Jump to content
Sign in to follow this  
Guest

Tips for those suffering with cloud shadow performance

Recommended Posts

Thanks Rob.  This is just what I'm interested in tweaking today.  I don't quite understand the cloud shadow 10,000 etc. slider, but it seems to be the on/off switch for cloud shadows, and it cuts my fps in half.  In other words, with it at 0 there are no shadows, and at 10,000, the first setting, there are shadows at great cost of framerate.  I'll try your recommendations when I get a chance.  Thanks.

 

This is probably just the thing SLI will help much with too, eh?

Share this post


Link to post
Guest

 I don't quite understand the cloud shadow 10,000 etc. slider

 

 

Just to be clear, I'm talking about setting weather cloud layer altitudes ... this is done in the weather section.  I'm NOT talking about setting Cloud Shadow distance values.

Hopefully this image will make it clear:

 

d1fe108d1b7304343ae934e81f5760c5.jpg

 

Cheers, Rob.

Share this post


Link to post

I assume that's only releavent if your not using a weather engine such as ASN or Opus?

Share this post


Link to post
Guest

 

 


I assume that's only releavent if your not using a weather engine such as ASN or Opus?

 

It's relevant to understand ... hopefully weather engines (ASN etc.) providers will update to support this performance optimization.  I don't think any of them currently do -- if they do, I'm not seeing the benefits nor the option in their settings.  

 

But yes, your average user will not be able to influence how ASN, Opus, etc. work ... so for those not wanting to mess with custom weather, 3 layer max works pretty good.

 

Cheers, Rob.

Share this post


Link to post

Intesting observations Rob - have you already cross-posted / mentioned this to Stephen @ Opus? 

Share this post


Link to post
Guest

 

 


have you already cross-posted / mentioned this to Stephen @ Opus?

 

No - pretty sure he/they might have already discovered this.

 

But there are still issues with time of day and cloud shadows ... I believe LM have confirmed ... don't know if they are planning to address.  My suggestion/tips are just here to help people work around it for now.

 

Cheers, Rob.

Share this post


Link to post

 

 


Hopefully this image will make it clear:

Rob,

Great information and thanks for that.  By the way, I am wondering if you can answer a question for me.  In your screenshot the base altitude seems to be always set to -1461, and that does not make much sense to me.  Do you know what that actually means?


spacer.png

REX AccuSeason Developer

REX Simulations

Share this post


Link to post
Guest

 

 


 In your screenshot the base altitude seems to be always set to -1461, and that does not make much sense to me.  Do you know what that actually means

 

That's percipitation ... rain doesn't always make it to the ground (real world) ... but if you DO want the rain to hit the ground, the best way to ensure that is a negative altitude.  In the image I posted, rain is not present so the -1461 is not relevant.

 

Cheers, Rob.

Share this post


Link to post

I have ASN and I found that decreasing cloud density from does more to improve rainy day performance than reducing cloud layers and cloud visibility distance.

 

I just tested this again in real weather in WA (near 4WA8), where it was heavily overcast and rainy steady.
Reducing cloud density from Ultra to Medium improved my FPS about 50%.

Reducing cloud layers from 5 to 3 and cloud visibility distance from 90 to 80 didn't improve my FPS more than 10%.

 

I guess this in another one of those things that varies greatly between systems.


~ Arwen ~

 

Home Airfield: KHIE

Share this post


Link to post
Guest

 

 


Reducing cloud density from Ultra to Medium improved my FPS about 50%.

 

Yes, you'll have a lot fewer clouds.  

 

Did your testing include leaving the cloud density at Ultra and moving layers to 3 and distance to 80 to note fps difference.

 

It's also important the same time of day is used ... for example (because of shadow cloud issue ... don't want to call it a bug) at 7am I will get 60% performance drop vs. at 11am - only change is time of day.

 

But for custom weather, my important find was the cloud layer base and tops values -- keeping the difference at around 1000 ft per cloud layer.  If I increase the difference (cloud layer thickness - NOT density) it would drastically impact performance.

 

example cloud layer type cumulus:

base 4000, top 5000 = 32 fps

base 4000, top 10000 = 15 fps

 

Cheers, Rob.

