In this exhausting analysis we look at each and every graphical setting Elite Dangerous has to offer. We’ll look at how I performed these tests, the test system and the effects both visually and in terms of framerate on each and every setting available.
- Core i9 10900k @ 5.2Ghz on all cores
- RTX 2080Ti @ 2.1Ghz
- 32GB DDR4-3600 RAM
- 512GB Samsung 970 Pro NVMe
Artist: Miguel Johnson
Elite Dangerous has a number of graphical options that affect not only the visual quality but also how the game performs. These can be used to meet target framerates from 60 frames per second to 144 or even 240 to match the refresh rate of your monitor. And while some of the settings have basic descriptions most don’t and as such it can be very difficult to know what each setting not only affects visually but how it will impact performance. I’m commander Exigeous and Today we’re going to take an exhaustive look through each and every setting to understand their visual and framerate impacts.
First let’s look at how I went about these test given Elite has no built in benchmark mode. The machine I used is a Core i9 10900k, a RTX 2080Ti, 32GB of 3600 Mhz RAM and a Samsung 970 Pro NVMe drive. For all tests the goal was to demonstrate the difference in each possible setting in terms of visual appearance and framerate impact. Given the difficulties in obtaining all this data it should be used as a guide to help you decide what settings to chose as while we’ll be looking at a lot of examples we certainly won’t cover every object that is affected by every setting, just the highlights. For each test I did all I could to recreate the situation exactly, then average the results over multiple runs and in many cases across different environments. I ran over 500 individual tests, recording over 50GB of video to put all these results together. To obtain the framerate data I used Riva Tuner which is part of MSI Afterburner and I’ve included a link to the spreadsheet with the full details in the description below. Now that we understand the how and why of these tests let’s jump in and look at each option available to us.
Texture Quality offers us 3 settings, High Medium and Low with very little visual difference between them. These apply to the textures on vehicles, here we’ll look at the minor differences in the SRV with the differences in ships being equally minor. If we look closely at the fine detail on the back panel we see the detail of the triangles is much more defined in high mode with medium and low being virtually identical. We see similar minor differences looking at the front where we see this same depth detail in the small features around the canopy. The framerate difference between high and low is a modest 4.5% with medium roughly 2% less than low. Given these minor differences I’d recommend high for all but low end rigs where I’d recommend low given the extremly minor difference between it and medium.
We also see that same subtle differences when looking at the legs where high gives us much more defined detail with medium and low again looking basically the same with much less and softer detail.
Let’s now look at effects quality, or FX in settings which offers us High, medium low and off. These settings affect much of the lighting elements from the glow inside stations to lens flares and the way some weapons, like plasma accelerators and thrusters are rendered. Let’s first look at it’s effects on space, where in all settings but off we see a light glow around basically everything with high and medium adding lens flares which are turned off when running in low. As such for VR players I’d recommend low as these lens flares can feel rather blury in VR amplying god-ray issues.
FX Quality also affects several weapons in the game, here we’re looking at plasma accelerators where high medium and low look virtually identical with off removing much of the glow that’s added. Setting this to off can greatly improve visility when using these weapons, especially when using thermal conduit which is like firing small suns.
The glow around thrusters is also controlled by FX quality with high medium and low all adding the same glow around thrustres, especially during a boost. When set to off this added glow is removed not only allowing you to see the thrusters and, if selected your sub-targetting reticle. Reducing this potentially extremely bright light source can be very beneficial for specifically PvP combat.
Looking at the performance impact of FX quality we see next to no difference, with all settings other than off costing us the same roughly 9% framerate impact. As such if you enjoy all these effects run high and if you’d like to simply turn off lens flares go with low. Again for intense combat situations and or VR you may want to go with off.
Now let’s look at Bloom where we have 4 options, Ultra, High Medium and Off. This affects the glow around light sources such as the large flood lights inside stations we see here. The setting level affects the amount of bloom with off removing these effects. As there is very little visual difference between them you may consider using medium to save some performance and to keep things from becoming too bright, especially in VR where this too can amplify the perception of god rays.
