasn't got a huge dynamic range.  I completely agree with everyone that says tonemapping has to be used, but I think LR has taken the correct route by utilising an eye adaption technique, rather than totally relying on tonemapping to get around the problem of low dynamic range on our display devices.