13.12.2016 4 Автор Алексей

 Обновляемся до новой беты 6190 с оффсайта.

Много слов от разработчиков под спойлером

We have been closely following your feedback these last couple months and we’ve worked hard on improving things in general. So, I’d like to share with you some thoughts about the changes we made. More specifically in relation to visibility and volumetric fog, we spent many man/hours in trying to fine tune and provide a significantly improved experience. I wanted to give you an update on how things have changed in this area.

Volumetric fog introduction in P3D (since verison 2.x) has significantly increased the realism level in all regards in the sim. This however has also brought up several challenges for us in order to get things right in all possible situations. There are obviously some technical limitations due to the global nature of volumetric fog and there are many things that need to work correctly at the same time in this regard and this is what makes it somewhat tricky. Let me share the details:

Volumetric fog is a global graphics process that is rendered in a specific way around the current camera position/angle in the sim and there’s not much we can do to improve it per se. This “global nature” of volumetric fog does not happen in reality, where haze may be present for example only to the east, over e.g. a city with polluted air etc. In P3D if you set a volumetric fog vis layer with 10SM (16 km) and let’s say 2000 feet upper altitude, then if you slew up, you will actually be able to see an area relatively close to the aircraft previously not visible. And whatever is beyond this point of transition is depicted as if thick fog is present. This is something we have to rectify using various tricks, but at the same time:

  • We have to conform to the metar. If the reported visibility is e.g. 1/8SM, then there is a minimum height needed for the volumetric fog to prevent the ground from becoming visible
  • We need to provide smooth transitions, when moving either laterally, vertically, or when conditions change
  • We need to avoid (or at least minimize) hard lines at the horizon. These have always been present (since fsx too), but they are exacerbated by HDR (and some extreme HDR settings sometimes users tend to select)
  • When the METAR gives us the “freedom” (i.e. it’s either “9999”, or 10SM), we have to take into account temperature, atmosphere stability, winds, presence of temperature inversion layers, cloud ceiling etc in order to correctly define the visibility layer attributes. We already do that, but keeping the balance is tricky sometimes and the logic is complex leading to bugs sometimes. Fortunately, we’ve nailed most of the latter recently with AS16 SP1.
  • We need to keep performance impact to a minimum (if possible)
  • Users’ subjective opinions.

    Some users do love haze, some consider it too much. While we tend to base all of our depiction on physics as I said taking into account atmospheric conditions, we do need to apply some bias based on user preference. The challenge however in this part is that users may want to use volumetric fog, or they may hate it. Proper 100% realistic depiction in all cases is not possible with either of them, but truth is that with volumetric fog on, things are significantly improved. Nevertheless, the work that we need to do (fine tune) has to be done with a completely different set of parameters when volumetric fog is on, compared to when it’s off. For this reason when AS16 was initially released, we decided to drop support for “vol fog off”, but a reality check (user feedback) has shown that this is not possible. The problem is that this doubles our work. Since P3D works similarly to FSX (when volumetric fog is off), we had to reintroduce the “Disable custom haze layer” option present in FSX. This has confused users even more, since they didn’t really understand what this means and they thought that by checking this option, they will get rid of haze all the times (either with volumetric fog on or off). While in reality what this option does is it only disables the old “fsx-style” thin ugly cirrus like cloud layer following the aircraft and intersecting in an ugly way with the terrain. And of course it’s irrelevant when volumetric fog is checked.

    Many users tend to select a lower maximum upper visibility value, in order to avoid the low “cloud draw distance” related cloud redraw (“cloud popping”) when at cruise altitude. While this is understandable, the problem is that e.g. 60SM is a completely different value in P3D compared to FSX (as we knew it). Traditionally (in FSX) users, selecting 90 miles of cloud draw distance and an upper visibility of <90SM used to get the nice blend effect where the cloud redraw in the distance would not be evident. Now however in P3D, “60SM” is a completely different story. P3D interprets these values almost literally and applies them to a certain defined vertical range. This is interpolated with the volumetric fog and may lead to significantly more hazy conditions when e.g. at FL350. And even selecting 200SM is too restricted as depending on the aircraft altitude, 200SM is significantly lower than the horizon distance and certainly lower than the distance to the sky (as calculated by P3D). To give you an idea, the “real” unlimited visibility value at FL350 is more than 32000SM! (it increases exponentially). Not doing that lead to some other side effects:

  • Cartoon clouds. Clouds beyond the visibility limit, were not culled, but they were rendered in an ugly low resolution way (remember the “white mountains” in the distance when beyond the current visibility setting? It’s the same thing).
  • Cloud popping still came up
  • You got a harsh line with the sky at the point where the “foggy” upper visibility layer ended
  • This may have made the sky “banding” issue more evident and the sky in general to look uglier. This is also of course related to the sky color/texture selected. For example the well known HDE sky sets (designed for FSX and previous), in the pre-sunset variation, have somewhat darker lower bands (close to the horizon). This looks ok, when visibility is unrestricted, but when haze needs to be rendered in front of the aircraft, the fog is significantly brighter. The end result being that 2 dark “lines” were rendered and the final result was not acceptable. Similar results may come up with other sky color/texture sets (btw, we worked hard with ASCA sky textures to eliminate such issues, but many users may prefer alternative ones)

Now, the good news is that we were able to address most of these things, but this now makes the max upper visibility option not relevant. With SP1, we’ve removed the “Upper visibility” slider altogether (in P3D only and when volumetric fog is on only). This may disappoint some users that wanted to have this hazy environment, but truth is that most of the users just selected a lower upper visibility value in order to reduce cloud popping in the distance. This time however we decided to address cloud popping in the distance in another way: We now take advantage of volumetric fog and (at least for broken and overcast clouds) we automatically adjust the volumetric fog so that a nice blend of the overcast clouds with the horizon is achieved. Still, however when cloud coverage is low (or the clouds to be drawn are high, e.g. a thunder-cell), cloud popping may become apparent. So, this is not a perfect solution (and we’ll keep working on it), but overall the results are much improved.

