Modeling the soot emissions of a Diesel engine is a challenge. Although it was part of many works before, it is still not a solved issue and has a substantial potential for improvement. A major problem is the presence of two competing effects during combustion, soot formation and soot oxidation, whereas only the cumulative difference of these effects can be measured in the exhaust. There is a wide consensus that it is sensible to design crank angle resolved models for both effects. Indeed, many authors propose crank angle based soot models which are mostly based on detailed first principles based structures, e.g. spray models, engine process calculations etc. Although these models are appealing from a theoretical point of view, they are all lacking of the required measurement information to validate all the complex model parts. Finally, most parts of the model remain at their assumed values and only a few parameters are used for calibration. Against this background in the actual work a gray box approach is presented, where the basic ideas of a separation into formation and oxidation in the crank angle domain are followed, however the remaining structure is determined in a systematic way. To this end a two-color spectroscopy in cylinder soot measurement was applied and a sequential input selection method was combined with a generic nonlinear optimizer to predict the crank angle resolved soot emissions and in particular the tailpipe values. The proposed approach was compared against the well known model of Hiroyasu  on the basis of testbench measurements of a Euro 4 passenger car Diesel engine, whereas a considerable improvement of the predicted tailpipe soot emissions was achieved.