The performance impact of bloom can be quite high for such a minor difference in visual fidelity. We see here that ultra costs us a 9% premium over no bloom, with high around 7% and medium just below 1%. As such medium would be my general recommendation for all but high end systems or if you prefer the brighter glow.
Let’s now talk about supersampling, that is where the game is rendered internally at a higher or lower resolution then scaled back to your native resolution. When using higher settings you can get a crisper image, negating the need for anti aliasing as the highly rendered then scaled down image will achieve basically the same effect. You can also render a lower resolution image that is then upscaled to native allowing you to run various settings at a higher level allowing you to reach a specific target framerate such as 60 or 90hz for a display or VR headset. When running higher than 1.0 the negative performance impact can be significant where running lower that 1.0 can improve framerate greatly at the cost of visual fidelity.
As noted the performace impact here both positve and negative can be extreme ranging from a whopping 140% for 2.0 above 1.0 down to a 15% savings when running at .5. For all but the highest end systems I recommend leaving this at 1 and for lower end systems you may consider using 0.875 if you aren’t hitting your target framerate and would like to run with various options at higher levels. VR players will see a setting for HMD Image Quality, this is essentially the same but rather than upscaling using Elite’s engine it will hook into your headsets API which generally results in higher quality. You can combine the two, running supersampling at 1.25 and HMD image quality a 0.875, negating each other but generating slightly higher image quality. This was popular in 2016 when VR headsets first launched but is generally no longer used.
Ambient occlusion is a shading and rendering technique that generates small shadows in parts of a scene based on ambient lighting. This is best seen inside stations as there is quite a bit of ambient lighting. In this scene if we look at the small shadow at the base of the wall inside the building we see that on high settings this is nice and crisp, generated in what appears to be native resolution. As we drop the quality to medium these areas are rendered at a lower resolution not only making them a bit fuzzier but also brighter as they are causing less shadow. At low resolution these are rendered even lower once again causing the scene to brighten. This effet can be seen in several areas in the scene, including the roofs and base of the railings.
In locations such as stations, where there are many light sources and as such many areas to be occluded the performance impact of high can be significant, it’s actually one of the most demanding of all settings, along with shadows. On average high ambient occlusion can cost us as much as 13% over off with medium the far better choice only costing us around 2% with very minor visual differences. Low only costs us 1% over off but given the minor difference between low and medium I’d generally recommend medium on all but low end systems. As even low adds quite a lot of visual fidelity I would generally recommend leaving ambient occlusion on in some capacity as the scene looks much better to my eyes than with it turned off.
Terrain Material quality has 4 settings, ultra high medium and low and it effects the generation of, well, terrain. Looking at the differences ultra terrain gives us the most overall detail, up close and at a distance. High reduces the detail at all distances with medium rendering close content the same as high but distant terrain at a lower quality. Finally low generates all terrain at a lower quality no matter the distance.
When we look at the performance impact ulta demands an 8% premium over low, where high costs us a close 6%. Medium only takes an additional 2% over low and given close terrain is rendered which such higher quality I would recommend this for all but low end systems. As planetary surfaces are some of the most demanding locations in Elite I would only recommend ultra on non-VR systems, for virtually all VR players medium is likely the best selection.
Terrain sampler quality, really planetary anisotropic filtering which is a technique similar to anti aliasing which smoothes jagged lines by reducing blur and preserving detail at extreme viewing angles. As this has been common setting in games for years it’s performance impact is realitively low. As we look at our four options, Ultra High Medium and Low we’ll see minor differences as textures are lightly blurred as we step down with low giving us the most blurred image.
While the impact of terrain sampling is rather low it isn’t free, ultra takes a roughly 6% impact over low, with high around 3% and medium clocking in right at 1%. For non-VR players I’d recommend ultra given the realtively low impact with VR players likely choosing High for it’s mix of quality and framerate, however VR players on GPUs lower than a 1080 may want to choose low here given the lack of optimizations on planetary surfaces.
