To explain the initial encoding of pattern information in the human visual system, the standard psychophysical spatial vision model is based on channels specific to spatial frequency and orientation, followed by divisive normalization (contrast gain-control). Schütt and Wichmann (2017, Journal of Vision) developed an image-computable implementation of the standard model and showed it to be able to explain data for contrast detection, contrast discrimination, and oblique and natural-image masking. Furthermore, the model induces a sparse encoding of luminance information. Reichert and Wichmann (2024, VSS, Journal of Vision) provided a modern, efficient, automatically differentiable implementation of this model enabling easy and fast training of model parameters on image data.

We extend Reichert and Wichmann's implementation by allowing learning of spatial and spatial-frequency parameters of contrast gain-control and the addition of different kinds of noise anywhere in the processing pipeline.

Using the same psychophysical data as Schütt and Wichmann, we compare our model’s predictions of contrast detection, contrast discrimination, and oblique and natural-image masking with the previous implementation. Moreover, we discuss possible interactions between the newly learnable parameters and effects of multi-stage noise on the image-processing pipeline.

An advantage of our model is its modularity and automatic differentiation because this facilitates rapid implementation, addition, and testing of new components of the standard spatial vision model. Furthermore, our framework allows the integration of this psychophysically validated spatial vision model into larger image-processing pipelines such as deep-learning and other artificial-intelligence models, where it may increase performance or energy efficiency due to its relative simplicity.