KA-50 Shkval targeting system

I have wondered for long time why Shkval works so differently between targets and between even KA-50 and Su-25T.

What I understand is that Vikhr missiles can't be "Locked" as some people say (there is own thread about that) as it just steers itself to middle of the laser beam, in otherwords it flies where ever pilot is guiding it.

Shkval is Soviet 'Fire and forget' targeting system, meaning that after the Shkval is locked to target and missile is fired, pilot doesn't need to keep targeting gate on target anymore like with SACLOS and its previous system MCLOS systems required. So all what pilot does is to fly aircraft without exceeding the Shkval targeting limits to maintain lock or the missile limits (launch altitude, launch speed etc).

In a sense Vikhr is exactly like many any other missile with a F&F feature like Hellfire as they too require target being maintained in line of sight by launch platform radar or laser or second party laser.

The Shkval itself is interesting as its lock capability is contrast based as far I understand. It doesn't even know what kind a target it is being pointed at. It can lock on anything that has contrast and is inside the targeting gate. After pilot press Shkval "Lock" the Shkval memories the contrast pattern inside the gate and keeps pointing at it.

Only when the contrast pattern in memory doesn't anymore match what the Shkval targeting gate has, does it lose the lock and Shkval changes to either of two modes, depending the targeting mode was used prior lock.

Either it keeps moving three seconds the last linear line to check if the pattern does appear and re-lock, or it doesn't and Shkval keeps inertial movement.

Or the second where inertial mode is enabled and Shkval gate stays at the point on ground (and target can continue moving away from that point).

As long the contrast pattern stays equivalent, Shkval keeps pointing at it. So radical target movement changing its silhouette or obstacle blocking the line of sight does cause Shkval lose the lock.

But how does the Shkval know what it is suppose to follow?

It doesnt by any other manner than the contrast pattern. Why pilot chosing the correct targeting gate size is crucial so the target pattern fills optimal area in gate.

But why does Su-25T Shkval then autolock on ground units and KA-50 doesn't? Is there some kind change done in Su-25T Shkval in reality to auto-lock on diamond shaped or square contrast patterns?

As my understanding is, it is from Lock-On level feature and shouldn't be there.

Same problem seems to exist with a KA-50 Shkval, disobedience to pilot commands to lock unless there is a unit inside a gate. Sometimes pilot can lock on some other contrasty objects like buildings or parts of buildings but i have never read or experienced that it could be locked on ground or anything when wanted.

this brings to other question, shouldn't the KA-50 Shkval lock everytime on anything that has high enough contrast pattern?

Like an enemy helicopter against blue sky, it being only a very contrasty shape against light grey background?

As then targeting gate being large enough to fill whole aircraft is important so the contrast pattern change would be a minimal. What brings a another question, does Shkval contrast lock pattern system allow specific amount of contrast pattern change?

like with a motion detection systems in secure cameras, the camera/computer takes series of shots in a second and I'd there is 5-10% (as example) change in contrast to previous two frames shapes, it detects movement. But if target change speed is slow enough, the system doesn't recognize the subject as a movement but as normal scene change.

The contrast pattern change system would be very logical as it does change if flying over or a side of the target instead toward it. Or if target is turning.

I have now last few days spent many hours flying quick missions in KA-50 only focusing to Shkval lock behavior.

And i have came to conclusion that it doesn't work as suppose at all.

I am highly probably wrong, but i can't find any reasoning why i can't get lock to hovering helicopter against light background or dark main battle tank on light ground.

or why i cant lock easily to anything very clear contrasty target?

Shkval shouldn't require lasering distance to target to achieve a lock, as it shouldn't need that data to control camera movement, as it follows contrast pattern and not inertial movement.

The accurate distance is required for other things like ABRIS map and datalink. But lock should be possible be acquired to anything where is enough contrast, no matter of targeting gate size or distance.

That way too it makes sense why laser is a switch instead automatic, as Shkval should be possible be locked without consuming laser.

Locking to helicopter as far as 15-20km should be possible as long there is enough contrast. To Shkval the target is just a specific pattern at specific direction and it follows it as long Shkval limits doesn't deny it.

Same way Su-25T Shkval should be possible used as now it auto locks to any unit inside gate, while even being inside a forest and pilot can't spot the unit.

Same way it looks that non-functional contrast locking disallows KA-50 pilot from locking Shkval to ground so easily while maneuvering.

