Sign in to follow this  
Rob Ainscough

ASN B5386 now available for P3D V2.4

Recommended Posts

Help AVSIM continue to serve you!
Please donate today!

FYI to all: 

 

At W52 (Orbx) airport using latest ASN live weather (5 layers) with P3D Cloud Density set to Max = 12 fps -- Cloud Density set to normal = 30 fps

 

Cheers, Rob.

 

EDIT: it does look amazingly realistic, but Cloud Density = Max with ASN is not something my system can handle. 

Share this post


Link to post

Hi Rob,

 

You do realize that by suggesting users to avoid using cloud density setting to max, will lead prepar3d to never select overcast cloud models. I understand the issue you get with the cloud overdraw induced performance drop, but this has to be seen in context with all the other settings selected in p3d.

 

And let me add my humble opinion about all this p3d discussion: It seems that most users want smoothness and the best graphics quality possible and this is acceptable. But by focusing on this, we do miss the main orientation and aim of p3d: It's a training tool above all. And by that trainees may need e.g. to practice non precision approaches in conditions with a metar "9999 OVC012" and the MDA at 1100 feet agl. In these conditions, the volumetric fog is irrelevant and P3D has to render a proper low overcast ceiling for the "trainee" to be able to decide about landing or not once breaking out of the ceiling. So, IMHO, it's much more important for users/trainees to set these options appropriately (cloud density to max), even if this means that they may have to sacrifice other areas of image quality and/or smoothness (e.g. disable cloud shadows).

 

That's  also the reason I insist that the "math" about the how much the clouds extend vertically should be correct, even if they are not so "dramatic" or "good looking". Beau fixed this with the shader fix (and this was incorporated in 2.4). I still got the feeling though the cloud base is not 100% respected. E.g if you just install vanilla P3D 2.4 and load it at the default airport/aircraft and just select the major thunderstorms weather theme (with no weather engine running), the clouds observed once airborne over the sea will still intersect the sea. This issue is present in fsx too (under certain circumstances), but was exaggerated in 2.3, is vastly improved in 2.4, but I think there's still room for improvement. My opinion is that we talk too much about microstutters, visual quality etc and we seem to not care so much for accuracy. I think this is the wrong approach here. And we make the (excellent) developers of P3D lose focus of what's really important in P3D development.

 

Just my opinion...

Share this post


Link to post

You do realize that by suggesting users to avoid using cloud density setting to max...

 

Yes I do ... but I don't believe I'm "suggesting" anything ... think you are reading a little too much into what I wrote.  I'm letting people know how the Cloud Density settings had a significant performance impact on my system in combination with my ASN settings.

 

Without ASN and using the built in P3D weather themes or my own (setup to overcast) performance of my system drops from 30 fps to 27-28 fps which I find manageable and a good compromise.  Performance starts to be a problem when there is input lag between ones input device (yoke, throttles, etc.) so there does need to be some balance between accuracy and performance.  

 

But either way, P3D can be setup to hit 60 fps with "9999 OVC012" which should meet training needs ... the RedBird motion simulator I use is FAA approved including weather conditions and does indeed operate at 60 Hz/fps (but this was done on an older ESP as I understand it ... but don't quote me).

 

As far as accuracy, it's a simulator, not real world - current technology only lets us go so far.  For example, an equally and just as important training would be how to identify icing on the wings ... in some aircraft the ONLY way to "discover" icing is to visually look at the wing and perform appropriate action.  XP10, FSX, P3D don't visually emulate ice on the wing/surfaces.  In fact, one can get their real PPL without ever having to deal with icing ... so obviously there is room to "grow" both in the real world and simulated world.

 

So in summary, I don't think LM have lost focus at all, they are simply living within the bounds of ESP and trying to extend those boundaries ... I think we're fortunate LM have provided this much and have plans to provide more.  My gut feeling (and that's all it is, I have no inside information on this) is the more complex issues are being saved for a version 3.x (be it 64bit or whatever LM decide).  So yes, more can be done and I hope will be done ... that's why I support the effort.

 

Cheers, Rob.

 

EDIT: my primary purpose of this post was to let folks know ASN is now compatible with P3D V2.4 because that is a question I get asked a lot (in these forums and elsewhere) ... so my post was intended as a single source for those many that were wondering, helps me avoid posting many times and in many different threads and even web sites and even my YouTube channel.

Share this post


Link to post

Hi Rob,

 

You do realize that by suggesting users to avoid using cloud density setting to max, will lead prepar3d to never select overcast cloud models. I understand the issue you get with the cloud overdraw induced performance drop, but this has to be seen in context with all the other settings selected in p3d.

