What you will need to do is research prospective sales for the product you plan, which could be tricky to do when you have no track record to draw upon, although you can use the sales popularity of other add-ons to gauge what would be a good choice, and the amount of freebie DLC downloads can also offer a guide to what makes a popular subject. But where actual numbers are concerned, it is a bit trickier to determine how you will do, since you are effectively an unknown quantity. Presently, Jointed Rail's NS Heritage models are selling well, and much of that is because they are current news in the real world of trains, and the locomotive those liveries appear on are ones which a lot of modern traction fans like, older less contemporary stuff is more likely to be subject to the whims of people's often not quite so esoteric tastes.
The screenie you posted of HP Trainz Marias Pass route is a good study case for that; as far as the scenery is concerned, it is accurate where DEM terrain elevation data is concerned, which makes it appeal to those who value prototypical realism, since the main appeal of the Marias Pass is of course tractive effort management required to get long trains up over Summit. But for most US rail fans, in terms of subject material, places like the Rathole, Marias and Donner being the legends they are, makes them no-brainers as subject material, since even people not necessarily that into BNSF would probably fork out money for a decent Marias Pass route anyway; I know I did, and I hate that orange BNSF scheme LOL, being more of an NS enthusiast myself when it comes to modern Class 1 railroads. Thus it is less important for a model of Marias to be peerlessly stunning from a visual modeling standpoint, or indeed where rolling stock is concerned, since there are no shortage of great BNSF models to be found, which is handy since such a product has had to balance eye candy with RAM usage owing to the fact that it covers about 200 miles of track through the mountainous terrain of the Glacier National Park, which has a lot of foliage about. Thus that product also has well designed sessions and the locations along the route relatively accurately custom-modeled, maintaining the prototypical plus point factor which it offers in being accurate as far as elevation is concerned, all this being the product's USP if you like.
Of course, modeling the entire Marias Pass route to any extent is no small creative task, but even with the work involved in having done so, it is still having to come in at a fairly inexpensive 20 Euro price point. If one works out the man-hours involved and cross references that with the price, minus overheads, and then also consider the number of units one has to sell to make a profitable venture where people might also take a cut as part of their fee for modeling, it doesn't take a rocket scientist to work out that you'd have to shift a lot of copies in order to make your hourly rate for the work involved be something which would make it anything more than an interesting side venture rather than something you could expect to make a living off. The reality is that even people who work for really successful third party simulator companies such as PMDG rarely do that alone to make a living, and it is often a labour of love as much as anything else.
These are the sort of things you will need to study carefully in order to determine if something is truly viable, and you will have to have at least some figures to hand if you are to offer modeler's a percentage, because 3 percent of something which you can realistically project will sell well might be a great deal to offer, whereas 20 percent of something which sells next to nothing, is nothing. And with no track record, you're probably going to have to offer a higher percentage if you can't talk turkey with some convincing research figures relating to what prospective sales might be.
Al