I can be very wrong, but i believe the DCS just doesn't include actual functions to do a correct contrast detection but it is cheated as it knows when there is a unit targeted, and then it just follows specific requirements for distance. This way example A-10C TGP or Maverick can work easily as when in thermal mode or point tracking, it is simply checked is there a unit inside cross.

To me it is frustratingly huge immersion breaker when KA-50 Shkval can't lock on clear object on sky only to find out that it would be always better use mouse and guide Vikhr without Shkval lock to enemy.

Can we expect in future Shkval system being fixed?

If we get trees that offers concealment from radars and lasers and depending thickness, cover from missiles and kinectic projectiles, the Shkval system locking should be working one.

Wow ... wall of text ... ;o)

Some remarks added in blue:

I have wondered for long time why Shkval works so differently between targets and between even KA-50 and Su-25T.

 

What I understand is that Vikhr missiles can't be "Locked" as some people say (there is own thread about that) as it just steers itself to middle of the laser beam, in otherwords it flies where ever pilot is guiding it.

Yeah, Vikhrs don't lock, they just beam-ride.

 

Shkval is Soviet 'Fire and forget' targeting system, meaning that after the Shkval is locked to target and missile is fired, pilot doesn't need to keep targeting gate on target anymore like with SACLOS and its previous system MCLOS systems required. So all what pilot does is to fly aircraft without exceeding the Shkval targeting limits to maintain lock or the missile limits (launch altitude, launch speed etc).

In a sense Vikhr is exactly like many any other missile with a F&F feature like Hellfire as they too require target being maintained in line of sight by launch platform radar or laser or second party laser.

The term Fire&Forget means that the pilot doesn't have to care about the weapon at all after launch. Radar Hellfires or Mavericks (except E model) are F&F, laser Hellfires and Vikhrs are not exactly F&F.

.... (snip) ...

 

But why does Su-25T Shkval then autolock on ground units and KA-50 doesn't? Is there some kind change done in Su-25T Shkval in reality to auto-lock on diamond shaped or square contrast patterns?

 

As my understanding is, it is from Lock-On level feature and shouldn't be there.

I would guess, it is because of the SSM of FC3 aircraft.

Same problem seems to exist with a KA-50 Shkval, disobedience to pilot commands to lock unless there is a unit inside a gate. Sometimes pilot can lock on some other contrasty objects like buildings or parts of buildings but i have never read or experienced that it could be locked on ground or anything when wanted.

 

this brings to other question, shouldn't the KA-50 Shkval lock everytime on anything that has high enough contrast pattern?

Like an enemy helicopter against blue sky, it being only a very contrasty shape against light grey background?

Yes, probably ...

As then targeting gate being large enough to fill whole aircraft is important so the contrast pattern change would be a minimal. What brings a another question, does Shkval contrast lock pattern system allow specific amount of contrast pattern change?

like with a motion detection systems in secure cameras, the camera/computer takes series of shots in a second and I'd there is 5-10% (as example) change in contrast to previous two frames shapes, it detects movement. But if target change speed is slow enough, the system doesn't recognize the subject as a movement but as normal scene change.

 

The contrast pattern change system would be very logical as it does change if flying over or a side of the target instead toward it. Or if target is turning.

 

I have now last few days spent many hours flying quick missions in KA-50 only focusing to Shkval lock behavior.

And i have came to conclusion that it doesn't work as suppose at all.

... but I assume, this is some limitation in the way it is modelled. The sim does not actually analyse the contrast - instead the sim already knows if what the Shkval is looking at an object (unit or building) or not.

Also: there is a panel where you can tune the optics of the Shkval - like adjusting contrast, switching into some sort of low-light mode (and others), etc. But all functions there are documented as "not functional" in the manual.

... snip ...

 

I can be very wrong, but i believe the DCS just doesn't include actual functions to do a correct contrast detection but it is cheated as it knows when there is a unit targeted, and then it just follows specific requirements for distance. This way example A-10C TGP or Maverick can work easily as when in thermal mode or point tracking, it is simply checked is there a unit inside cross.

Yep, that is my understanding as well.

To me it is frustratingly huge immersion breaker when KA-50 Shkval can't lock on clear object on sky only to find out that it would be always better use mouse and guide Vikhr without Shkval lock to enemy.

 

Can we expect in future Shkval system being fixed?

If we get trees that offers concealment from radars and lasers and depending thickness, cover from missiles and kinectic projectiles, the Shkval system locking should be working one.

