How to clean up your ARKTIKA.1 graphics and get supersampling working

[MadMax1998]

Hello Arktika players! There has been a bit of discussion on the web about Arktika's blurry graphics, with people attempting to force supersampling (using the OculusTrayTool) but having no luck. When I played the game for the first time, I too noticed how blurry and "out of focus" the environment was, especially the farther away an object is. So I dug into the game's configuration file, which you can find at

C:\Users\<username>\Documents\4A Games\Arktika.1

where <username> is to be replaced by your own Windows account name. The file is called user.cfg and can be opened with any raw text editor such as Notepad. If you're using Notepad++ or another editor which shows line numbers, you can follow along with mine; makes it easier to find what you need to change in the file.

Let's see how we can clean up our image in Arktika.1 !

WHY does Arktika.1 look blurry?

As pointed out by some, the game uses a dynamic form of undersampling/supersampling based on the quality preset you chose and the speed the game is running at. Example: if you go with the "high" quality preset, the maximum supersampling factor will be set at 1.4, while the lowest the game is allowed to go will be 0.875 -- effectively giving you a pretty clear picture if your hardware allows it but dropping to below native resolution if performance is struggling.
This means that the higher you go with the quality, the sharper your image will look. But obviously that also increases the detail of the game, most notably lightmap resolution and ambient occlusion, so depending on your hardware you might want to go with "medium" or "low" presets and still want a clearer picture -- especially "low" quality is pretty hard to look at because it renders only at native resolution, dropping down to a sampling factor of 0.5 (!) when performance decreases.

There is more to the story, though. On top of the dynamic resolution, Arktika.1 also employs both FXAA (fast approximate anti-aliasing) and TAA (temporal anti-aliasing). Judging from my tests I'd say that FXAA does nothing to reduce pixels, but it *does* create a slight "fog" on top of everything. TAA however is the *real* problem here and the sole reason why things in the distance seem to be "out of focus" and blurry. TAA is always engaged no matter the quality and smooths out the pixels you might otherwise see. Now, TAA can be pretty awesome; in fact Unreal Engine 4 has very solid TAA that doesn't affect image clarity very much. But 4A Games' implementation of it is not very good... it does significantly reduce pixels and make the image look smooth, but also makes objects even 1 foot away look blurry. It kind of "swallows up" the detail of things. Turning off TAA is the first step to a much clearer image, as if someone gave you a pair of glasses when you're nearsighted!
Now I must point out that the TAA looks a lot better the higher your supersampling is. It's going to be horribly blurry on "low" settings because of low resolution, but will be tolerable on "extreme" because of the high supersampling. In general, since supersampling never gets rid of all pixels, TAA on top of it is a good idea, but only if your resolution is already pretty high before TAA is applied. And it will never be a perfectly clear image still.

WHAT do I need to do?

Open the user.cfg file with a text editor as mentioned above, then look for these variables. The numbers in front are line numbers; use a text editor with line numbers to easily follow along.

Overview of settings

110 r_adapt_q 1 << controls whether dynamic supersampling is active

121 r_aq_max_ss 1.39999998 << maximum allowed SS factor, medium quality
122 r_aq_max_ss_ex 1.39999998 << maximum allowed SS factor, extreme quality
123 r_aq_max_ss_hi 1.39999998 << maximum allowed SS factor, high quality
124 r_aq_max_ss_lo 1. << maximum allowed SS factor, low quality
125 r_aq_min_ss 0.75 << minimum allowed SS factor, medium quality
126 r_aq_min_ss_ex 1. << minimum allowed SS factor, extreme quality
127 r_aq_min_ss_hi 0.875 << minimum allowed SS factor, high quality
128 r_aq_min_ss_lo 0.5 << minimum allowed SS factor, low quality

180 r_fxaa 1 << FXAA setting, medium quality
181 r_fxaa_ex 1 << FXAA setting, extreme quality
182 r_fxaa_hi 1 << FXAA setting, high quality
183 r_fxaa_lo 1 << FXAA setting, low quality

262 r_quality_level 2 << quality preset to use, from 0 (low) to 3 (extreme)

302 r_taa_enabled 1 << TAA setting
303 r_taa_jscale 0.75 << TAA scaling (this does not seem to affect anything)

319 r_vr_start_recenter 0 << bonus: whether the game should recenter around you when booting

FXAA and TAA

Alright, so before we tackle supersampling, let's remove that blurry mess that is TAA. Find these lines:

180 r_fxaa 1
181 r_fxaa_ex 1
182 r_fxaa_hi 1
183 r_fxaa_lo 1

and:

302 r_taa_enabled 1 << TAA setting

Now set all of them to 0 instead of 1. This will eliminate the useless (?) FXAA and especially turn off TAA completely. Now go back into the game and admire the clarity! Of course you will also see pixels now, which we will fix with...

