How to · 4 min read
Make your AI-generated images interactive
Generating a good portrait takes seconds now. What you do with it afterwards is usually nothing: it goes in a folder or on a feed. Soft Touch gives those images a second life by making them touchable, whether they are photographs or fully generated.
It does not care where the image came from
Soft Touch never asks whether an image is a photo or a render. If skin is visible and a face or body can be recognised, it works. Painterly or illustrated styles respond too, as long as there is a skin-toned surface for the detector to find.
Because the detection runs on your device, there is no upload step and no waiting for a server. You choose the file, the model reads it locally, and a second later the picture is ready to touch.
Tips for images that respond well
Front-facing or slightly turned faces give the most convincing results, because the depth of the cheek is easiest to estimate. Very small faces in a wide scene may not be detected; if that happens, use the “point to the face” button and tap the face to re-run detection there.
Soft, evenly lit skin shows the blanching effect most clearly. Heavily stylised lighting still works, but the estimated light direction can be off — there is a slider to set it by hand.
Pose and idle motion
Besides touching, you can switch to Pose mode and drag the limbs of a full-body image, or leave the picture alone and watch it breathe, glance around and sway slightly. These small movements make a generated character feel present in a way a still image never does.
A note on consent
Use your own images, images you generated, or images of people who have agreed. Do not edit photos of other people without permission, and do not share edited images without the consent of the person shown. Images of anyone under 18 are not allowed.