Yeah ... but the keyword here is if ... :o)

You can definitely adjust the Shkval's brightness and contrast with the knobs—if you turn brightness all the way down, then turn contrast all the way up, then adjust brightness until you can see an image, you can see at night. (It shouldn't work, because there's noise in the video system that isn't modeled, but it's a handy way to pretend we have a Ka-50N.)

You can definitely adjust the Shkval's brightness and contrast with the knobs—if you turn brightness all the way down, then turn contrast all the way up, then adjust brightness until you can see an image, you can see at night. (It shouldn't work, because there's noise in the video system that isn't modeled, but it's a handy way to pretend we have a Ka-50N.)

Yes, you are right, I did not express myself correctly. You can tune the brightness and the contrast of the TV display, but afaik it does not make a difference for the Shkval or it's ability to lock onto targets. It only helps you as pilot to see better. At night, you still have to deal with drastic reduced max lock on range and will have to use the launch override to be able to actually engage a target without a lock.

But what I meant was the rotary knob below the brightness and contrast knobs. The manual says, it deals with optical filters - which are not implemented.

Wow ... wall of text ... ;o)

 

Some remarks added in blue:

You didn't get change to use quote? Makes my quote impossible for some reason....

Well, i only touch to parts.

The Hellfire L-model is F&F in that purest form with its own active millimeter radar seeker at homing phase (from launch to that point it is inertial guided), but the F&F is as well that you don't need to guide missile as the system takes care of it, just lock, fire and forget (focus to other things). And that Shkval fills as well. Sure I have the opinion that the forget part requires that launching platform can be destroyed or gone, but if pilot/gunner doesn't need to do guidance, it is still same to them. But it isn't this topic

And I forgot to ask about the screen adjustments affecting to Shkval anyways on others? I have tried to adjust Shkval but my understanding is it would not affect it at all, as those are TV settings instead the targeting data controls. It would mean that the video feed would need to be a digital so it would be possible be adjusted that manner, being analog signal doesn't allow such adjustments to signal before it is fed to display, after Shkval system has got own feed. Unless controls goes to both.... who knows.

You didn't get change to use quote? Makes my quote impossible for some reason....

 

Well, i only touch to parts.

 

The Hellfire L-model is F&F in that purest form with its own active millimeter radar seeker at homing phase (from launch to that point it is inertial guided), but the F&F is as well that you don't need to guide missile as the system takes care of it, just lock, fire and forget (focus to other things). And that Shkval fills as well. Sure I have the opinion that the forget part requires that launching platform can be destroyed or gone, but if pilot/gunner doesn't need to do guidance, it is still same to them. But it isn't this topic

 

And I forgot to ask about the screen adjustments affecting to Shkval anyways on others? I have tried to adjust Shkval but my understanding is it would not affect it at all, as those are TV settings instead the targeting data controls. It would mean that the video feed would need to be a digital so it would be possible be adjusted that manner, being analog signal doesn't allow such adjustments to signal before it is fed to display, after Shkval system has got own feed. Unless controls goes to both.... who knows.

If the TV settings affect the Shkval data processing in the real thing ... I don't know. But in DCS they don't (see my reply to Fishbreath).

.....You can lock on Aircraft with the shkval to engage them......just saying.

Also try the most smallest Target frame size.

 

I can be very wrong, but i believe the DCS just doesn't include actual functions to do a correct contrast detection but it is cheated as it knows when there is a unit targeted, and then it just follows specific requirements for distance. This way example A-10C TGP or Maverick can work easily as when in thermal mode or point tracking, it is simply checked is there a unit inside cross.

Yes, they say real time image processing is too high CPU hungry for now.

It is weird as in Ka-50 times and other lawn mowers - 1980-1990 era computers installed in these helicopters were slower than nowadays cell phones.

Regarding size of Shkval gate and locking - I think there is bug because try to set it at minimum size and lock object far away. It will do but how is it possible if resolution of whole system is really low - how could it distinguish "pixels" in locked gate size which are not contrast enough in such distance?

As far I know in reality it could lock road trepassing (white "zebra"), puffy cloud ect. I don't know how it can keep lock on something when angle of view changes - so the image in buffer which is used to compare is old. There must be some kind of margin..... and that all with very weak CPU when we look from today's POV.

.....You can lock on Aircraft with the shkval to engage them......just saying.

Also try the most smallest Target frame size.

Yes you can but it is about pure luck when the lock happens and it is limited to Shkval laser range etc.

