MARVEL Tōkon: Fighting Souls Best PC Settings

Start at your display’s native output resolution, set Vertical Sync to Off for the first latency-and-tearing comparison, turn Motion Blur Off for clarity, and leave the native 60 FPS battle lock untouched. Run Auto Graphics Quality Settings, enable Launch Shader Warmup, then use Execute Shader Warmup after the August 10 patch. Stress-test the same assist, super, and stage-transition route. If sustained GPU load misses 16.67 ms, lower verified quality rows one step at a time or test an available scaling option. When misses remain at lower render load, investigate CPU/system load, first-use shaders, transitions, bugs, or rollback separately. Your library may show Marvel Tokon without the accent; these MARVEL Tōkon best settings use the same method, not a guarantee.

Build status at the August 20 cutoff: the August 10 PC patch is deployed; the follow-up announced August 19 is pending and must not be described as live.

Best MARVEL Tōkon PC Settings for Stable 60 FPS

On Graphics Settings, select the intended Output Display and set Display Mode to a working option; observed Borderless Window is not proven fastest. Set Brightness Settings for visibility, keep Resolution native initially, and compare Vertical Sync Off with the intended display path.

For Anti-aliasing Type, begin with a native-quality option shown on your GPU. DLAA, DLSS Balanced, TSR, and Off were retained; FXAA and FSR have weaker corroboration, and availability varies. No in-game Frame Generation toggle was observed. Choose the AA/upscaling path before judging Sharpness Quality or Scaling Resolution; captured values of 0 and 100 are not universal recommendations.

Run Auto Graphics Quality Settings once and inspect its result. Near minimum spec, start at 1080p/native and lower one verified tier only if GPU-bound. A balanced system should retain native output and the auto profile. Keep a higher profile only when five warm stress repeats hold near-even 16.67 ms pacing. Use Custom Graphics Quality Settings for controlled A/B tests; Revert to Defaults restores the baseline.

Why the 60 FPS Battle Lock Changes the Goal

The official Steam page says “60fps (locked in battle).” Your target is stable pacing at 60 FPS—16.6667 ms per unique battle frame—not an uncapped counter. Preserve the native lock and do not stack it with a driver cap and RTSS. A 57 FPS or generic 95%-of-refresh cap deliberately undershoots this fighter’s target.

At 120 or 240 Hz, each 60 FPS game frame divides evenly across refreshes. At 144 or 165 Hz, it does not. Variable refresh rate (VRR) may improve cadence when its range includes 60, but compare tearing and latency with Vertical Sync intentionally. Higher panel refresh does not raise the 60 FPS battle simulation ceiling.

A 60 average can hide repeated 33.33 ms intervals. When you choose a realistic FPS target, judge displayed cadence, a defined 1% low, and the frametime distribution—not average alone. Measure menus, lobbies, cinematics, Episode Mode, and transitions separately if their behavior differs.

Run Shader Warmup After the August 10 Patch

The deployed August 10 update reduced named startup CPU and audio overhead, stopped repeated audio-input checks linked to sudden drops, addressed certain-resolution frame-rate trouble, improved automatic profile detection, and added official warmup guidance.

1. Open Options → Graphics Settings.
2. Set Launch Shader Warmup to On.
3. Select Execute Shader Warmup and let it finish.
4. Run one unscored exposure pass through the exact characters, effects, super, and transition in your test.
5. Keep that pass separate, then collect warm repeats from the same state.

Idling in Training does not prove every relevant effect or transition asset is warm. Warmup can reduce first-exposure compilation hitches; it cannot fix sustained GPU/CPU pressure, VRAM streaming, rollback, transitions, or unrelated bugs. Re-running it after a driver or major graphics change is a cautious test, not a guarantee. The August 19 follow-up remained announced but not deployed at the cutoff.

What to Lower First When the GPU Is the Limit

Prove sustained GPU pressure in the same offline effects route before cutting quality. First compare one step lower on Post-processing Quality, Global Illumination Quality, Reflection Quality, or Shadow Quality. Test Effect Quality cautiously in the same assist-and-super sequence so hit sparks, projectiles, silhouettes, and competitive telegraphs stay readable. Keep Anti-aliasing Quality at the auto/default tier until the AA method and scale are fixed, then A/B one step.

Leave Texture Quality at the auto-selected tier while repeat tests show memory headroom. Lower it one step only for reproducible VRAM or streaming pressure; no universal 6, 8, or 16 GB cutoff is supported. None of these rows has a retained isolated FPS saving or proven cost ranking.

If the GPU still misses 16.67 ms, compare an available Anti-aliasing Type upscaling mode or reduce Scaling Resolution where exposed. Hold output resolution, route, and other settings constant. If lower render load does not change the same spike, stop lowering graphics. That order is safer when you need to fix FPS drops.

GPU Load, CPU Spikes, Rollback, or a Game Bug?

Do not call every symptom “lag.” Use seven lanes:

1. Sustained GPU/render load: the same effect-heavy phase misses repeatedly, GPU work stays high, and reducing resolution, scaling, or one verified quality row helps.
2. Sustained CPU/system/game-thread load: misses persist after render-load reductions while CPU, game-thread, audio, or device activity rises. The August 10 fixes did not prove every system issue solved.
3. Shader/first-exposure hitches: a character, effect, super, stage, or pipeline spikes once and improves on a warm repeat. Keep this separate when diagnosing PC stuttering in games.
4. Game-lock/simulation/presentation misses: the PC misses 16.67 ms, causing judder or stalled/duplicated presentation. Do not invent an exact slowdown formula or raise battle simulation to 120/144 FPS.
5. Rollback/network corrections: online-only teleports, snaps, rollback-counter spikes, or connection-dependent degradation. Compare offline first; record region, ping, loss, jitter, and cross-play where exposed. Work to reduce input delay only after separating local response from network correction.
6. Transitions/intros/camera cuts: isolate intros, supers, stage changes, KO/round transitions, and loading boundaries. A transition spike does not prove the whole fight is unstable.
7. Game/data/stability bugs: settings, saves, replays, ranks, achievements, UI, and crashes are not graphics optimizations. The August 19 announcement named future fixes, but they were pending at cutoff.

The selector guides evidence. Local misses affect online play; frametimes and ping are separate.

Official PC Requirements and Low-End Limits

Official tiers omit targets. “Low-end” means near minimum; lower settings only after repeatable GPU or memory evidence.

Run a Repeatable 60 FPS Match Test

No built-in benchmark was documented in the retained current sources, so use a controlled manual route.

Use a fixed Training script: movement → chain → assist → tag → special → super → reset. Repeat overlapping assists/projectiles → rapid tag → super/camera cut → stage transition → post-transition assist/super. Alternate A/B order, change one item, and collect five warm repetitions. Run a matched local round through start, exchanges, super, KO, and post-KO.

Log build/date, hardware, driver, OS, resolution, settings, AA/upscaler/scale, refresh/VRR/V-Sync, route, capture tool/version, exclusions, GPU/CPU/VRAM/RAM context, and online ping/loss separately. Capture displayed/presented FPS, one defined 1% low, median/P95/P99/P99.9 frametime, and counts above 20, 25, 33.33, and 50 ms. Never pool first-exposure and warm runs.

Finished the match test? Use Hone to reduce avoidable Windows background load, then repeat the same route. Results vary.

MARVEL Tōkon PC Settings FAQ