NOTES: Sorry, just to be clear in terms of the terrain work slider differences the cpu is largely irrelevant, the slider slides between allocating a small amount of gpu time or a large amount of gpu time. The work the gpu does would be done 100 times slower on the cpu if we did use it, which we don’t
Next up we have terrain quality, which handles the rendering of planetary textures at distance. This effects surface elements that at outside your current area and while the impact of visual quality is rather apparent the effect on framerate isn’t too bad with ultra roughly 5.5% over low, high just over 3% and medium about 2.5%. Given the minor differences in framerate on most systems I’d recommend Ultra, but given the changes don’t affect your immediate surroundings if you’re looking for a few frames running in high is likely the best choice.
Depth of Field
Depth of Field seems to only effect the 3rd peson views, such as the free camera, specifically allowing and setting the detail level for depth when using the zoom and blur functions. While these settings are bound in controls the graphical depth of field setting is what affects there quality levels, or to simply disable them by setting to off. You can see there are fairly minor changes between high and medium in terms of quality with the framerate impact very significant. High settings here cost us a enormous 32.8% over off with medium nearly impercevable between off. As you’re generally not going to care about framerate in the external camera, especially if you’re just shooting still shots for most I’d recommend taking the perofmrnace hi t of high, however if you are shooting video and not getting 60 fps using medium will likely be your best choice.
Environment quality primarily affects the skybox when in space, wether you’re in normal space or supercruise. Seen here the rendering of objects such as the galactic core and nebula shows the differences between ultra high medium and low. Given these are generally rather simple textures anyway the performance difference between them is fairly minimal. For Ultra settings we see a roughly 7% increase of low, with High costing us about 6%. Medium is only about 3% over low meaning for virually all systems I’d recommend Ultra. Given there is basically no performance difference between ultra and high if you are looking for a few frames I’d likely suggest low as these aren’t elements you’re generally focusing on.
The Galaxy Map offers us 3 options, High Medium and Low. These effect the quality of the background starfield as well as nebula, very similar to the environment settings. While these do make significant changes to the galaxy map for most it won’t be too big of a concern given it doesn’t affect active gameplay. Looking at those numbers high costs us almost 16% over low with medium costing around 10%. Again as the galaxy map isn’t used for active gameplay for almost all systems I’d recommend high where as there is little visual difference between medium and low I’d recommend that selection for lower end systems that are having trouble hitting a target framerate.
Now let’s look at the rendering of a station, here we see similar results with FXAA again giving what to my eyes are the smoothest lines, however with some small loss of clarity. We can see that with both SMAA and FXAA we have a slight reduction in brightness due to the smoothing of lighter colored objects such as the grate at the rear of the pad. When we look at MLAAX4 and MLAAX2 we see slight differences over no anti aliasing with SMAA a nice balance of smoothness and clarity. This makes the choice a bit more difficult that before, however personally given the performance difference and that I find the smoothing of HUD elements more important overall given they are front and center I still personally choose FXAA.
Let’s start fairly simply, by looking at the performace difference in the various presets. We have four options for 2D, Ultra High Medium and Low and the same four for VR Ultra High Medium and Low. As there are many setting differences between them let’s focus on the performance impact of each. Starting with the 2D presets we see that Ultra costs us 10% over low, with high about 6% and medium around 3%. In a future video we’ll look at how you can customize each of these to set your own presets. Due to VR headsets locking a maximum framerate it isn’t really possible to test in VR, as such I tested the difference in those presets in 2D, so take these with a grain of salt. As we are looking at the difference in impact of each setting selected this should still give us some idea of the performace differences. Here we see Ultra’s impect is just over 7% compared to low, with high costing around 4% and medium just over 1%. As these performance differences are rather minor for VR players using SteamVR I’d highly suggest the fpsVR app to confirm your framerate in real-time.