It isn't difficult for me to even get target gate kept over enemy aircraft while maneuvering (tweaked settings for minijoystick, combination using Shkval, HUD and HMS) but the locking not being possible is frustrating as either never getting it or burning laser because already on second or third loadout for Vikhrs.

Sometimes same thing happens for ground units, never getting a lock.

The smallest is the best option now in DCS, but in reality it is worst as the contrast pattern inside the targeting gate is most likely to change radically in short time. While the largest covering target gives change to maintain lock as pattern change would be much smaller between updates.

That has as well led me to believe that the Shkval targeting mode button for Moving Ground Target is not just to tell target being moving so Shkval can estimate lead to it, but it is to configure Shkval locking pattern and ease the pilot task to get a lock.

Example, you move targeting gate front of a target instead trying to keep gate on it. You enable moving target -mode and adjust gate to target size. When the target moves inside gate, the contrast pattern of ground/background changes to pattern of target and Shkval knows that new contrast pattern is a moving target and locks on it.

As for contrast based targeting the system needs to get optimal contrast pattern what to follow.

It doesn't know which changing pattern is the target (need to send examples) and what is background so switching algorithms would make things easier. And that would as well help to lock on aerial target as usually they are against clear background and moving fast. It would allow pilot to move gate front of a target and wait it to fly inside gate that allows Shkval autolock changing contrast pattern.

Yes, they say real time image processing is too high CPU hungry for now.

It is weird as in Ka-50 times and other lawn mowers - 1980-1990 era computers installed in these helicopters were slower than nowadays cell phones.

 

Regarding size of Shkval gate and locking - I think there is bug because try to set it at minimum size and lock object far away. It will do but how is it possible if resolution of whole system is really low - how could it distinguish "pixels" in locked gate size which are not contrast enough in such distance?

 

As far I know in reality it could lock road trepassing (white "zebra"), puffy cloud ect. I don't know how it can keep lock on something when angle of view changes - so the image in buffer which is used to compare is old. There must be some kind of margin..... and that all with very weak CPU when we look from today's POV.

Our PCs could probably do the image processing quite well - but our CPU(s) also have to do a ton of other things to do as well in DCS. Also the algorithm would have to work somewhat reliably to replicate the real behaviour ... but under different conditions: our world does in fact look quite differently than the real world. So there would probably a lot of research and experiments needed to actually simulate the image processing.

As for the changing image of a moving target: maybe the comparasation between the current image and the reference image is not done with a fixed, initial image, but with the image a few moments ago as the reference image.

Yes, they say real time image processing is too high CPU hungry for now.

It is weird as in Ka-50 times and other lawn mowers - 1980-1990 era computers installed in these helicopters were slower than nowadays cell phones.

 

Regarding size of Shkval gate and locking - I think there is bug because try to set it at minimum size and lock object far away. It will do but how is it possible if resolution of whole system is really low - how could it distinguish "pixels" in locked gate size which are not contrast enough in such distance?

 

As far I know in reality it could lock road trepassing (white "zebra"), puffy cloud ect. I don't know how it can keep lock on something when angle of view changes - so the image in buffer which is used to compare is old. There must be some kind of margin..... and that all with very weak CPU when we look from today's POV.

I don't know what is real Shkval monitor resolution but I would guess it is around high end TV from that era as military budget, so around 768x576 as PAL TV cameras had that already and it would support the quality: Russian FLIR - Laser guided anti tank missile - 9…:

The contrast based pattern lock isn't hard to do in reality. Especially if you have a dedicated computer to process the data.

As we know, even the KA-50 HUD is a slow to respond on changes, around 10-15 times a second: KA-50 In Cockpit Flight Video:

So I would guess that is the contrast pattern processing speed per frame and it is shown as each update on target gate move on Shkval screen.

I made a ugly (rough?) example how I believe the contrast detection targeting works.

First is the sample of the view on Shkval monitor, then how it more likely would look for Shkval if the targeting gate is around the view. (The last sample is how it could be today with improved processing power).

The processing power from late 80's or early 90's isnt today's standard impressive. But thinking about shkval itself, computer only needs to deal binary data (black and white, not even different shades) and to get accurate enough, only 16x16 grid is required. If grid is example 32x32 the accuracy is much better but processing is slower.

Skhval targeting:

If we guess that the targeting grid is 16x16, binary colors and it gets updated 10-15 times a second, it doesn't require much processing power to do the tracking.

And if the grid always stay 16x16, tracking can be done on any target size because optical zoom and Shkval targeting gate resizing.

