Nate--IRL-- Posted May 10, 2012 Posted May 10, 2012 Good stuff, thanks. Nate Ka-50 AutoPilot/stabilisation system description and operation by IvanK- Essential Reading
Harzach Posted May 10, 2012 Posted May 10, 2012 (edited) Quick track file attached. Air start, default summer weather with a cloud base of 2000M. Graphics settings: GPUs are in Crossfire/AFR mode. Sorry for the crazy zoom-fest in the beginning, I didn't have time to get properly situated with TIR and all, plus my absolute zoom (on the TMWH slider) was not neutral. Nate, I think you mentioned that you know what is causing the cloud "stutter", so I'll just offer my observations regarding the "spinning". It seems that DCS uses cloud "sprites" (similar to that which MSFS used), which continually orient themselves toward the pilot. Somehow, the pilot's view angle/relative position/etc also have a part in determining this orientation. The closer you get to a cloud, the more apparent the effect is. Skim the top of a cloud and scan your view left-right a few times and you should see it quite clearly. This was captured in the track.funky clouds.trk Edited May 10, 2012 by Harzach
Shaderhacker Posted May 10, 2012 Posted May 10, 2012 (edited) Quick track file attached. Air start, default summer weather with a cloud base of 2000M. Graphics settings: GPUs are in Crossfire/AFR mode. Sorry for the crazy zoom-fest in the beginning, I didn't have time to get properly situated with TIR and all, plus my absolute zoom (on the TMWH slider) was not neutral. Nate, I think you mentioned that you know what is causing the cloud "stutter", so I'll just offer my observations regarding the "spinning". It seems that DCS uses cloud "sprites" (similar to that which MSFS used), which continually orient themselves toward the pilot. Somehow, the pilot's view angle/relative position/etc also have a part in determining this orientation. The closer you get to a cloud, the more apparent the effect is. Skim the top of a cloud and scan your view left-right a few times and you should see it quite clearly. This was captured in the track. Pardon me for jumping in.. I'm a graphics programmer and what you describe is very common in games. I have yet to see any game render out real volume clouds (not to mention light scattering and self-shadowing) as it takes up too much texture memory and will still have a limited number of layers (i.e. you'll still see a popping affect when going through them). The orient toward camera sprites are used to cut down rendering time since the GPU can do it so easily. This is done for hair primitives in the animation industry most of the time as well. In short, I think the only way to remedy this would be to implement the same algorithm that the Ubisoft developers did for their clouds in H.A.W.X. It will get rid of the orient-to-camera sprite rendering. However, the stuttering is definitely a bug or hardware issue. Edited May 10, 2012 by Shaderhacker
Nate--IRL-- Posted May 10, 2012 Posted May 10, 2012 Nate, I think you mentioned that you know what is causing the cloud "stutter", so I'll just offer my observations regarding the "spinning". It seems that DCS uses cloud "sprites" (similar to that which MSFS used), which continually orient themselves toward the pilot. Somehow, the pilot's view angle/relative position/etc also have a part in determining this orientation. The closer you get to a cloud, the more apparent the effect is. Skim the top of a cloud and scan your view left-right a few times and you should see it quite clearly. This was captured in the track. The strange thing is they don't do that at all for me - I am absolutely stumped why they do it for you. The stutter again I don't see, but it was a feature of earlier Lock On titles. I'm uploading a short video from that track to Youtube (Slow upload give it time) qqxAOfdyFWg Humour me and try, if you haven't already tried, a single card instead of Crossfire to see if that makes a difference. Nate Ka-50 AutoPilot/stabilisation system description and operation by IvanK- Essential Reading
Harzach Posted May 11, 2012 Posted May 11, 2012 Humour me and try, if you haven't already tried, a single card instead of Crossfire to see if that makes a difference. No difference.
xxJohnxx Posted May 11, 2012 Posted May 11, 2012 I think I have the same problem as the op. Yesterday I flew trough bad weather in DCS BS2, and I could clearly see the cloud spinning arround its center, when moving my view. However I haven't noticed this behaviour in DCS A-10, only in BS2 and bad weather. But I will do some tests later, if it also happens in DCS A-10. I have v1.1.1.1 installed. My system: i5 760 GTX 560 Ti 1GB 8 GB RAM Intel Mainboard I can make a track if it is nessesary. Check out my YouTube: xxJohnxx Intel i7 6800k watercooled | ASUS Rampage V Edition 10 | 32 GB RAM | Asus GTX1080 watercooled
Nate--IRL-- Posted May 11, 2012 Posted May 11, 2012 I think I have the same problem as the op. Yesterday I flew trough bad weather in DCS BS2, and I could clearly see the cloud spinning arround its center, when moving my view. However I haven't noticed this behaviour in DCS A-10, only in BS2 and bad weather. But I will do some tests later, if it also happens in DCS A-10. I have v1.1.1.1 installed. My system: i5 760 GTX 560 Ti 1GB 8 GB RAM Intel Mainboard I can make a track if it is nessesary. Try to see if the options are different between Warthog and BS2 - Also, more importantly, see if they are using different profiles for your graphics card settings. Since it is happening in one and not the other, on the same computer, I think we may be close to the reason behind it. Nate Ka-50 AutoPilot/stabilisation system description and operation by IvanK- Essential Reading
ALDEGA Posted May 11, 2012 Posted May 11, 2012 It's not a stutter in the sense of sudden performance change, it looks like the cloud objects change depending on distance, call it LOD's if you must. The clouds consist of "sprites" which obviously rotate along with the viewport. Same thing goes for other effects like smoke of burning vehicles or the smoke of the GAU-8.