So, what are the results? Performing many test flights with SP1 in P3D these last few weeks I really like very much the general feeling we achieved mainly due to the new visibility changes, the new cloud textures, the new blending with the horizon, volumetric fog handling (in combination with post process/shader apps like Reshade and PTA, which by the way I’ve been using recently and I took care so that all fine tuning took into account these commonly used utilities). Still however there is room for improvement. Especially since the general quality has increased significantly, this makes a couple of old standing depiction issues become more prominent:

  • Cirrus clouds: These are very nice when we look at them from the ground (and in general when we are below them), but when flying at cruise altitude they tend to ruin the immersion as they render as ugly lines (a reminder that cirrus clouds are just simple 2d textures always “facing” the ground). When above them they may intersect with lower cloud (z-order issue), also affecting depiction in a negative way. A thought about improving this is to detect such situations when injecting the weather (if the e.g. the aircraft is above FL200) and skip loading any cirrus clouds that are not at least 1000 feet above the current aircraft altitude. So, in SP1 we also included this “filter”. Note, that this will not make the issue go away completely but I think it will minimize it to a tolerable level
  • Cloud texture repetition: Since ASCA was released, this has been one of the main things that simmers used to criticize us . So, there are some things here that need clarification: This is a P3D bug (omission is a more accurate word). We’ve forward that information (what exactly is wrong) to LM and expect that it will be addressed soon. This problem (that btw does not exist in fsx or P3D version 1.x) is present whatever texture/cloud model someone uses. It’s more apparent with ASCA for a single reason: ASCA cloud models are larger in general by design (that’s the way we achieve these huge TCU and thunderstorm clouds thanks to the work Diana made). So, this single bug in the geometry calculations may be more noticeable. But it is present with default clouds or other 3rd party ones too. Do you remember the long threads about “pop corn” clouds? The core reason is the same. We can hide the problem up to a point (by using softer textures), but the problem will always be there until fixed by LM. In ASCA SP1, the problem is still there, but hopefully it’s now somewhat less noticeable.
  • Sky textures/colors: This is a 100% subjective thing. While I do like our current ones (they are very realistic at least for the current latitude I’m living at), after the initial negative feedback we got, we (Damian and Diana) deliberately made them more “neutral”. Some (like myself) may like it, but some may find them too dull and there will be cases of users that simply prefer something else. So, the major thing Damian actually changed in ASCA SP1 was to let other sky color definitions (freeware for example) to be incorporated in our dynamic selection processes. This is something that may bring to the community the best possible solution, with no compromises. Imagine a user located at northern Canada (or Norway) taking a nice high res picture of the sky there in several weather conditions. Then (after proper processing) he can assign it to the “high latitude – X” texture slot. Similarly a user located to a tropical zone, could make his own set. Then we can all share the best of all this and when we fly from e.g. Miami to Sweden the sky colors will gradually change (also depending on the weather conditions) to match real life as much as possible. This is one of the things that if properly taken advantage of (by the community), could really increase immersion significantly. Now, having said that, that does not mean that our own efforts in improving sky colors end here. We’ll still work on providing more and more of these as time goes by.

That’s it. We’re waiting for your feedback. Please keep a note that we read everything (both the positive and the negative feedback ) and we always try to improve things based on your valuable observations.

For B6171: One thing I wanted to add: You may notice a contradiction between what this post says and the “re-appearance” of the upper visibility slider. Based on your feedback we considered it important to give you back some control. One thing however that needs clarification is the way now the upper visibility slider works (when volumetric fog is enabled):
– It only works as “expected”, when the value is <= 100. In this case upper conditions are (and remain) hazy up to 40000 feet and users can control the level of actual “front” visibility
– If the value is 100-199, then it works similarly to the way it worked in SP1 beta1. It’s “hazy” up to about 15000 feet and it quickly clears up when the aircraft altitude reaches 20000feet. This depending on the sky texture used may give either very realistic or problematic results.
– If the value is 200 (which is now the default), it clears up very quickly, when getting above the surface visibility ceiling. In most cases it will be completely clear by 8000 feet of altitude when conditions are very foggy on the ground (and in most other cases it will clear up at much lower altitudes).

So please provide us your feedback on this area. I just wanted to explain in detail how exactly it works now, so that you can get more or less predictable results during your tests.

For B6190: Please note that we’ve tweaked things further with maximum upper visibility based on feedback. When set to 200 miles, only a single visibility layer will be used which can improve performance. Default is now 199 miles and is the maximum upper visibility depiction possible with 2 visibility layers, providing smoother visual visibility depiction (elimination of hard lines).

Thank you!!!

1 085 Просмотров