The pilot is required to adjust the gate around the target. That is cropping the area away that has complex contrast noise pattern, only focusing to pilot own target he spotted visually trough Shkval video. So the pilot does the hard work visually. Then resize gate to fit target and Shkval adjust itself to that area to get optimal contrast, if it does, it informs pilot of lock.

The shkval isn't good at low light, because contrast is so low that shkval can't adjust itself to get high contrast scenery. And for that the low-light television mode would be required.

A thermal camera already gets this same thing naturally via monochrome/B&W conversion. So it is easier to track subject.

Today Shkval kind targeting can be simulated easily on client computer that is targeting. It would not require at all heavy CPU processing as even 32x32 would be possible to do thousands of times per second without impact. Doing it when player activates lock with 10-15 frame per seconds has no impact.

Contrast lock detects edges of an object and then steers the camera to point in the middle of the left, right, upper and lower edges. For the edge detection the system simply measures if there's sharp enough and large enough change in image intensity in all directions so that it can be reliably tracked. If the edge is too vague the measured position of the edge will jump around too much between measurements for the system to be able to track it. Early EO sensors did this with analog circuits which I think Shkval would be using also.

I have no doubt this wouldn't be doable on current systems for single sensor. The problems is that you would need to model every contrast locking sensor the same way which could be a problem. Imagine multiple AI A-10s ripple firing mavericks at the same time. Other thing is that there could be efficiency problems with rendering pipeline to get the sensor images to the algorithm handling lock simulation in a timely manner.

Contrast lock detects edges of an object and then steers the camera to point in the middle of the left, right, upper and lower edges. For the edge detection the system simply measures if there's sharp enough and large enough change in image intensity in all directions so that it can be reliably tracked. If the edge is too vague the measured position of the edge will jump around too much between measurements for the system to be able to track it. Early EO sensors did this with analog circuits which I think Shkval would be using also.

 

I have no doubt this wouldn't be doable on current systems for single sensor. The problems is that you would need to model every contrast locking sensor the same way which could be a problem. Imagine multiple AI A-10s ripple firing mavericks at the same time. Other thing is that there could be efficiency problems with rendering pipeline to get the sensor images to the algorithm handling lock simulation in a timely manner.

This would be something to be best implemented in a multithreading/multicore version of DCS. Nice little, relatively sharply outlined tasks that have well defined interactions with the environment. Something like 1-2 World cores, 1 sound core, 1 Player aircraft core and 1 sensors core ... or something like that. But well, this is only day dreaming these days ... :o)

Our PCs could probably do the image processing quite well - but our CPU(s) also have to do a ton of other things to do as well in DCS. Also the algorithm would have to work somewhat reliably to replicate the real behaviour ... but under different conditions: our world does in fact look quite differently than the real world. So there would probably a lot of research and experiments needed to actually simulate the image processing.

 

As for the changing image of a moving target: maybe the comparasation between the current image and the reference image is not done with a fixed, initial image, but with the image a few moments ago as the reference image.

Todays PC have a lot of power, they are mostly limited only by programmers will and their financial resources.

We don't have to do image processing of whole area visible by Shkval, only center of it (gate size) because the target is always in the center, just gate size changes.

I don't know what is real Shkval monitor resolution but I would guess it is around high end TV from that era as military budget, so around 768x576 as PAL TV cameras had that already and it would support the quality: Russian FLIR - Laser guided anti tank missile - 9…:

 

The contrast based pattern lock isn't hard to do in reality. Especially if you have a dedicated computer to process the data.

 

As we know, even the KA-50 HUD is a slow to respond on changes, around 10-15 times a second: KA-50 In Cockpit Flight Video:

 

So I would guess that is the contrast pattern processing speed per frame and it is shown as each update on target gate move on Shkval screen.

I've heard it could be even lower, around 256 lines. Regarding quality we have now Full HD standard in DCS. It is wrong, I don't know why they don't want to degrade quality of image, if even in old FC1 it was more realistic by just layered filter which degraded resolution.

Same goes for Ka-50.

 

I have no doubt this wouldn't be doable on current systems for single sensor. The problems is that you would need to model every contrast locking sensor the same way which could be a problem. Imagine multiple AI A-10s ripple firing mavericks at the same time. Other thing is that there could be efficiency problems with rendering pipeline to get the sensor images to the algorithm handling lock simulation in a timely manner.

