VTOKU Cam gives you cinema-style control over how the shot looks: real-world occlusion so the avatar sits in the room, lens-accurate depth of field, autofocus that follows your subject, an anime color grade, and a graphics-quality setting. These live under Settings › Graphics, and they apply to the viewfinder and every output (recording, HDMI, NDI, SRT, VRL Link).
Occlusion and masking
With occlusion on, the avatar is hidden behind real-world objects and people in front of her, so she sits inside the scene instead of floating over it. Open Settings › Graphics › Masking.
- Occlude in the live camera, hides the avatar behind real-world objects in the viewfinder and every output, using the device's depth sensor. Requires a LiDAR-capable device.
- Sharpen with neural depth when still, when the camera holds still it occludes with an on-device neural depth model for crisper edges, then falls back to LiDAR the moment you move.
On-device and private. The neural depth model (about 50 MB) downloads on first use and runs entirely on the device. No camera or depth data is sent to VTOKU (see the Privacy Policy).
You can also send the occlusion or avatar matte to a compositor as a separate NDI source. See Send a mask over NDI.
Depth of field
Turn on Depth of field to defocus the background and foreground around your subject. It is modelled on the real lens, so the avatar blurs at the same rate as the room behind her. At ×1 it matches true phone optics; raise Blur strength to open the effective aperture for a shallower, more cinematic look.
Autofocus and focus tracking
The real lens focus is controlled from the FOCUS dial (tap the dial to toggle auto and manual, drag to rack focus by hand). On top of that, VTOKU Cam tracks focus for you:
- Tap to track focus, tap anywhere in the viewfinder to lock focus on that point; tap the reticle to release it.
- Follow the avatar, frame the avatar and focus racks to her whole body, not just her head.
- Face-priority for real people, when a real person is in shot the hardware autofocus prioritises their face.
- Share the shot, a virtual avatar and a live person can both stay in focus, so a VTuber and an IRL guest can appear together.
Screen FX
Screen FX applies an anime color grade over the whole frame, in the viewfinder and every output. Open Settings › Graphics › Screen FX, pick a preset, then fine-tune it. Adjustable parameters include intensity, exposure, contrast, saturation, bloom, gradients, vignette, grain, color fringing, anamorphic flare, and dust motes. Save your own settings as a preset to reuse later.
Graphics quality
The Quality setting trades visual fidelity for performance: Auto matches your device, and Low / Medium / High let you set it by hand. Lower quality is lighter to render, which helps keep streaming smooth and the device cooler. You can also tune the avatar's ground shadow strength, length, and softness here.
Outdoor mode
Outdoors and in large rooms the LiDAR sensor runs past its range, so the noisy depth it reads can pull on the world anchor and make the scene drift. Turn on Outdoor mode under Settings › Camera to drop LiDAR depth and track from the camera and motion sensors alone, which holds steady far from walls. Placement and live lighting still work. Real-world occlusion, depth-of-field focus, and ground reflections turn off while it is on, and tracking restarts when you toggle it.
The TRK status tells you when tracking is struggling and why: LOW DETAIL when the scene is too plain to track, SLOW DOWN when the device is moving too fast, and RELOCATING while it finds its place again. If you sit in LOW DETAIL for a few seconds, the app suggests Outdoor mode once.
HDMI monitoring
Connect the device to an external display over USB-C or HDMI, then turn on HDMI under Settings › Output to mirror the live program to that display, clean of the on-screen controls. Turn on Include microphone audio to route sound to the display as well; the phone speaker is restored when HDMI output is off. HDMI output is off at every launch.
Tips
- Occlusion and the neural depth model need a LiDAR-capable device; on other devices the avatar renders without real-world occlusion.
- Filming outdoors or in a large space? If the scene drifts, turn on Outdoor mode. It trades occlusion and focus tracking for steadier pose.
- Set your look (depth of field, Screen FX) before you hit record or go live so the take is consistent.
- If streaming feels heavy or the device runs hot, drop the Quality setting a notch.
Next: Recording takes →