Modify example in order to get larger than 244x244 resolution on combination image

In March I was able to generate images larger than 244x244 in my project, based on that example. But then I havelost my code and don’t remember what was the deal allowed me to generate images with more resolution than 244x244. Now attempt to increase width and height say to 2440x2440 fails due to

Map<String, INDArray> activationsContentMap = vgg16FineTune.feedForward(content, true);

probably expect the NDArray of the specific format fitting the size of the specific layer of the vgg16. So how I was able to produce the larger images? Or probably there was another model (like vgg19) in the previous examples and its layers were larger?

https://github.com/eclipse/deeplearning4j-examples/blob/master/dl4j-examples/src/main/java/org/deeplearning4j/examples/advanced/modelling/styletransfer/NeuralStyleTransfer.java