Character Lock — How to Get Perfect Consistency in Every Photo

If you have ever generated several photos of the same companion and noticed her face changing from image to image, Character Lock is the feature you have been looking for. It is VelourCrush's answer to one of the hardest problems in AI image generation: keeping a character perfectly consistent every single time.

Here is why photos look different without it. Standard AI image generation works from a text description of your companion — her hair, her eyes, her body type. But a text description leaves room for interpretation, and the model fills that room a little differently on every generation. The result is a companion who is recognizably similar but not identical from photo to photo. Small differences in the face, the exact shade of her hair, or the shape of her features add up, and the illusion of one consistent person starts to slip.

Character Lock fixes this by training a personalized model of your specific companion. Instead of relying on a text description alone, the system learns what she actually looks like and locks that identity in place for every future generation.

Training is simple. You upload one photo of your companion — the anchor image that defines who she is. From that single photo, VelourCrush generates 15 variations of her, capturing her from different angles and in different expressions so the model learns her identity thoroughly rather than memorizing one frozen pose. Those variations become the training set. The whole training process takes roughly 15 to 20 minutes, and you do not have to babysit it — you can leave and come back.

Once training finishes, everything changes. Every photo you generate afterward is unmistakably her. The same face, the same features, the same person, in every act, every setting, and every pose. Her identity carries into video too, so your clips look like the same companion you see in her photos. Character Lock turns a collection of similar-looking images into a consistent, believable person.

Character Lock costs 500 coins to train. It is a one-time cost per companion — once she is locked, every future image benefits from it at no extra charge beyond the normal per-image coin cost. Given how many photos most users generate over time, it pays for itself quickly in quality.

You will find Character Lock on the character profile page. Open your companion's profile, look for the Character Lock option, upload your anchor photo, and start training. About 15 to 20 minutes later, she is locked in — and every photo and video from that point forward looks exactly like her.