work in progress Improved FLIR improvement discussion

[Flappie]

@Taz1004 I tried but I cannot find a Caucasus area with these different ground textures. Can you please sare a short track?

[Taz1004]

8 hours ago, Flappie said:

@Taz1004 I tried but I cannot find a Caucasus area with these different ground textures. Can you please sare a short track?

Track is attached.  There are more extreme cases but I chose this spot to keep both trucks in close proximity to reduce outside factors.  And I didn't compare between different type of surface like grass vs tarmac in case it might be the temperature of the surface heating up the unit (I don't think this is simulated in DCS but still).  These are both on grass.  And I know these two parts of grass use different noise image.  And the difference becomes more apparent if I alter the terrain noise image.

And I used Hornet this time thinking maybe it's auto exposure of A10C-2's TGP messing with it.  So this one I tried on Hornet with ALG turned off.  And same result.  The difference will be much more noticeable if you take screenshots and compare.

Test_TerrainFLIR_1.trk

 

The question is... why is the terrain the units are sitting on affecting the IR signature of the unit.  Even when it's same type of surface.

Edited April 16, 2023 by Taz1004

[Taz1004]

9 hours ago, Flappie said:

@Taz1004 I tried but I cannot find a Caucasus area with these different ground textures. Can you please sare a short track?

By the way.  I think terrain noise should be turned off in FLIR.  At least the very fine TV static looking noise.  Noise is supposed to simulate variation in grass color etc.  Grass temperature doesn't change if it's slightly darker grass or lighter grass.  Most real FLIR doesn't show these type of noise at close up.  This is what I was initially trying to do.

 

Edited April 16, 2023 by Taz1004

[Flappie]

Thank you very much. I added your findings to an ongoing report.

[NineLine]

On 4/15/2023 at 2:08 AM, Temetre said:

Of course you dont want to ruin the game, we all want DCS to be the best it can be.

But if you replace an older feature, that was inferior but works, with a new one, that doesnt fully work because its incomplete, then that kinda feels like you broke that part of the game. 

It improved, but after the IR-update, I had so many cases where it was nearly impossible to make out targets via the pod using IR mode.

 

That is your opinion, but I do not agree that the current offering is inferior. There are some things still being tuned but I do not think its as serious as you suggest. 

Some of the examples you gave such as the level of cold vehicles against the background I do not think are correct observations. The terrain is meant to change with temperatures as well as vehicles. So the base temps of the ground might vary depending on the mission, same with 'cold' vehicles. 

Anyways, you need to include tracks of your issues so we can see all the temps and the states of the vehicles and which vehicles you are having issues with otherwise there is no way to diagnose what is going on. 

Thanks.

11 hours ago, Taz1004 said:

By the way.  I think terrain noise should be turned off in FLIR.  At least the very fine TV static looking noise.  Noise is supposed to simulate variation in grass color etc.  Grass temperature doesn't change if it's slightly darker grass or lighter grass.  Most real FLIR doesn't show these type of noise at close up.  This is what I was initially trying to do.

 

 

I have flagged this suggestion with the devs as well, I would be interested to see what it would do to the look, thanks. 

[Temetre]

vor 4 Stunden schrieb NineLine:

That is your opinion, but I do not agree that the current offering is inferior. There are some things still being tuned but I do not think its as serious as you suggest. 

I havent said the current offer is inferior, and it has gotten a lot better with the last updates. Ive said so, I gicredit for that.

But it took a long way to even fix massive bugs like the "cold at start" issue, which, together with other issues, made the FLIR experience extremely frustrating, in a way that had nothing to do with realism. So in at least a few aspect the big FLIR update made things worse, aka inferior. Id say thats self-evident.

vor 4 Stunden schrieb NineLine:

Some of the examples you gave such as the level of cold vehicles against the background I do not think are correct observations. The terrain is meant to change with temperatures as well as vehicles. So the base temps of the ground might vary depending on the mission, same with 'cold' vehicles. 

Anyways, you need to include tracks of your issues so we can see all the temps and the states of the vehicles and which vehicles you are having issues with otherwise there is no way to diagnose what is going on. 

Thanks.

But how can a cold metal vehicle have the same temperature as a grassy field? The surface of a tank is clearly completely different than... grass. Eg in the sun, a tank might get very hot, while at night, its probably colder than the ground. This difference in temperature has to count for something.

I might give a track later, but what Im talking about seems so obvious... like, were literally talking about spawning "cold at start" vehicles that dont move. They always look like that, almost invisible above grass. My track could literally be five seconds long.

