Let me bullshit here a bit and think of a few limiting biological factors for flight which might be broken by a roc/dragon:
Lift vs. weight or energy
You need more lift if you have more weight. You get either lift by speed or wingarea.
Lift is defined as: L = 1/2airpresure*Airspeed^2*area*liftcoeficient
The obvious thing here is to get Airspeed. There are 2 ways to get that powered or unpowered. Unpowered for example is used say with Kites, a strong wind, ridgelift (air flowing up a hill), jumping of a cliff from there you can go into a soar say thermals. This is thought limited by windspeed (you are F*ed if you need Tornadospeeds) or the hight of your drop. To stay aloft you either need constant winds or thermals that push you up.
Powered is done by running, flapping your wings etc. . It has a few constraints thought. Stored energy in the form of fat means that you need more fat if you need more energy thus weight. The energy-density of fat isnt really great. