Last week we proudly announced the release of four stunning, pre-trained models from Lundin GeoLab in OpendTect’s Machine Learning solution. In last week’s blog post, we showed an example of one of these models: SimpleHmult, a 3D Unet model that attenuates horizontal multiples.

Desmile; another pearl in OpendTect’s library of pre-trained Machine Learning models

Today, we present Desmile. Like SimpleHmult, Desmile is a 3D Unet. Its purpose is to remove steeply dipping events such as migration smiles from post-stack 3D seismic volumes. The model was trained on synthetic examples with and without migration smile artefacts. The shape of the training examples was 128x128x128 samples. The slider shows a seismic section before and after application of Desmile (courtesy Lundin GeoLab).

For more information or to request a Machine Learning demo license, please email This email address is being protected from spambots. You need JavaScript enabled to view it..

The next post in this series of posts is available.