Look here for an example.
The conventional simulation is an exercise in geometric planning. This means that you are suck in a 2D world making assumptions about where things are. Now this is not all bad - for instance, where is the bottom of the axillary lymph nodes? The 'old' teaching was that the mantle field needed to extend down to the bottom of the scapula to include the entire axillary nodal chain. Now it is hard to verify this with CT until you have a patient with nodes down that far … BUT … take it from me, this is 100% accurate! I have seen this on a CT scan - nodes almost down to the inferior scapula!
So all that knowledge is not 'old hat'. It has to be mixed with your CT knowledge. But I digress!
In conventional-simulation-land you can only make marks on the Xray and you have to set up and image all your radiation fields in the simulator, because you would have no software to help you.
So if you wanted to treat the prostate you could anything/everything to identify the prostate boundaries, but you had to mark it all on the geometry offered by the Xray. So contrast in the urethra, bladder and prostate - along with symphysis pubis bone - would define your prostate reasonably.
I have included some conventional planning pictures below. The edges were drawn with ruler and usually crossed bony landmarks (L5/S1 junction, mid-pituitary fossa, inferior border of clavicles, midline cord block, etc, etc).
You can simulate this simulation process (sorry!) by having a DRR of your treatment area printed and then marking the treatment field on the printed DRR without looking at the CT. In fact, why don't you go and do this exercise for a head and neck patient. The old texts will give you lots of help - Kian Ang's book has all the recipe's to follow.