Some tank duel tests (CMBN)

[Simple Wine]

On 1/16/2022 at 8:39 PM, KGBoy said:

Are we sure this is the result of zeroing? I would be curious if the non-firing tank had an Armour Target Arc on that spot if this would improve its' hit ratio?

I'm sure and the non-firing tank had an armor target arc as the other two tanks had. I used this technique to ambush enemy armored vehicle in my games (you need just fire at the ground from the MG for a minute to have an acceptable zeroing). At the distance of 1.5-2 km the ability to hit first is an advantage.

[Redwolf]

43 minutes ago, Simple Wine said:

I'm sure and the non-firing tank had an armor target arc as the other two tanks had. I used this technique to ambush enemy armored vehicle in my games (you need just fire at the ground from the MG for a minute to have an acceptable zeroing). At the distance of 1.5-2 km the ability to hit first is an advantage.

How did you determine that your first-shot-hit-probability went up?

[Simple Wine]

24 minutes ago, Redwolf said:

How did you determine that your first-shot-hit-probability went up?

I made a test which was published here earlier. 

[Bulletpoint]

10 hours ago, Simple Wine said:

I'm sure and the non-firing tank had an armor target arc as the other two tanks had. I used this technique to ambush enemy armored vehicle in my games (you need just fire at the ground from the MG for a minute to have an acceptable zeroing). At the distance of 1.5-2 km the ability to hit first is an advantage.

How far from the zero-in target point does the tank get an advantage? That is, what if the enemy tank pops up 200m to the right and 150m more distant?

[Drifter Man]

On 1/17/2022 at 4:39 AM, KGBoy said:

Are we sure this is the result of zeroing? I would be curious if the non-firing tank had an Armour Target Arc on that spot if this would improve its' hit ratio?

I doubt that the author of the test would deliberately subvert his setup in that way  Nevertheless, I have seen no indication that the target arcs improve spotting* or hit accuracy. See point 4 in my initial post in this thread. They just restrict the area in which the unit will engage targets, that's all.

*other than as a result of the unit turning in the direction of the target...

 

[Drifter Man]

2 hours ago, Bulletpoint said:

How far from the zero-in target point does the tank get an advantage? That is, what if the enemy tank pops up 200m to the right and 150m more distant?

TRPs seem to work in a 50m radius for artillery. I don't know whether this applies to non-artillery units as well, but I assume it does.

They work even after the unit has moved, including on-map artillery. I have been able to call missions on TRPs with on-map mortars that have moved from their initial positions. I have no explanation for the information on the page @Erwin linked to other than that it might be for CM1. In CM1, units that have moved from their initial setup positions lost the ability to use TRPs.

[Bulletpoint]

33 minutes ago, Drifter Man said:

2 hours ago, Bulletpoint said:

How far from the zero-in target point does the tank get an advantage? That is, what if the enemy tank pops up 200m to the right and 150m more distant?

TRPs seem to work in a 50m radius for artillery. I don't know whether this applies to non-artillery units as well, but I assume it does.

Yes, that is for TRPs. I was asking about this new (to me) method of using MGs to range in tanks..

[Drifter Man]

1 hour ago, Bulletpoint said:

Yes, that is for TRPs. I was asking about this new (to me) method of using MGs to range in tanks..

I see - in that case I don't know.

[Drifter Man]

On 1/17/2022 at 10:56 AM, John Kettler said:

Congratulations on what has to be the most comprehensive testing and analysis done by a regular CM player in the entire history of Combat Mission going clear back to the Beta Demo for CMBO!

Thanks - the credit should really go to my old backup laptop, which has managed to run CM practically 24/7 for months on end. Lenovo rules.

[Drifter Man]

On 1/17/2022 at 9:00 PM, Bulletpoint said:

This is also how I thought they worked. But recently I was told that TRPs improve accuracy for even units that have moved. I had a discussion with someone about it, maybe it was @Drifter Man?

I found our discussion of this around this post:

@IanL subsequently confirmed that this is how TRPs work due to a design limitation. They can be used even by units that have moved from their initial position.

[Simple Wine]

6 hours ago, Bulletpoint said:

How far from the zero-in target point does the tank get an advantage? That is, what if the enemy tank pops up 200m to the right and 150m more distant?

I don't know, but expect it to be better then without preliminary zeroing if the target is at considerable distance from the tank.

[herr_oberst]

On 1/23/2021 at 9:23 AM, Bulletpoint said:

Bushes don't give advantage, and in fact they give a disadvantage.

Hmm... bushes messing with the gunners view, given most gunners are lower than turret commanders?

[Drifter Man]

On 1/17/2022 at 4:02 AM, Bufo said:

How many iterations do you do?

Sorry, I forgot to answer this question.

10,000 trials for each spotting test - that means loading the test save 10,000 times, calculate the action phase, let run for up to 7 seconds (until the first spotting action) and collect data of which enemy tanks (in which position) were spotted.

1000 trials for the duels.

[Sgt.Squarehead]

17 hours ago, Drifter Man said:

Sorry, I forgot to answer this question.

10,000 trials for each spotting test - that means loading the test save 10,000 times, calculate the action phase, let run for up to 7 seconds (until the first spotting action) and collect data of which enemy tanks (in which position) were spotted.

1000 trials for the duels.

OMFG! 

[chuckdyke]

17 hours ago, Drifter Man said:

1000 trials for the duels.

