Back to searchlights

Further to above, I have gone through a whole lot of different searchlights I have; I have yet to test all but did test ones called
dr_searchlight_sb and dr_searchlight_sw
Both use an effect <fx_searchlight ClassName="GroupEffect" Effect0="searchlight_halo" Effect1="searchlight_fireball" effect2="searchlight_reflected_light" effect3="searchlight_shockwave"/>
which I note was in my effect file.

tested in facility tk_flak_battery_1 replacing gc_searchlight_white in 2 different missions


dr_searchlight_sb and gc_searchlight_white showed hardly visible light BUT illuminated the aircraft, mainly the top side, but somewhat on the correct underside.

dr_searchlight_sw showed full lights which moved; quite erratically at first, but then somewhat tracked the aircraft BUT no illumination of the aircraft.:dizzy:


I have yet to test a "g_searchlight" which uses an effect called

<fx_ambientlight ClassName="GroupEffect" Effect0="Light_l"/>
<Light_l ClassName="Lighting" LightType="Point" Lifetime="9000" InitialDelay="0" FadeInTime="0" FadeOutTime="3" StrobeInterval="0.01" StrobeRitm="1" PosX="0" PosY="2.1" PosZ="0" RotY="0" RotX="0.0" Diffuse0="0 0 0" Diffuse1="231 228 254" Specular0="0 0 0" Specular1="231 228 254" Ambient0="0 0 0" Ambient1="151 143 254" Range="12" FallOff="3" Attenuation0="1" Attenuation1=".01" Attenuation2="0" Theta="10" Phi="30"/>
 
Illumination of aircraft works differently with Ankors shaders than with the original GC concept. I'll try to explain the differences.

GC Searchlights
-Model functions as a flak gun with the beam always pointing the same direction as the gun node.
-Beam illumination is accomplished by making the beam out of highly reflective but mostly transparent material. Ambient lighting present with stock shaders was enough to get the beam model to reflect the light. This does not work as well in Ankor's shaders because the ever present ambient light source that illuminated the beam in the stock shaders has been removed in favor of more realistic lighting in Ankor's shaders.
-Aircraft illumination is accomplished by having the gunstation in the model calling for a searchlight gun in the xdp that fires a shell that produces a lighting effect, but no damage or other visual effect. It does not necessarily light the aircraft from below through the beam, but from wherever the searchlight "flak shell" happens to burst (above, below, to the side, etc.)

Ankor Shaders Searchlights
-Model functions as a flak gun with the beam always pointing the same direction as the gun node, just as above. HOWEVER, the beam (and moving gun node) must also be aligned along the negative X axis, i.e. it starts near X = 0 and ends near X = -3000 or -5000 etc depending on desired length. Note: the beam in game will display at twice the length of the beam that is actually modeled.
-Beam illumination is accomplished by mapping a semi-transparent texture to the beam. The name of this texture is added to the TextureMagic.ini in the shaders30 folder under the [Searchlight] section. Ankor's shaders will see the texture name and know to illuminate it.
-Aircraft illumination is accomplished automatically by Ankor's shaders any time an aircraft passes through the beam. If the beam has been modeled as specified above, the light will always shine from the source at the searchlight itself, and only illuminate objects within the beam, and always from the correct direction. No lines need to be added to the effects.xml whatsoever to make this work. A searchlight gun is still required to be called in the xdp, but its only purpose is to give the AI gunner a weapon to point at you and try to shoot you with. It does not and should not contribute to the visual effect in any way.

Hopefully that clears up some confusion about what we're trying to accomplish and what the components and requirements are. A successful searchlight will have to be modeled according to those specifications and track enemy aircraft reasonably aggressively and accurately, such as how current flak guns operate for us now.

