Watch a car turn a corner in GTA 6 and a 60 FPS clip can show its tail lights moving in smaller steps. Drive the same corner through a capture preview and the steering can feel late. Both observations make sense. The console renders 30 frames per second, and a PC running Veyra can fill the gaps after those pictures leave the console.
Where the extra pictures come from
Picture two consecutive console frames. In the first, the car is beside a lamp post. In the second, it has passed it. At a steady 30 FPS those views are about 33.3 ms apart. External interpolation uses the pictures to estimate where the car would appear between them. The controller still sends your steering input to the console. The PC has no way to ask the game to simulate an extra turn of the wheel.
| Display path | New console pictures per second | Ideal spacing between distinct displayed pictures |
|---|---|---|
| Direct console output | 30 | 33.3 ms |
| PC preview with 2× interpolation | 30 | 16.7 ms (60 FPS) |
| PC preview with 4× interpolation | 30 | 8.3 ms (120 FPS) |
Those intervals assume even delivery and a display fast enough to show every image. They describe motion on the screen. Input-to-display latency is the time from moving the control to seeing its effect. An FPS counter alone tells you neither that delay nor how often the engine samples input or updates physics.
GTA 6 FPS Explained: Why a Faster GPU May Hit a CPU Limit
Check for repeated frames before turning up the multiplier
Step through a recording of a steady camera pan with generation off. A 60 FPS capture or Remote Play stream carrying 30 distinct pictures can show A, A, B, B. The first pair gives the interpolator no movement to work with. The next pair spans the whole jump. In Veyra, if you confirm that pattern on a 60 FPS capture, enable Remove duplicate frames and choose 60→30 before selecting 2×. Check a moving scene again before trying 4×.
Follow a lamp post as it crosses the edge of the car, then watch the minimap during the same pan. A capture app sees the final picture, with the interface already drawn over the street. Look for a bent edge, a doubled outline or text that briefly smears as the background moves. Judge these at playing speed as well as frame by frame.
Measure steering delay on the display you play on
Use the same save, controller, monitor, refresh rate and display game mode for each pass. Choose a refresh rate the whole setup supports and use a 120 Hz or faster screen to judge 4× motion. Keep the controller connected to the console. Use a repeatable button action with a clear first visible response for timing, then use a slow pan for the separate motion check.
- Start with the console connected directly to the monitor. Film the controller and screen together in slow motion. Count camera frames from the same point in the button press to the first visible response, then multiply by 1,000 divided by the camera frame rate to estimate milliseconds. Repeat several times and record the median and range.
- Switch that monitor to the PC capture preview with generation and enhancement off. Repeat the presses. The difference from direct output estimates the cost of the whole preview path, including capture, buffering and display scheduling. It cannot isolate the capture card on its own.
- Test 2× and then 4× with everything else fixed. Record the delay alongside how evenly the pan moves and whether the minimap stays clear. A queue setting or neural enhancement pass deserves its own run. Change one at a time, then repeat the button test.
- Read small differences against the spread of repeated presses and the camera's time resolution. If steering feels late through the preview, try playing on the direct output while recording the processed view separately.
What to ask of a DLSS or FSR frame-rate claim
Ask for the rendered FPS with generation off, the rendered and displayed FPS with it on, and the latency measured in each mode. Keep upscaling quality and latency settings fixed where possible. Super Resolution can reduce the work needed to render each frame. Frame generation adds pictures between renders. An in-game implementation can use engine motion data that an external capture lacks, so two modes labelled “2×” need their own comparison.
DLSS 4.5 vs DLSS 5: Test Visual Quality, FPS and Latency
FAQ
A processed clip can contain 30 console renders and 30 intermediate pictures each second. Check the direct console output or a capture with generation disabled to establish the source rate.
The multiplier sets how many pictures the PC aims to display from the same console frames. Capture, waiting for source frames and presentation all take time. Compare button-to-screen measurements on the two previews to find the difference on your setup.
Passthrough carries the console picture. Veyra generates frames on the PC and sends the processed view to the PC display. Match the monitor and its settings when comparing the two paths.