Nate--IRL-- Posted May 11, 2012 Posted May 11, 2012 I understand about the performance change, not exactly LOD, but sufficient for describing this. The thing is I don't see it. Neither do I see Rotating Sprites for any effect. I want to know why this is. Nate Ka-50 AutoPilot/stabilisation system description and operation by IvanK- Essential Reading
ALDEGA Posted May 11, 2012 Posted May 11, 2012 I think we may be close to the reason behind it.By design™
Nate--IRL-- Posted May 11, 2012 Posted May 11, 2012 By design™ No, I mean why the orientation of those sprites stays fixed to the game world co-coordinates in WARTHOG and fixed to the screen co-ordinates in BS2, on the same computer. There has to be a reason for it. Nate Ka-50 AutoPilot/stabilisation system description and operation by IvanK- Essential Reading
Harzach Posted May 11, 2012 Posted May 11, 2012 In short, I think the only way to remedy this would be to implement the same algorithm that the Ubisoft developers did for their clouds in H.A.W.X. It will get rid of the orient-to-camera sprite rendering. However, the stuttering is definitely a bug or hardware issue. IL2 (also Ubi) had some decent volumetric clouds, as I recall.
ALDEGA Posted May 11, 2012 Posted May 11, 2012 No, I mean why the orientation of those sprites stays fixed to the game world co-coordinates in WARTHOG and fixed to the screen co-ordinates in BS2, on the same computer.It's the same for DCS:BS and DCS:WH as far as I can see. Clouds, smoke, contrails, fire, it's all the same. In attachment you can see the same method used for thick clouds (same as Lockon, 2003 AD). These are fixed so they don't move. You can also notice that the canopy isn't fully rendered. It was only rendered up close and rotating the view affected how much of the canopy was rendered ...
Nate--IRL-- Posted May 11, 2012 Posted May 11, 2012 (edited) It's the same for DCS:BS and DCS:WH as far as I can see. Clouds, smoke, contrails, fire, it's all the same. . Not for xxJohnxx above - he has both scenarios (Spinning and non-spinning) on the same computer. I know they use the same method I want to know why it works correct for one and not the other. Nate Edited May 11, 2012 by Nate--IRL-- Ka-50 AutoPilot/stabilisation system description and operation by IvanK- Essential Reading
kokoweems Posted May 11, 2012 Posted May 11, 2012 I see the spinning clouds on the aircraft start-up training mission. The clouds behind and above the A-10. It is quite annoying.
xxJohnxx Posted May 11, 2012 Posted May 11, 2012 (edited) Okay, I did some tests now, and I am not sure if it is exactly what I have said above. But first my settings: DCS A-10: Textures: High Scenes: High Civ Traffic: Low Water: Medium (didn't see spinning clouds in high either) Visib range: High Heat blur: On Shadows: Medium Resolution: 1920x1080 Aspect Ratio: 16:9 Monitors: 1 Screen Res. of Cockpit Displays: 1024 every frame MSAA: 8x HDR: Normal Shadow Tree: False VSync: False FullScreen: True Cockpit Shadows: True TSAA: False DCS BS2: Textures: High Scenes: Medium Civ Traffic: Medium Water: Medium (didn't see spinning clouds in high either) Visib range: High Heat blur: On Shadows: Medium Resolution: 1920x1080 Aspect Ratio: 16:9 Monitors: 1 Screen Res. of Cockpit Displays: 1024 every frame MSAA: 8x HDR: Normal Shadow Tree: False VSync: False FullScreen: True Cockpit Shadows: True TSAA: False Okay, I have attached two missions, one for the A-10 and one for the Ka-50. Both have the same weather. Additionally I have attached the track from BS2. Every time I saw clouds spinning, I shake my head left and right, which makes the effect visible on my system. After I passed the cloud I fired some rounds with the gun, just to make sure that you notice that I have seen the effect here. Test results Ka-50: I flew through clouds at nearly maximum speed, and several seconds before I entered the cloud, I saw that it started spinning. In the test mission I made I had to be really close to see the effect, however in the second mission of the Operation Madvedev II Campaign I could see the spinning effect from a distance of 500 meters, or even more. Test resulsts A-10: In the same weather, in the A-10 I tryed to fly the same way like in the Ka-50. With flaps extended and gear out, I flew with about ~ 220 kph (110 knots) through clouds, and I wasn't really able to make out the effect. I saw it only once, just in the second I passed the cloud. However the effect is not like it is in BS2, where I can see the cloud spinning, before I am in the cloud. I hope I was able to provide enough informations to help you. If you need anything more, just name it, and I will do my best to help. Best regards, John BTW: I swear I can fly the Ka-50 better than I did in the track...:music_whistling:Clouds_Test_Kamov.trkClouds_Test_Kamov.mizClouds_Test_Hog.miz Edited May 11, 2012 by xxJohnxx Check out my YouTube: xxJohnxx Intel i7 6800k watercooled | ASUS Rampage V Edition 10 | 32 GB RAM | Asus GTX1080 watercooled
ALDEGA Posted May 11, 2012 Posted May 11, 2012 You moved the view rather quickly so I couldn't really see it in that track. Do you apply specific settings in the video driver related to textures?
xxJohnxx Posted May 11, 2012 Posted May 11, 2012 Okay, sorry for moving my head to fast there. You may can see the clouds if you hit alt + Z (or Y, depending on keyboard), to run the track at half or even less speed. I have NVIDIA Inspector installed, but I haven't changed anything there for DCS and the _GLOBAL_DRIVER_PROFILE (Base Profile) should still be default. I have the GeForce290.53 - [r290_27-18] drivers installed if that matters. John Check out my YouTube: xxJohnxx Intel i7 6800k watercooled | ASUS Rampage V Edition 10 | 32 GB RAM | Asus GTX1080 watercooled
Recommended Posts