That is dedication and you do all of us a tremendous service. 

[Drifter Man]

4 hours ago, chuckdyke said:

That is dedication and you do all of us a tremendous service. 

Not dedication, just automation. It's about 200 lines of code.

[Sgt.Squarehead]

6 hours ago, Drifter Man said:

Not dedication, just automation. It's about 200 lines of code.

I'd love to know how you did that.....Probably wouldn't understand a word of it if you told me, but all the same! 

[Bulletpoint]

@Drifter Man Are you planning to test the Puma's spotting, too? They are rarely used due of rarity, but are sometimes seen in scenarios.

Yesterday had a scenario where my four Pumas had lots of trouble spotting suppressed AA guns just apprx 400m away... so I was wondering how well those things actually spot.

Edited January 21, 2022 by Bulletpoint

[Drifter Man]

On 1/22/2022 at 3:44 AM, Bulletpoint said:

@Drifter Man Are you planning to test the Puma's spotting, too? They are rarely used due of rarity, but are sometimes seen in scenarios.

Yesterday had a scenario where my four Pumas had lots of trouble spotting suppressed AA guns just apprx 400m away... so I was wondering how well those things actually spot.

Yes - German heavy armored cars are on the list.

I used the PSW 234/2 in a QB recently, by the way. We had loose rarity and main gun limit to max. 50 mm, in CMBN. Interesting vehicles, worked okay for me.

[Drifter Man]

On 1/22/2022 at 1:00 AM, Sgt.Squarehead said:

I'd love to know how you did that.....Probably wouldn't understand a word of it if you told me, but all the same! 

The first part is to control the game. If you know programs like AutoClicker that let you specify a sequence of mouse clicks and repeat it... you can program something similar with python, which has tools that allow you to emulate keystrokes and mouse clicks (PyAutoGUI). So, I have a loop that clicks in various places on the screen, pressing the buttons I need in the right sequence and with the right time delays in between, like a human would.

The second part is to process what is happening on the screen. For this I save a screenshot (by having Fraps running in the background and by emulating an F10 press). Again, python has tools (PIL) to load an image file and, for example, look at individual pixel values. I know where to expect the tank unit icons that have been spotted. If there is a bunch of black pixels nearby, it means that there is a solid contact on an enemy unit in that position. No black pixels = it is either a tentative contact or nothing.

The screenshot taking part is quite inefficient - I would prefer reading the pixels right off the screen, without saving anything to the disk. But I haven't found any library that would do that with CM. They just return the main menu screen, same as when you use printscreen in the game.

I have no idea how the magic works inside those libraries I linked to. I just have some basic understanding of how to use them.

I hope some of those words make sense 

Edited January 23, 2022 by Drifter Man

[Sgt.Squarehead]

16 hours ago, Drifter Man said:

I hope some of those words make sense 

The words themselves are just fine! 

[Bulletpoint]

OK, I went back and replayed the little Puma scenario a couple of times. Two observations:

1) It seems to spot an AA gun, you need to spot the crewmembers, not the actual gun. So when suppressing the gun, it becomes nearly invisible to the Pumas. Not completely sure this is how it works, but this is how it seems like to me.

2) Pumas don't fire their main gun against AA guns. They fire one shot when they spot the target, but then just keep firing MG. This also seems to fit with the hypothesis that the Puma sees the target as infantry.

[Drifter Man]

On 1/24/2022 at 6:35 AM, Bulletpoint said:

OK, I went back and replayed the little Puma scenario a couple of times. Two observations:

1) It seems to spot an AA gun, you need to spot the crewmembers, not the actual gun. So when suppressing the gun, it becomes nearly invisible to the Pumas. Not completely sure this is how it works, but this is how it seems like to me.

2) Pumas don't fire their main gun against AA guns. They fire one shot when they spot the target, but then just keep firing MG. This also seems to fit with the hypothesis that the Puma sees the target as infantry.

I've never thought about it that way, but I think I've seen similar things happen - I doubt they are specific to AA guns vs Puma, it is more likely that they apply to all crewed weapons vs AFVs.

I agree that units see the crew and not the gun - when the crew take cover and vanish from sight, the unit is unable to spot the gun itself and loses a target to shoot at.

I also think that the AI tries to choose the weapon that "makes most sense" to it in a given situation. When the crew is up and vulnerable, it may be worth a HE. Once they take cover, the AFV will use MG to keep them suppressed, although we might prefer using more HE to blast the gun out of existence before it gets the chance to shoot back.

My tests won't help here. I will only tell you how well the Puma can spot compared to other AFVs.

[Bulletpoint]

14 minutes ago, Drifter Man said:

My tests won't help here. I will only tell you how well the Puma can spot compared to other AFVs.

Yes, I know.. didn't want to derail your thread.

When it comes to spotting other AFVs, I think the Puma spots quite normally. Haven't noticed anything special about it.

[Drifter Man]

On 1/25/2022 at 10:06 PM, Bulletpoint said:

Yes, I know.. didn't want to derail your thread.

When it comes to spotting other AFVs, I think the Puma spots quite normally. Haven't noticed anything special about it.

Nah, you can't derail this thread with just one bulletpoint 

I don't expect any surprises about their spotting, either. It is unfortunate that they don't come up more often in QBs - German armored cars in general - they are interesting vehicles. Too expensive. Stuarts can do more for less.