 

And let me add my humble opinion about all this p3d discussion: It seems that most users want smoothness and the best graphics quality possible and this is acceptable. But by focusing on this, we do miss the main orientation and aim of p3d: It's a training tool above all. And by that trainees may need e.g. to practice non precision approaches in conditions with a metar "9999 OVC012" and the MDA at 1100 feet agl. In these conditions, the volumetric fog is irrelevant and P3D has to render a proper low overcast ceiling for the "trainee" to be able to decide about landing or not once breaking out of the ceiling. So, IMHO, it's much more important for users/trainees to set these options appropriately (cloud density to max), even if this means that they may have to sacrifice other areas of image quality and/or smoothness (e.g. disable cloud shadows).

 

That's  also the reason I insist that the "math" about the how much the clouds extend vertically should be correct, even if they are not so "dramatic" or "good looking". Beau fixed this with the shader fix (and this was incorporated in 2.4). I still got the feeling though the cloud base is not 100% respected. E.g if you just install vanilla P3D 2.4 and load it at the default airport/aircraft and just select the major thunderstorms weather theme (with no weather engine running), the clouds observed once airborne over the sea will still intersect the sea. This issue is present in fsx too (under certain circumstances), but was exaggerated in 2.3, is vastly improved in 2.4, but I think there's still room for improvement. My opinion is that we talk too much about microstutters, visual quality etc and we seem to not care so much for accuracy. I think this is the wrong approach here. And we make the (excellent) developers of P3D lose focus of what's really important in P3D development.

 

Just my opinion

well said, i share your sentiments, I start with all sliders at minimum but wx at maximum because its important for me, I then crank up the scenery and eye candy at my fav airport (EGLL) until the FPS is affected, thats where I leave it, wx trumps everything else for me

 

OH and PS, Rob, thanks for your superb input to the platform, I really am pleased your on the team as you represent the sim world views well

Share this post


Link to post

 

 


OH and PS, Rob, thanks for your superb input to the platform

 

Most welcome ... Flight Simulation is a lot of different things to a lot of different people, some focus on aircraft systems, others weather, some obstacles/terrain, some procedures, and some everything.

 

But clouds are eye candy too and I'm a firm believer in eye candy whether it's clouds, or mesmerized by a weather radar, or staring at an MFD/PFD course line, or looking out the window at realistic terrain with traffic, or searching for other aircraft in my air space, or sitting at an airport watching traffic come and go, or observing clouds move across the land/water, etc. etc. 

 

I'm sure some of us would like it all and the kitchen sink (I know I would) but we live with compromises and make the best out of it.

 

But ultimately what I think it really all boils down to is my obsession with pushing buttons and turning knobs  :Peace:

 

Cheers, Rob.

Share this post


Link to post

Hi Rob,

 

 

Yes I do ... but I don't believe I'm "suggesting" anything ... think you are reading a little too much into what I wrote

 

 

well, you know when a highly respected individual as you makes comments like that, they may be misread as indirect suggestions ;) Not only by me, but by other not so knowledgeable users too, that's why I wanted to clarify what exactly this option does.

 

 

Without ASN and using the built in P3D weather themes or my own (setup to overcast) performance of my system drops from 30 fps to 27-28 fps which I find manageable and a good compromise.  Performance starts to be a problem when there is input lag between ones input device (yoke, throttles, etc.) so there does need to be some balance between accuracy and performance.  
 
But either way, P3D can be setup to hit 60 fps with "9999 OVC012" which should meet training needs ... the RedBird motion simulator I use is FAA approved including weather conditions and does indeed operate at 60 Hz/fps (but this was done on an older ESP as I understand it ... but don't quote me).

 

 

You seem to imply that something specific to ASN brings performance down. We've been working with LM developers closely recently and they know exactly how ASN works and what we try to achieve. And while I understand that ASN may be the best indirect "beta tester" for P3D (stress testing it and indirectly leading to all these improvements we've seen recently in cloud related performance), I'd like to investigate this further to see if it's something wrong on our side. So please provide further details about this fps drop incident (best would be for this to be done on our support forums to avoid cluttering this thread/forum). 

 

Thanks!

Share this post


Link to post

Here ya go Kostas,

 

Video is self contained but suggest you watch in HD.  Frame rates in upper left corner.  ASN and P3D graphics settings are all in the video.  Cloud textures are P3D defaults.  P3D Date/Time is 9/30/2014 17:49.  ASN using historical (see video).

 

 

Download the flight (fxml) and weather (wx) files from here:

http://robainscough.com/uploads/ASNOvercastFlightWeather.zip

 

ASN is in network configuration and the as_btstrp.dll is listed as v 1.0.0.12  (9/30/14)

 

Cheers, Rob.

Share this post


Link to post

Hi Rob,

 

thanks for this. It helps being specific. I can see in this video that you have the "Repeating overcast texture pattern fix" in ASN checked, while we recommend against it in P3D for performance reasons.

