Interpreting results about boostingmonoculardepth

[yaksoy]

Okay. I don't think it's easy to tell which one has worse 'depth' in this example.

[Slazzo]

Just to confirm, I wasn't saying is worst, I should be more specific, the one on top fits better to the original scene, the depth spreads linear on reconstruction, it doesn't match but is close to ground truth.
The second doesn't look linear on reconstruction, most of the data is concentrated in planes, doesn't spread uniformly, I was assuming there must be some gamma issue when interpreting the result.

Assumed models trained on same dataset will give very much similar result, I can provide more details about scene reconstruction.