Bucic

The frame times turn into sawtooth even with LOD set to 0.1 and Tree Visibility Slider turned down almost to minimum. It seems that the bug is triggered MAINLY by the close proximity to trees. The only required factor is: the trees are out of the lowest LOD. That's it.

OK, I've just confirmed 100% similarity of my current vs pre-December updates configuration. That includes both in-DCS and in-nVcpl settings. To better illustrate the problem, compare the DCS performance metrics (notice the frame time sawtooth pattern!) between the scene where barely a 1 km radius is visible in the scene vs the other shots. The last screenshot I took deliberately at the exact moment the distance to the trees corresponds to start of the performance dip. https://1drv.ms/f/c/66cf2646234394fc/EkB24ygz9mBLqN6BJlpqAMoBauptRXW6rOmTCLgQXMurkg?e=CEWV8j

That's (almost) besides the point. I'm comparing the same hardware and DCS configuration to just ~2 updates back. There's clearly some choking happening universally with trees. I don' t have an issue with bringing the tree LOD down. I just wouldn't want a performance issue of the infamous Lock On flares caliber to slip under the radar and grow into a permanent performance issue component of low discoverability.

Not likely. 6500 or even ~7000 was nothing out of the ordinary on my system still at 100 FPS. Could you please clarify your question on GPU usage? My system always runs "GPU limited" with DCS. EDIT: The GPU utilization is ~90% and the temps rise rapidly.

A sensible studio adjusts techniques to available computing power. Not the other way around. ED's choice to cling to rendering of individual (and overlit beyond reason) trees even for forests is of their of their own doing. As far as I know SpeedTree is a model generation toolkit, not a rendering technique. Also, there's no reason to assume the performance drop is an outcome of a deliberate implementation. Same as with the broken MSAA.

I went from some ~May build straight to December.

Just a heads up - you may be misled into thinking a terrain mod (this or that) is causing a performance drop for you. You may actually be experiencing this https://forum.dcs.world/topic/366205-performance-trees-huge-performance-hit-especially-up-close/ Also a reminder: MSAA is currently broken.

Just a heads up - you may be misled into thinking a terrain mod (this or that) is causing a performance drop for you. You may actually be experiencing this https://forum.dcs.world/topic/366205-performance-trees-huge-performance-hit-especially-up-close/ Also a reminder: MSAA is currently broken.

TLDR: I finally discovered why my performance dipped so badly compared to just 2 months back. My system is 100+ FPS capable. Usually capped at 60 FPS. First I've noticed I can no longer fly 50+ FPS over Marianas. I just thought its looks has been ramped up and ignored it. Then I started to notice stuttering while flying low level. It turns out it's not just stuttering. As it is now DCS turned from 100+ FPS to almost unplayable, just like that. Especially that switching from DLAA to MSAA x4 didn't change the situation by even a bit. Just a reminder: MSAA is basically completely broken anyway in the current build. I included a modless track file but also recorded a video on a separate flight to show you what I'm seeing. I doubt track files contain performance metrics of original player's flight and nVidia recording has negligible impact on performance on my system. Specs: 1440p LG UltraGear 27GP850-B G-Sync Intel i3-12100F RTX 2070 8GB MSI, latest nVidia studio driver 32.00.0015.6636 (English) Driver Version: 32.0.15.6636 64 GB RAM Motherboard MSI PRO B660M-A WIFI DDR4 (MS-7D43) 2024-12-19 18:39:54.538 INFO APP (Main): options.graphics = { ['volumetricLights'] = 1; ['AA'] = 'DLAA'; ['forestDetailsFactor'] = 1; ['rainDroplets'] = true; ['LensEffects'] = 0; ['box_mouse_cursor'] = false; ['anisotropy'] = 4; ['water'] = 0; ['maxfps'] = 59; ['motionBlurAmount'] = 1; ['BlurFlatShadows'] = 0; ['outputGamma'] = 2.1; ['aspect'] = 1.7777777777778; ['lights'] = 2; ['LODmult'] = 1; ['MSAA'] = 2; ['messagesFontScale'] = 1.25; ['canopyReflections'] = 1; ['width'] = 2560; ['visibRange'] = 'High'; ['DLSS_PerfQuality'] = 1; ['useDeferredShading'] = 1; ['clutterMaxDistance'] = 0; ['textures'] = 2; ['cockpitGI'] = 1; ['terrainTextures'] = 'max'; ['height'] = 1440; ['multiMonitorSetup'] = '1camera'; ['shadows'] = 4; ['defaultFOV'] = 55; ['shadowTree'] = false; ['chimneySmokeDensity'] = 10; ['secondaryShadows'] = 0; ['SSS'] = 0; ['fullScreen'] = true; ['preloadRadius'] = 61200; ['scaleGui'] = 1; ['Scaling'] = 0.66; ['DOF'] = 0; ['clouds'] = 1; ['sceneryDetailsFactor'] = 1; ['Upscaling'] = 'OFF'; ['Sharpening'] = 0.7; ['motionBlur'] = 0; ['ColorGradingLUT'] = 0; ['SSLR'] = 0; ['effects'] = 3; ['SSAO'] = 0; ['maxFPS'] = 60; ['sync'] = true; ['heatBlr'] = 0; ['forestDistanceFactor'] = 1; ['flatTerrainShadows'] = 0; ['civTraffic'] = ''; ['ScreenshotExt'] = 'png'; }; Track and log (as usual, dxdiag dump only on demand via PM): https://1drv.ms/f/c/66cf2646234394fc/EkB24ygz9mBLqN6BJlpqAMoBauptRXW6rOmTCLgQXMurkg?e=CEWV8j