The AI doesn't need it at all, for AI it is enough to just aim to target and depending skill level variate the release and aiming accuracy or reaction speed to aim.

Only players would need it to experience the difficulty and possibilities.

The sample I shown is just made with simple pixelisation and then then threshold 50% what is done on decades olds compute d security cameras and some targeting systems as well today in modern image recognitions algorithms as well.

It is very accurate and very fast way to recognize subjects from video or photo. It can even be used to recognize objects by colors.

16x16px goes very far and can be applied even in very crucial situations like cars early warning systems avoiding other cars or humans (or animals).

Sure some might use edge detection but it is far more complex calculations to be done and use.

As for the changing image of a moving target: maybe the comparasation between the current image and the reference image is not done with a fixed, initial image, but with the image a few moments ago as the reference image.

That is what I wrote, as we can do it like on security cameras indoor as otherwise you an cheat them, why they use predefined sample. But outdoor and anything tracking something moving requires sampling from few previous seconds at top and then sometimes even to original sample (security cameras).

Like compare last 10-15 frames and subject can turn pretty fast without losing a lock.

Our PCs could probably do the image processing quite well - but our CPU(s) also have to do a ton of other things to do as well in DCS. Also the algorithm would have to work somewhat reliably to replicate the real behaviour ... but under different conditions: our world does in fact look quite differently than the real world.

The old computers could do it pretty fast and analog computers did it well. But the impact really is very very small as we are talking about processing center of targeting gate and very limited resolution and time it is in use. Even today DCS runs well enough (until something blows up next to you) and it is just with a two threads (graphics+simulation & sound, as far I know, internally dozens of threads fighting about priority) while good multitasking programming should give huge improvement to current system, releasing even more processing power from each client computer to do simple calculation from their own computer. Same way as rest of the simulations, we don't need to send data of all aircraft functions to every client and server, it is enough to do simulation on client and send result to server and other clients. Like no need simulate fuel tank fill rate or fuid momentums on server and other clients, it is enough client does it and then send result how aircraft behave at that moment to others.

And I would draw a line to the simulation depth/accuracy as replicating the algorithms the original engineers and mathematics designed and build own computer for it did, isn't viable. The current system looks and behaves badly after longer time using it, but it has done its work well enough.

It is enough to know how it should work (contrast lock, lock on everything where is contrast and in what situations comes problems etc) simply offer that.

Todays PC have a lot of power, they are mostly limited only by programmers will and their financial resources.

We don't have to do image processing of whole area visible by Shkval, only center of it (gate size) because the target is always in the center, just gate size changes.

exactly as the gate is always adjusted by pilot around target, the low resolution and sampling is enough. In my video the first sample is how huge the full view would required to be. The last sample is like only gate s size is calculated in two ways.

I've heard it could be even lower, around 256 lines. Regarding quality we have now Full HD standard in DCS. It is wrong, I don't know why they don't want to degrade quality of image, if even in old FC1 it was more realistic by just layered filter which degraded resolution.

 

 

Same goes for Ka-50.

I like to keep it in 512 and skip frames.

The 256 lines sounds way too low to identify anything.

In all videos and pics provided by claiming it to be "shkval" it seems its quality is really low - in this movie object is 2-3 km and it looks bad so I more less imagine how it would look like in 10 km :D

I don't know if ED knew what is the real resolution of this system in Ka-50 when they were modelling module.

Regarding 250 being too low - again the video and pics aren't too good evein in such close object, so perhaps?

In all videos and pics provided by claiming it to be "shkval" it seems its quality is really low - in this movie object is 2-3 km and it looks bad so I more less imagine how it would look like in 10 km :D

I don't know if ED knew what is the real resolution of this system in Ka-50 when they were modelling module.

Regarding 250 being too low - again the video and pics aren't too good evein in such close object, so perhaps?

The quality in those are actually good as you can recognize target, target directions, markings and even individual parts of those units at that size on monitor.

It isn't the distance that is problem, but the object size relatively to gate.

The display can be low resolution, but we don't know how accurate the camera itself is. Low output doesn't mean input is low. As the Shkval isn't tracking it trough the monitor, but trough the feed that is sent to computer before monitor.

And soviets claiming 10-12km engagement ranges would require a good tracking/targeting capability, that is now for some reason dropped to 6-8km.

There's a cockpit video around here somewhere that shows actually really good resolution and even colour on a Ka-50 shkval display. The original displays may have been low res, doesn't mean that all I them were. Remember we're basically flying a prototype. Each was different