Supersampling/resolution

Now you could start by turning off dynamic supersampling altogether, like so:

110 r_adapt_q 0

However, that means the game will render at native resolution constantly and won't allow you to supersample at all - not even with OTT. If you're happy with that, great -- if you'd like to use some SS, leave the above setting on, find the quality preset you want to use and change the min/max variables to your desired SS factor, like so ("high" preset):

123 r_aq_max_ss_hi 1.2
127 r_aq_min_ss_hi 1.2

By setting both to the same factor, we force the game to run at the same resolution constantly. 1.2 seems to be a good sweet spot between image clarity and performance for me, running a GTX 1070. I use it in most other games and it seems to be a lot clearer than no SS (factor 1.0), where 1.4 is only a little bit better (smaller pixels) but has a big performance impact.
You are free to change the other quality presets as well and then switch presets in the game to see the effect immediately. I did this for mine:

121 r_aq_max_ss 1.20000005
122 r_aq_max_ss_ex 1.39999998
123 r_aq_max_ss_hi 1.29999995
124 r_aq_max_ss_lo 1.
125 r_aq_min_ss 1.20000005
126 r_aq_min_ss_ex 1.39999998
127 r_aq_min_ss_hi 1.29999995
128 r_aq_min_ss_lo 1.

Notice that the game rounds down/up on my values after you run it once; this is okay. Effectively, these settings mean:

"low" preset: SS factor of 1.0 (no SS)
"medium" preset: SS factor of 1.2
"high" preset: SS factor of 1.3
"extreme" preset: SS factor of 1.4

If you set the "min" and "max" variables for each preset to different values, you can define the maximum and minimum allowed resolutions as you desire. So you might want to go and do something crazy like 2.0 supersampling on "max" but, as a failsafe, keep 1.2 as "min" so the game can still perform well. Experiment!

More tweaks

Once you've found your optimal supersampling setting, you might want to try and turn TAA back on to eliminate the pixels. However, as I said above, this really only looks good enough on high resolutions -- think 1.5 and higher SS factors. Since TAA is essentially a "free" method of anti-aliasing, it won't hurt performance, though. So if you can indeed run that high of an SS factor, then adding TAA on top may just give you the best picture. If clarity is your chief concern, don't use it (I don't).

There are a lot of other variables in the user.cfg file and you can see by the "_ex", "_hi" and "_lo" suffixes which variables affect which quality preset. For example, you could go and turn ambient occlusion ON for "low" quality, like so:

155 r_esm_ao_lo 1

Or you could turn OFF shadows, like so:

268 r_shadows_lo 0

One thing we apparently cannot fix here is the world scale of the game. It's up to the devs to fix it for us.

Performance

Overall I'm a bit concerned with Arktika's performance on my system. I do get the occasional microstutter (one dropped frame visible in the Oculus performance HUD) in Oculus Home and games such as Robo Recall, and I've narrowed it down to just a software issue. I optimized the crap out of my system and I use my Windows 10 install (pre-Creators Update) for gaming only. In general I'm pretty happy with the performance of my VR system but Arktika.1 really doesn't run all that well. Apparently I'm not the only one who feels this way as I've seen comments under reviews saying the same. It's weird that the reviews themselves never mentioned performance issues!

Anyway, on "medium" settings with supersampling at 1.2 the game runs at 90 FPS *most* of the time but it will drop the odd frame occasionally. By that I mean once or twice every five seconds or so. Even when I face a flat wall and the performance HUD shows the GPU time in the greenish zone, there can be sudden, momentary spikes into the red. Just like that! The game really needs some optimization if you ask me.
Firefights in bigger areas are the most demanding and can bring my FPS down into the 70s, but ASW never comes on. My GPU load rarely exceeds 60% in GPU-Z; for that I'd have to increase the SS factor (which I might). It seems like the game is actually not *that* demanding on CPU/GPU (I'm running a modest i5-2500K at 4.5 GHz and it rarely hits 70%), so if the engine were optimized better for VR then we might actually see pretty smooth performance even on more modest systems.