 

In the video at the specified moment the station influencing the weather at your position was this one:

 

KVUO 011553Z AUTO 14003KT 10SM BKN006 OVC029 12/10 A3025 RMK AO2 SLP243 T01170100

 

 

In this case ASN interprets these 2 clouds as this:

 

1) 7/8 (broken) stratus cloud with cloud base at 600 feet agl with the top at 5600 feet

2) 8/8 (overcast) stratus cloud with cloud base at 2900 feet above ground level with the top at 8913 feet. 

 

I actually performed some tests on my machine (with many P3D settings much lower than yours, but ASN settings exactly as the ones you have. I only made sure I kept the cloud density to max in all these tests)

 

- First test at W52 (default) with clear skies. FPS = 30 fps (locked)

- Second test with ASN historical weather at 2014/10/1 1557Z creating the above conditions: 19-24 FPS depending on the view angle 

- Third test, I shut down ASN and manually set the weather to the above clouds. The only thing I needed to change was the cloud top of the broken cloud to 2800 feet (since P3D manual weather dialog does not allow cloud overlap): 18-25 FPS

- 4th test: I unchecked the "Repeating overcast texture pattern fix" in ASN as it is not recommended to be on in P3D. Result (using ASN in the same conditions): FPS=30

 

So comparing apples to apples (stratus overcast clouds with stratus overcast clouds, not thunderstorm ones), it's clear to me that the "repeating pattern fix" is responsible for this performance drop and will become apparent only when stratus overcast clouds are depicted. 

 

Technically with this "fix" ASN attempts to deal with the well known light related line patterns. This works very well in FSX, but we were aware of possible performance issues in P3D since version 2.3 

 

It is the same issue Zach is talking about here:

http://www.prepar3d.com/forum-5/?mingleforumaction=viewtopic&t=8190.1

 

Stratus overcast clouds are specifically prone to these ugly line/stripe patterns (especially during dusk/dawn). It seems to me that the shader fix made by LM devs makes our own fix obsolete (thus this option should be disabled/hidden in ASN for P3D). Can you confirm my findings Rob? Just uncheck this option and re-test in the same conditions.

Share this post


Link to post

Can you confirm my findings Rob? Just uncheck this option and re-test in the same conditions.

 

Hi Kostas,

 

Confirmed!  FPS has returned ... only a 1-2 fps difference now between Cloud Density Medium and Max.  

 

 

"Repeating overcast texture pattern fix" in ASN checked

 

Just a thought, might want to remove this option for P3D users given it's impact on performance (60% drop in FPS is pretty drastic).  Also, I don't think you need the cloud base offset any more for P3D users?

 

But I appreciate you taking the time and your suggestion is very informative and I hope others will benefit from this info as much as I did.

 

Cheers, Rob.

Share this post


Link to post

@Kostas (HiFi),

 

I was just about to respond to Rob, because that option is checked off by default for the ASN P3d version (ASN also has a note for the checkbox, warning not to use it in P3d unless you want performance to decline). I have a weaker system than Rob's and I don't see the performance hit with the recent ASN beta that he does/did. 

Share this post


Link to post

Here ya go Kostas,

 

Video is self contained but suggest you watch in HD.  Frame rates in upper left corner.  ASN and P3D graphics settings are all in the video.  Cloud textures are P3D defaults.  P3D Date/Time is 9/30/2014 17:49.  ASN using historical (see video).

 

Hi Rob - after setting my ASN to your video settings and P3D  meus settings 80 miles on clouds distance and maximum amount of clouds I no longer have vertical clouds - tried ASN and P3D weather choices and I can't reproduce these vertical clouds - excellent - thinking it was your ASN settings - probably repeating overcast texture pattern fix in ASN settings ticked on - thanks brother

 

Hmm just noticed another update downloading now - and will load it up

Share this post


Link to post

Well seeing those vertical clouds again so not fixed - has to be 2.4 build I guess and a slight stutter while panning other than this all looks good - hopefully this will get sorted out

Share this post


Link to post

 

 


Just a thought, might want to remove this option for P3D users given it's impact on performance (60% drop in FPS is pretty drastic).  Also, I don't think you need the cloud base offset any more for P3D users?

 

Yes, of course. As ASN for P3D official release is closing (remember we are still in open beta), we'll make sure all these things are properly adjusted.

Share this post


Link to post

 

 


Well seeing those vertical clouds again so not fixed - has to be 2.4 build

 

Yeah, not sure what to say about the Vertical clouds (they were present in v2.3 also - see my video on that - but seems to be a little more frequent in v2.4) ... based on my understanding of how clouds are displayed, I don't think they are "vertical" clouds in intent of the word ... "I think" they are clouds that simply aren't being correctly orientated to the current view port and/or is a by-product of other optimizations to clouds to encourage a more volumetric "feel".  

