Is it possible to use different groups of LDR images to resolve the camera response function ?

Hello, Thanks to OpenCV, we achieve to generate one HDR and recover the camera response. For our use case, we need to retrieve the camera response accurately. We noticed that from one HDR to another with the same device, the recovered response function is not the same. So it’s not stable and depends a lot of the content of the image. In our context, we cannot add more constraint on the shooting process, but we have many multi-bracketing LDR images of different locations. Is it possible to use all these groups of LDR images to resolve the response function? Thanks