Conclusion

Hey, if this helped you get better image quality out of the game, please share this post so others can see it! I'm not on Reddit and there doesn't seem to be an Arktika.1 forum or place of discussion, so I hope people will find this post here. Now back to shootin' raiders everyone!

[JJJake]

Arktika.1 is a really GREAT game in so many ways!   BUT I too have been TERRIBLY frustrated by all the BLUR ... even after SuperSampling at 1.8.  THANK YOU  MadMax for supplying such an exhaustive method for correcting a problem that should NOT have been there in the first place!  

Beyond just Arktika, the "resolution/blur thing" may not be killing VR/Oculus, but it is severely impeding its broad market acceptance.  Everybody who I have treated to the Oculus/VR experience is "wowed".  BUT they all ask, "is it supposed to be this blurry?!!"   Well, in some games, Yes.  In others, No.   Unfortunately, devs have not standardized a way to make VR a simple "plug 'n play" experience that maximizes image quality.  There are many millions "out there" who want to buy into and fully enjoy VR.  BUT all too often you have to be something of a "tech geek/wizard"  to extract/squeeze the best performance from games and/or the system. Arktika seems to be a perfect case in point.

I've been a Mac guy for over 20 years.  I got my first PC Windows (recently upgraded with a 1080TI video card) machine with my Rift.   A couple decades' experience with those wonderfully intuitive Macs did not prepare me at all for all the tech hoops one has to go thru when wrestling with "the control options" a newbie is burdened with PCs/Windows ... and the Rift!  Following the instructions above provided by MadMax may be a cake walk for experienced PC-ers.   And I MIGHT be able to successfully implement those changes.   However, every time I start mucking around with such settings I feel like I'm stumbling thru a minefield.  We'll see.

The thing is --- why oh why do we have to search and search for "3rd party HELP" with such games?!!  THANKS again Max.  But shouldn't 4A Games be providing such info right at their website ... or something?!!

BTW, examples of devs who got it deliciously right IMO are the guys who did Batman and especially Serious Sam --- those games look really FANTASTIC super sampled with the simple, onboard graphics options! 

[Comic_Book_Guy]

Until they fix the wonky scaling and giant hands, all this is rather moot for me.

[MadMax1998]

JJJake said:

Following the instructions above provided by MadMax may be a cake walk for experienced PC-ers.   And I MIGHT be able to successfully implement those changes.   However, every time I start mucking around with such settings I feel like I'm stumbling thru a minefield.  We'll see.
[...]
The thing is --- why oh why do we have to search and search for "3rd party HELP" with such games?!!  THANKS again Max.  But shouldn't 4A Games be providing such info right at their website ... or something?!!

Jake, it's rather easy to do; you basically just edit a text file. It's no different from how you'd do it on a Mac. Make a backup of that file before you mess around with it, then you can easily restore it. (You can also delete it and the game will regenerate it.)
I *would* provide a file that players can drop into their Arktika folder but my settings are for my rig and I want you to experiment and find your own settings that match your monster GPU!

As for your other sentence -- 4A shouldn't put this info on their website, instead they should make an in-game settings menu where we can tune these things *without* editing config files!

[huliqan]

r_dx11_tess 0
is This tessellation?how much can be set?

[MowTin]

JJJake said:

Beyond just Arktika, the "resolution/blur thing" may not be killing VR/Oculus, but it is severely impeding its broad market acceptance.  Everybody who I have treated to the Oculus/VR experience is "wowed".  BUT they all ask, "is it supposed to be this blurry?!!"   Well, in some games, Yes.  In others, No.   Unfortunately, devs have not standardized a way to make VR a simple "plug 'n play" experience that maximizes image quality.  There are many millions "out there" who want to buy into and fully enjoy VR.  BUT all too often you have to be something of a "tech geek/wizard"  to extract/squeeze the best performance from games and/or the system. Arktika seems to be a perfect case in point.

I agree that the low res panels are the biggest issue. At the high end, there is enough power for higher resolution panels. That's essentially what happens when we supersample. We're rendering at a higher resolution and then downscaling. 

[ebgb08]

Hello! I don't seem to have the user.cfg file. Did they perhaps change it since you guys found this fix?

[MadMax1998]

UPDATE: newest version of Arktika.1 works differently! Supersampling is now different from the values mentioned in the first post! Please read the first post to learn the method of how to edit the user.cfg file, but ignore the supersampling values; instead, use these!
I would edit the first post but this is no longer possible.