 

But problem was definitely in 2.3 as well as 2.4.  I can't recall if I saw it in 2.0 - 2.2?

 

Cheers, Rob.

Share this post


Link to post

Yeah, not sure what to say about the Vertical clouds

 

Thanks for the explanation Rob - must have got lucky in 2.3 as I really didnt see one of these and now all the time - is the 2.3 cloud file a workaround for now maybe ?

Ok I replaced 2.4 with 2.3 cloud file and minor stutter is gone and not seeing vertical cloud issue - one flight but seems better tried various weather themes so seems at first glance that the 2.4 cloud file does need to be looked at I think

Share this post


Link to post

Hot on the heels of yesterday's P3D 2.4 fix comes today's RC1 release, for both FSX and P3D. :smile:

 

http://www.hifitechinc.com/forums/showthread.php?3502-ASN-SP1-OPEN-BETA-RC1-now-available!-FSX-and-P3D

RC2 (Build 5391) has now been posted. This build will probably be the same as ASN SP1 release version for P3d according to HiFi. This brings to a close a successful open beta period, which I wish that more 3rd party authors would offer.

But problem was definitely in 2.3 as well as 2.4.  I can't recall if I saw it in 2.0 - 2.2?

 

 

There's enough posts here and there on the internet that indicate to me that this bug was even present in FSX. My guess is that a combination of some code changes to cloud.fx in P3d 2.4 plus more realistic low-level stratus cloud textures (such as found in REX 4 Texture Direct) are making this issue more apparent. Flatter clouds would look far worse when incorrectly rotated 90 degrees than puffy clouds would.

 

I suppose that there is a possibility that this is a feature and not a bug that has been with us since FSX. Since the clouds in ESP are not volumetric, but rather a layering of adjacent sprites, it's at least remotely possible that ACES introduced a very low rate of occurrence of 90 degree flipped clouds, in order to give the bottoms of the lowest cloud layer a more irregular appearance. 

 

Have you played around with the advanced cloud settings in P3d 2.4 (without an external weather engine active) in order to see if this bug is more prevalent in the lowest cloud layer or is it common even if you manually create 5 or 7 cloud layers?

Share this post


Link to post

I have done my own weather and not noticed any vertical clouds ... also done some of the P3D weather themes and have not noticed any vertical clouds ... key being I haven't noticed (meaning they could be lurking and I just didn't see them).  

 

So far I've only seen them with weather engines.  However, it could very well be I just didn't hit the right combinations or settings of my custom weather such that I would see a vertical cloud.  Weather engines are probably more likely to hit the situation that triggers the vertical clouds since their coverage is much more detailed and accurate than any theme or something I could create with built in P3D weather.

 

Right now I'm running 5 cloud layers in ASN and cloud density set to Max in P3D.

 

Cheers, Rob.

Share this post


Link to post

I have done my own weather and not noticed any vertical clouds ... also done some of the P3D weather themes and have not noticed any vertical clouds ... key being I haven't noticed (meaning they could be lurking and I just didn't see them).

 

Like some others I did not see them prior to 2.4 and this is only using my own weather [as you say they may have been lurking] but I do see them occasionally in 2.4. So it can and does happen with custom made themes. Maybe there is a greater chance of it occurring with certain combinations of density and cloud height, that weather engines are more likely to use.

 

Despite not seeing them in 2.0 - 2.3 I have seen them in FSX and therefore wonder if its because LM have pulled the distance at which clouds rotate to the viewpoint closer to the viewer... it seem we now have a system somewhere between FSX and what LM first implemented.

Share this post


Link to post

As a commercial pilot let me add something to the mix here.  When the accuracy of clouds and what breaking out looks like for those pilots that want to use a sim for a trainer and to log time, let me  give some perspective.  When I first started flying IFR it was in a airplane with a instrument instructor, he would make me wear a hat that had an overextended brim that would keep my focus on the instrument panel.  The way to look at instruments is very simple, you either see or you don't see the runway.  When you reach that number that on the approach plate has the letters DH, you may not descend below that number on the altimeter if you do not have the field in sight.  If you can see the field and as you are approaching you lose the field again, GO AROUND!  You don't try and fly the approach anymore, you just execute the missed approach procedures.  So while all the attention is being made on Prepar3D about cloud accuracy, why not worry other more important things like stability, VFR visual qualities, terrain accuracy, aircraft flight dynamics, or here is something crazy - airfields look like they are supposed to. When it comes to clouds and instruments, a professional pilot only cares that they can shoot the approach, they break out of clouds prior to Decision Height, and that they land safely - (Take offs are optional - Landings are always mandatory). 

Share this post


Link to post

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this