Let’s start fairly simply, by looking at the performace difference in the various presets. We have four options for 2D, Ultra High Medium and Low and the same four for VR Ultra High Medium and Low. As there are many setting differences between them let’s focus on the performance impact of each. Starting with the 2D presets we see that Ultra costs us 45% over low, with high about 28% and medium around 8%. In a future video we’ll look at how you can customize each of these to set your own presets. Due to VR headsets locking a maximum framerate it isn’t really possible to test in VR, as such I tested the difference in those presets in 2D, so take these with a grain of salt. As we are looking at the difference in impact of each setting selected this should still give us some idea of the performace differences. Here we see Ultra’s impect is just over 32% compared to low, with high costing around 10% and medium around 8%. As these performance differences are rather significant for VR players using SteamVR I’d highly suggest the fpsVR app to confirm your framerate in real-time.
Again these should be considered general guidelines to help you understand the presets
Elite describes Terrain LOD blending as determining how smooth transitions between LOD, or levels of details are handled. We have 3 settings, ultra which blends patch vertex and texture data between levels, High which seems to only take effect when you’re near a surface and off which performs no blending. This was exceedinly difficult to test as there isn’t really a way to show an A to B difference visually. As such I performed 18 tests, 2 for each setting while approaching a planet and flying over a very large crater, and both the SRV challenge and tutorial from the training menu. After collecting and averaging all this data together the differences seem to be extremely minor with Ultra showing a roughly 3% framerate cost over off with high just over 1%. As such and given this will help reduce terrain popin, which can be rather obvious in Elite for almost all systems I’d recommend Ultra.
Terrain Work was another that was very difficult to test with Elite describing it as controlling the amount of GPU resources that are used to calculate the terrain. Just like terrain LOD blending I performed the same 18 tests and just like LOD Blending I saw essentially no difference no matter the slider position. This could be due to my rather powerful 2080Ti so I welcome your results in the comments below. As you see here while there are variations in the minimum and maximum framerate the average results are essentially identical. As such if you are seeing frame rate dips while terrain is being generated set this to it’s lowest level otherwise I’d recommend running this at maximum.
And finally we have the second most elusive effects to track down in Elite, Volumetric Lighting. This is generally a technique used to add lighting effects such as light rays shining through an open window or what’s generally called god rays. While there are several examples of this type of lighting in Elite I can find no locations where the various setting choices of Ultra High Medium and low have any percevable effect visually. Much of what you’d assume would be affected, such as light in icy rings, is actually controlled by either FX or shadows rather than volumetric effects. However the setting choices have a significant impact on performance causing me to wonder if this is due to a bug. I’ve tested locations such as Thargoid structures, icy rings, stations and guardian sites and in none of them can I see a visual difference. I’ve reached out to Frontier for clarification but sadly don’t expect a reply. Regardless these settings do affect framerate rather dramatically with ultra costing us the most of any other setting at 22%. What’s odd is high medium and low are virtually identical with only a 2% variation between them. So like material quality if you can find a demonstratable difference in them I welcome it in the comments below and will produce an updated video covering them. For now given the performance I’ve seen I have to recommend high here or really anything other than wasting frames with Ultra.
Material quality was another that was very difficult to work out, from all my testing the only thing I can find that is affected by this are the atmospheres of planets. I tested with a number of earth likes and gas giants showing me minor visual differences between Ultra High Medium and Low. Not only are the visual differences minor so are the performance differences with ultra costing us only 3% over low with high costing around 2.5% and low right around 1%. Given these minor differences and that the apperance of ultra is nicer for all but the lowest end of systems I’d recommend ultra here.
This was by far my most ambitious project thus far, taking well over 50 hours of research, filming, scripting and editing and over 500 individual tests so if you made it this far and found it valuable I hope you’ll consider subscribing. With the right settings Elite Dangerous can be both a visually beautiful game which performs very well on a wide range of hardware. I hope you enjoyed this deep dive as I look forward to hearing your thoughts in the comments below. Once again this has been commander Exigeous of edtutorials.com reminding you to fly dangerously and thanks for watching.