We successfully got the searchlight modeled correctly to produce the correct visual effects, but said searchlight failed to track targets as well as a flak gun in spite of being set up just like a flak gun. This has led me to wonder if there is a conflict with how Ankor's shaders require the beam to be on the negative X axis and whatever CFS3 needs in a model to make the gun track a target correctly, i.e. does CFS3 require the gun node to be aligned with a specific axis other than the negative X axis, (Y axis perhaps?) or can the gun node be aligned on any axis and still track targets correctly?
 
According to the SDK, there are specific axes for guns:

gun#_l_r
gun rotation, Y-axis in line with barrel, blue arrow pointing up, red arrow pointing right and green arrow pointing towards the gunner

gun#_fore_aft
gun elevation, Z-axis to the right of the gunner, so blue arrow to the right of the gunner, red arrow pointing up and green arrow pointing towards the gunner, child of gun#_l_r,

gun_grp#b#
emitter for gun, child of gun_grp#b#, Y-axis points towards target
 
Could it be that the searchlight.y and searchlight.x references in the code I posted above just need to be swapped? Or perhaps change float3 vec = worldPos.xyz - arrSearchLights.vWorldPos.xyz; to float3 vec = arrSearchLights.vWorldPos.xyz - worldPos.xyz;
 
Thanks Joost, that seems to confirm my suspicions. It may be that this was intentional for WOFF, to prevent the searchlights from being accurate, as I don't think there was much in the way of technology to help direct them in WWI. By WWII there was much better technology in place to get the beams on target.

Andy, I agree, somehow the shaders have to be changed to put the beam along the (positive?) Y axis rather than the negative X axis. Hopefully it's as you suggest with a simple edit of the .fx file.
 
Thoughts

I have an XP install and the win10 with Ankor's shaders. I develop in XP and test in win10. This is what I have found, in XP the lights track about the same if not exactly the same as they do in the win10 w/ the new shaders. So as far as I can tell there is no accuracy difference, appearance yes, but how they track not really. Without the older versions of the lights, the planes that I used to test the lights (all in the win10 with the new shaders) were only silhouetted by the light beams and never "light up", that is until I installed the older effects code the planes were just dark images in the light beams. Once I installed the GC code the planes were then in light and reflected the bursts.

The beams themselves have taken on a different look with the new shaders and some work (In my opinion) is needed to make them appear more realistic if that is even possible.

I have a set of lights that we worked on last summer or so but as with all the versions of the lights they do not track.

Want you to think about something for minute - the guns we have on planes even if the rate is high still have limited accuracy and tracking ability - I believe that is a coding scheme by MS to represent the difficulty in aerial combat. I offer that unless there is new code developed for a gun that addresses this or a work around, I doubt any improvements will be possible in how these lights track - limitations of the sim. Of course I just push polys around and coding is not my skill so I am most certainly handicapped by my limited knowledge.

So for me, I like the older version since in my install (most likely I have done something wrong) the reflected light although random at times is far superior to the silhouetted appearance without these light bursts.
 
gecko said:
Andy, I agree, somehow the shaders have to be changed to put the beam along the (positive?) Y axis rather than the negative X axis. Hopefully it's as you suggest with a simple edit of the .fx file.
Click to expand...

Unfortunately it is impossible, there's a bit of code inside the dll which specifically gets light direction from the X axis.

I don't know how they implemented tracking in WOFF, but it is quite bad at times. I recall being very annoyed when I tested those lights that sometimes they couldn't even track a Zeppelin at 1000m.
 
Lights

Given what Ankor just added to the conversation, tracking seems random at best and once in a while even a blind squirrel finds a nut. For me, illumination of the plane is far more important and if that happens sporadically and only briefly then it is the best we can do, unless someone else can develop a work around. I will tighten up the models we worked on last year to get the beam height correct and the textures working as best as I can. I would be happy to work with anyone on this.

Ted
 
The tracking on the WOFF searchlights is poor because the beam was modeled along the wrong axis for what CFS3 needs.

The GC ones tracked targets pretty well because their beams were modeled along the Y axis.