Edited April 17, 2023 by Temetre

[Moonshine]

here are some tracks showing the "cold-at-start" option, one BTR on the tarmac, one in the grass next to it. compare those units with the static map units as they are placed per default

tested the following scenarios:

- Summer, 20°C daytime

- Summer, 20°C nighttime

- Winter, 0°C daytime

- Winter, 0°C nighttime

 

did not test spring and autumn, should not be needed though.

 

what i observed:

during winter, obviously there is snow on the ground (at least on Caucasus). this makes it easier to pick out even cold at start vehicles.
during summer, at times it is near impossible to find them as they really blend in with the exact same color as the background has unless you already know where they are (like in this scenario, the waypoint is directly on it, hence easy to spot)

HOWEVER: in all of those scenarios, the static units next to the BTRs i places are BY A LOT easier to find

 

winter_cold-at-start_night_0°C.trk summer_cold-at-start_night_20°C.trk winter_cold-at-start_day_20°C.trk winter_cold-at-start_day_0°C.trk

Edited April 17, 2023 by Moonshine

[NineLine]

22 hours ago, Taz1004 said:

Track is attached.  There are more extreme cases but I chose this spot to keep both trucks in close proximity to reduce outside factors.  And I didn't compare between different type of surface like grass vs tarmac in case it might be the temperature of the surface heating up the unit (I don't think this is simulated in DCS but still).  These are both on grass.  And I know these two parts of grass use different noise image.  And the difference becomes more apparent if I alter the terrain noise image.

And I used Hornet this time thinking maybe it's auto exposure of A10C-2's TGP messing with it.  So this one I tried on Hornet with ALG turned off.  And same result.  The difference will be much more noticeable if you take screenshots and compare.

Test_TerrainFLIR_1.trk 597.04 kB · 3 downloads

 

The question is... why is the terrain the units are sitting on affecting the IR signature of the unit.  Even when it's same type of surface.

 

I was able to reproduce this with your track, I will ask the team what is going on here and see if it's a bug or some reasonable science, I am assuming some sort of bug maybe related to lighting that FLIR should be ignoring. 

[Temetre]

vor 20 Stunden schrieb Moonshine:

here are some tracks showing the "cold-at-start" option, one BTR on the tarmac, one in the grass next to it. compare those units with the static map units as they are placed per default

tested the following scenarios:

- Summer, 20°C daytime

- Summer, 20°C nighttime

- Winter, 0°C daytime

- Winter, 0°C nighttime

 

did not test spring and autumn, should not be needed though.

 

what i observed:

during winter, obviously there is snow on the ground (at least on Caucasus). this makes it easier to pick out even cold at start vehicles.
during summer, at times it is near impossible to find them as they really blend in with the exact same color as the background has unless you already know where they are (like in this scenario, the waypoint is directly on it, hence easy to spot)

HOWEVER: in all of those scenarios, the static units next to the BTRs i places are BY A LOT easier to find

 

winter_cold-at-start_night_0°C.trk 305.37 kB · 0 Downloads summer_cold-at-start_night_20°C.trk 422.11 kB · 0 Downloads winter_cold-at-start_day_20°C.trk 336.04 kB · 0 Downloads winter_cold-at-start_day_0°C.trk 301.74 kB · 0 Downloads

 

Dang, I was just preparing to figure out how to make a track and do the report. But thats actually way better than what I wouldve delivered!

 

[Moonshine]

20 minutes ago, Temetre said:

Dang, I was just preparing to figure out how to make a track and do the report. But thats actually way better than what I wouldve delivered!

 

I used the scenario you described and made the 5s test. Then i remembered ED saying that seasons and temperature do have an effect as well, hence i added more tests with different season and time settings (to see if night/day alone changes something) in order to provide comparisons. This can be taken even further by conducting even more tests with a similar approach (altering the units this is tested with and more weather settings like sun vs rain or just clouds and even different maps. and from Taz' post above even the surface the units are placed on seem to have an effect, hence why i placed one on grass, one on tarmac.) but the two most extreme examples (summer/winter) show the most significant differences. Everything in between will show the same tendency with smaller amount of impact on the outcome

Edited April 18, 2023 by Moonshine

[Temetre]

vor 29 Minuten schrieb Moonshine:

I used the scenario you described and made the 5s test. The i remembered ED saying that seasons and temperature do have an effect as well, hence i added more tests with different season and time settings (to see if night/day alone changes something) in order to provide comparisons. This can be taken even further by conducting even more tests with a similar approach but the two most extreme examples (summer/winter) show the most significant differences. Everything in between will show the same tendency wirh smaller amount of impact on the outcome

What seems peculiar is that I would expect that summer would have tanks sticking out a lot, considering how much they heat up in sunlight.