Any attempts of actually validating the AI output?

An even easier alternative to excluding kneeboard from reflections would be to make that page not that contrasting and ugly. I assume it's a bitmap. I haven't really dived into KB yet, does anyone have a clue on what could be that file path?

At least in the horizontal plane. If not possible, please at least review their FOV and coverage.

As I said, I've seen that reflection many times already on YouTube so it's not a singular case. https://youtu.be/uUsGB4alxO8?t=323

@Joe1978Please file a separate bug report. There are at least two already confirmed with lighting adjustment. See https://forum.dcs.world/topic/365211-remastered-rwr-brightness-knob/

Track and spec here, DXdiag dump - on demand. https://1drv.ms/f/c/66cf2646234394fc/EoUZvQnnm7VMufmpAk7CiwsB5P_v-1d1E9rcwDdbhk3JMg?e=Nk3DKK Remastered + Channel Map NVIDIA GeForce RTX 2070 with latest Studio Driver Win 10

Remastered. No mods apart from maybe Kill the bloom at the time of recording. I'll make sure to turn that off while recording a short track. Easy task. I've seen that reflection in some YouTube videos.

I'm not sure anymore whether this was the sound of the intake doors or was it a coincidence. In any cas I haven't heard it even once in Remastered.

It works but it's both cumbersome and unpredictable. Caging is a bit easier. Or at least faster

I switched to 100% 1:1 config for both axes and it's clear to me the "full ailerons AFTER commanding high G" is the trigger here.* Trying to break the wings using concurrent pitch+roll doesn't have nearly as strong effect as far as the risk of damage goes. I won't be exploring this any further without access to higher sampling rate telemetry data. *with prior pitch down being a strong catalyst, which is understandable

Middle mouse button (scroll) click to open full-resolution versions.

The second issue is an outright bug across the cockpit, including RWR which has already been reported.

Speaking of pinterest, here's a set of uBlock lines to... improve the pinterest experience in Google Search. Should be easy to also have it for DDG: ! Remove Pinterest from Google search results google.###search .g:has(a[href=".pinterest."]) google.*##a[href*=".pinterest."]:nth-ancestor(1) ! Remove Pinterest from duckduckgo.com search results duckduckgo.###search .g:has(a[href=".pinterest."]) duckduckgo.*##a[href*=".pinterest."]:nth-ancestor(1)

Me neither. Never heard of the concept.

Since the reflection can be static I'd suggest either excluding the kneeboard from the reflection or put a generic, low-contrast placeholder page there. That ugly first page is an issue on its own. It sticks out like a sore thumb