Share this post


Link to post

Yes, you'll have a lot fewer clouds.  

 

Did your testing include leaving the cloud density at Ultra and moving layers to 3 and distance to 80 to note fps difference.

 

Actually I still had full overcast . . . as I was flying below the cloud level (which is prudent when you are flying a Piper Cub in limited visibility).

 

I tested first with cloud density at Ultra, changed cloud layers from 5 to 3 and cloud visibility distance from 90 to 80. When I saw little improvement, I reduced cloud density and then I changed cloud layers back to 5 and cloud visibility distance back to 90 (and my FPS remained essentially the same). I was flying both times with live ASN weather . . . loading a saved flight. My two flights were only about 15 minutes apart, and the weather condition were unchanged (according to the ASN weather conditions panel).


~ Arwen ~

 

Home Airfield: KHIE

Share this post


Link to post

Oddly, I went to experiment a little bit last night, and used 1 cumulus layer, 4/8,  1000-3500 feet. I tried increasing tops to 4500, 5500, and 6500, but saw no apparent increase in the height of the clouds.

 

Am I missing something here?

 

thanks

Andrew

Share this post


Link to post

It's relevant to understand ... hopefully weather engines (ASN etc.) providers will update to support this performance optimization. I don't think any of them currently do -- if they do, I'm not seeing the benefits nor the option in their settings.

 

Hello Rob,

I am getting a consistent 30fps locked, unlocked it is higher but jumps about a bit.  with clouds set at high, distance set at 90  with cloud shadows @ 60,000m. ASN beta 3 layers no overcast enhancement.

This is with a lowly GTX570.

The results:

cloud_cover.jpg

sun_peeping_through.jpg

 

Share this post


Link to post

Hi,

 

a small comment by me. First Rob, as noted by Andrew, setting the cloud thickness (either programmatically or through fsx/p3d menus), does not match in most cases (especially if cumulus clouds are used) the actual "visible" thickness you'll get. The actual thickness is determined by the cloud model selected by the sim (and one of the parameters taken into account is the "dictated" thickness). So, restricting the thickness to 1000 feet will force p3d to select a "lighter" model (with less plane/sprites included). The aggregate sprite/plane count in the scene in addition to the sun angle are the decisive things affecting the frames in this case (and this was also the reason reducing the cloud density had such a significant impact in Arwen's case. In the latter case however, obviously overcast conditions cannot be depicted properly. 

 

IMHO, artificially reducing the cloud thickness just to preserve frames is not the way to go here (we spent many- many developer hours for example to be able to produce realistic storm/hurricane depictions). There are other optimizations that can be done and we'll work with LM to test case/fine tune things moving forward. Having said that, I'd like to say that we're very excited with all the new innovations and the way LM is pushing forward the sim development (and in particular "our" part- weather in that regard). 

 

@mad dog, it might be interesting to share your prepar3d.cfg file. There must be another setting used in combination (in addition to cloud shadows), that brings our machines to their knees  (I don't know, AA setting? tesselation?). Getting these perfectly acceptable frames with your graphics card is worth the investigation. I've got a seriously weak developer machine (6 yrs old Q9550 and a 660GTX) and just by enabling the cloud shadow to 10km in ASN manually produced OVC conditions, the gpu load (in window mode as measured by NVI), jumps from 35-40% to 85-90%, but still keeping ~30fps. In full screen mode it drops to about 20-22. But when in dusk/dawn it drops down to 8-9 fps (showcasing that the time of day is equally important to other things in determining the final fps count).

 

Don't forget that (as emphasized by LM devs), being more GPU bound means that the performance is much more resolution dependent (e.g. some super sampling AA settings effectively double this resolution). 


Kostas Terzides

 

devteam_bannerA.png

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  
  • Tom Allensworth,
    Founder of AVSIM Online


  • Flight Simulation's Premier Resource!

    AVSIM is a free service to the flight simulation community. AVSIM is staffed completely by volunteers and all funds donated to AVSIM go directly back to supporting the community. Your donation here helps to pay our bandwidth costs, emergency funding, and other general costs that crop up from time to time. Thank you for your support!

    Click here for more information and to see all donations year to date.
×
×
  • Create New...