And speaking about light: One factor that also seems to hurt FLIR-performance a lot is the lack of shadows, idk if its just a graphics setting thing. Ive read that in Desert Storm, enemy tanks where often discovered by their shadow in satellite pictures.

 

Honestly, Im just trying to make sense of IR-footage currently. In IRL footage, thermal images seems as clear or more clear than TV footage (regardless of resolution). In DCS, its the opposite, IR images are way more washed out than TV, and everything that isnt seperated by temperature difference is harder to see.

Like, in real life IR has less contrast, but the contrast existing is way more pronounced and seperated. DCS also has stuff like ground textures adding a lot of visual noise in IR, which I dont think is as visible in in real life.

Edited April 18, 2023 by Temetre

[BammBamm]

Want to see something funny?  Load up the Hornet with the SLAM-ER and the datapod.  Get it all set up, inst fuze, man launch, too, etc...  Then launch the missile.  watch it travel through the DDI.  Then at about 14.0 seconds before impact, notice the screen change.  Now, hit F6 (follow missile view) and see a high quality picture in that mode.  Switch back to cockpit F1 and it goes back to an all dark screen with a couple hot objects.  Keep switching back and forth.  Easily reproduceable.

See attached image.

I can't wait for the next retort that this is by design realistic and real pilots jump out of their cockpits onto the missile to get a better guidance picture just like Harry Potter on his Nimbus 3000.

[NineLine]

2 hours ago, BammBamm said:

Want to see something funny?  Load up the Hornet with the SLAM-ER and the datapod.  Get it all set up, inst fuze, man launch, too, etc...  Then launch the missile.  watch it travel through the DDI.  Then at about 14.0 seconds before impact, notice the screen change.  Now, hit F6 (follow missile view) and see a high quality picture in that mode.  Switch back to cockpit F1 and it goes back to an all dark screen with a couple hot objects.  Keep switching back and forth.  Easily reproduceable.

See attached image.

I can't wait for the next retort that this is by design realistic and real pilots jump out of their cockpits onto the missile to get a better guidance picture just like Harry Potter on his Nimbus 3000.

Could you supply your track, please? Thanks. 

[Taz1004]

Noticed another thing today.  When I quit while looking at FLIR, it momentarily blinks to what looks to be correct FLIR.  I put together a video below.

What's interesting to me is that nothing else changes.  Terrain, trees, cockpit... nothing changes.  Only FLIR changes during that 1 frame blink.

 

[Mausar]

Any news about improving FLIR..?

Truck Flir december -12 celsius.trk Truck Flir August +50 celsius.trk

[Jenson]

Hello, I encountered the same issue and would like to report it. Also attached with some screenshots comparing the TV vs IR mode. Hope this could be solved soon, otherwise it's impossible to see anything, especially when using SLAM/SLAMER. 

Target is an SA-5 S 200 site, set on red alert status. Summer Time, temperature 25℃

ATFLIR View

 

 

SLAMER View

 

[Mearcat]

On 7/9/2023 at 9:13 PM, Taz1004 said:

Noticed another thing today.  When I quit while looking at FLIR, it momentarily blinks to what looks to be correct FLIR.  I put together a video below.

What's interesting to me is that nothing else changes.  Terrain, trees, cockpit... nothing changes.  Only FLIR changes during that 1 frame blink.

 

That's an interesting find, so it looks like there is a graphics item that is being rendered that shuts down first just before the sim exits to the menu. If that could be ID'd then a temporary work around would be to disable that graphic setting maybe.

Also another thing is a lot of these FLIR videos everyone has linked shows the terrain being a lot darker and more uniform (flat) in colouration. It could be the operators gain settings but would another temporary fix be to change the terrain maps IR textures to 30-50% strength? That would reduce the background noise to make vehicles more visible until the units have had their proper fixed IR textures done. Obviously the % change would need experimenting with to get the best balance, just need enough to reduce the background so a vehicle is distinguishable. One of the posts above had a similar effect with the "noise" texture map on one of the terrains.

But working on the terrain textures would be the quickest and simplest method giving the greatest impact. There are maybe 6-8 official terrain maps (guessing) compared to thousands of object files.

Just an idea.

Edited November 29, 2023 by Mearcat