Unfortunately without an update to the dll we have to choose between correct tracking and Ankor's improved visual effects. I'll review this over the weekend when I have time to look again at the different models we have.
 
Just worth considering.
The accuracy of ( I assume radar controlled) searchlights was never very high in WWII; I am assuming less than 50% chance of one particular aircraft being lit up. I'll try and did into that.
As I mentioned in the posts above, the lighting up of the aircraft is often on the top side which is inaccurate and must also be resolved.

Ted what was meant to be in that zip file or is it irrelevant now?

EDIT: QUOTE

A recent analysis of searchlight operations led to the following conclusions: German antiaircraft defenses rely mainly on unseen methods of control but augment their fire by visually controlled guns, using searchlights only when there is little or no cloud. Among aircraft coned by lights for more than 20 seconds (and therefore probably engaged visually), the percentage damaged has been about twice as high as among planes illuminated for a shorter period. There was no evidence that those coned for more than 20 seconds were subjected to more intense antiaircraft fire than others. The risk of being illuminated by searchlights seemed to be about the same at all bombing altitudes (6,000 to 20,000 feet). On one occasion when conditions were favorable for searchlights, there were 70 to 80 bombers over a target at one time and they were effectively coned (i.e., for more than 20 seconds) at the rate of about one per minute. Heavy antiaircraft fire in coordination with searchlight cones is extremely accurate and destructive. Once a cone centers on a plane, it ignores all other aircraft and proceeds methodically to direct the destruction of the one it has caught.

ALSO

On clear nights, when in belts to aid fighter interception, the usual tactic was to illuminate the target by directing beams vertically too produce a wall of light against which enemy bombers would be visible to fighters attacking from the rear, or to compel the bombers, as they ran the gauntlet of lights, to fly so close that they became visible from the ground, thus enabling other lights to engage them.
 
Last edited:
perspective

James - I prefer the older searchlights and the illumination they provide albeit seemly om all surfaces at times. Unless the revised lights actually illuminate the planes then the older models are superior in my mind. Tracking will never be precise or reflect actual operations. I have the lights we made last year and never tested them since I did not have the new shaders working. Dan will look at these and see what else needs to be done as he said the textures were fine, so I believe that we could combine the two types if you want, personal preferences.

Need to stop fooling around with this stuff and finish the planes I have...really distracting this other stuff is....LOL
 
Hold the phone folks, I got something to work!

Ted, I was using one of the models you made a few months back that was fully compatible with Ankor's shaders and got it to accurately track targets and hold them in the beam, with skill level determining how fast they could acquire a target and how long they could hold it, but it worked on all skill levels. I set up the searchlight gun as a regular cannon instead of a flak gun firing a bursting shell and set up the searchlight vehicle picktype as "support_truck". This kept the AI from trying to create a field of flak bursts around its target and instead focused on it directly. Targets were nicely illuminated from below when within the beam, and searchlights within the same facility worked together to successfully "cone" targets consistently. On ace setting it seems like they matched James' description of the radar guided lights fairly well. Searchlights in separate facilities operated independently. I suspect that if I added flak guns to the searchlight facility that they would concentrate their fire on whatever target the searchlights had coned since the facility seems to pick the target for all the lights.

So it looks like we should be able to have our cake and eat it too after all. It also seems like we can have it track with a variety of degrees of accuracy ranging from a semi-random search patterns of individual searchlights to radar-guided cooperation of multiple searchlights and flak. I only had time for a few short tests this evening, but I'll flesh it out more over the weekend.

ny3czOg.jpg
 
That looks fantastic!:ernaehrung004::applause:. Now I have talked to Theo Boiten on 12 o'clock high forum and another guy. He has and is releasing stuff on searchlight assisted results (expensive to buy but I might). The point being that. if possible, we need 2 sets date related. 1. Prior radar controlled (tracking by sound) and 2., radar controlled as the flak. I am assuming, from what you said, that is possible. The picture looks really great!
 
You must log in or register to reply here.
Back
Top