No line numbers this time, but these changed anyway with the last patch, and may change again with future patches.
The new user.cfg is not like the old one; if you still have the old file from before the February 2018 patch, delete it and run the game to recreate it!

Here's how to set supersampling factors in the new version of Arktika.1:

ALL quality presets at 1.0 SS (no SS):

r_aq_max_ss 1.4
r_aq_max_ss_ex 1.
r_aq_max_ss_hi 1.
r_aq_max_ss_lo 1.
r_aq_min_ss 1.4
r_aq_min_ss_ex 1.
r_aq_min_ss_hi 1.
r_aq_min_ss_lo 1.

Let's use this as our baseline. What is happening here is that, by contrast to the straightforward way from my original post, the supersampling factors are no longer independent from each other if you want them to be static. On top of that, they have been scaled internally, which makes it even more complicated to figure out what numbers you need to put where.

The new rules for supersampling are:

- all values are to be multiplied by 0.69 << this scale applies to all SS numbers you write into the file
- if you use static SS (min/max same value), the *highest* value in the whole list will be used for *all* quality levels
- if you use dynamic SS (min/max different values) on a quality level, its values will be used exclusively

I have no idea what 4A Games have done with the SS code, but things are unnecessarily complicated now. Let me be clearer by some examples:

ALL quality presets at 0.69 SS (as values are multiplied by 0.69):

r_aq_max_ss 1.
r_aq_max_ss_ex 1.
r_aq_max_ss_hi 1.
r_aq_max_ss_lo 1.
r_aq_min_ss 1.
r_aq_min_ss_ex 1.
r_aq_min_ss_hi 1.
r_aq_min_ss_lo 1.

ALL quality presets at 1.25 SS:

r_aq_max_ss 1.8
r_aq_max_ss_ex 1.
r_aq_max_ss_hi 1.
r_aq_max_ss_lo 1.
r_aq_min_ss 1.8
r_aq_min_ss_ex 1.
r_aq_min_ss_hi 1.
r_aq_min_ss_lo 1.

ALL quality presets at 1.5 SS:

r_aq_max_ss 1.
r_aq_max_ss_ex 1.
r_aq_max_ss_hi 1.
r_aq_max_ss_lo 2.2
r_aq_min_ss 1.
r_aq_min_ss_ex 1.
r_aq_min_ss_hi 1.
r_aq_min_ss_lo 2.2

As you can see here, it does not matter *where* you put the number for 1.5 SS (2.2) as long as both min and max values are the same on the same quality preset.

THREE quality presets at 1.5 static SS and ONE at dynamic SS (1.0-1.25), by its own values:

r_aq_max_ss 1.
r_aq_max_ss_ex 1.
r_aq_max_ss_hi 1.8
r_aq_max_ss_lo 2.2
r_aq_min_ss 1.
r_aq_min_ss_ex 1.
r_aq_min_ss_hi 1.4
r_aq_min_ss_lo 2.2

As you can see, the "high" preset was given two different values, making it scale dynamically. This will *not* affect the other quality levels or be affected by them. The other three levels still go by the highest value of 2.2 and are static.

Again, the *highest* number in the list defines the SS for all presets *unless* a preset has its own dynamic scale.

I hope this clears things up and you can reset your SS levels as you wish. Of course it is no longer possible to have four different static SS levels for the quality presets unless all of them are dynamic. 😕

I'd like to link to the wonderful Arktika.1 scale fix mod made by idk31 since this simple mod has transformed the game for me; it's just fantastic! Finally we can get proper scale in this game! I highly recommend everyone check it out to perfect your game experience.

Here are some more settings from the config file I've found, FYI.

r_af_level 1 << this *seems* to control anisotropic filtering and I assume "1" is 4x, "2" would be 8x and "3" 16x

r_fullscreen on << set this to "off" to run the game's monitor mirror in a borderless window!

r_vr_mirror_scale 0.2 << this seems to be the zoom factor of the monitor mirror compared to the VR render

r_vr_mirror_hor 1366 << horizontal monitor mirror resolution

r_vr_mirror_vert 800 << vertical monitor mirror resolution

[nrosko]

This is ridiculous, why they make this so difficult. 

[Fazz]

nrosko said:

This is ridiculous, why they make this so difficult. 

Good point!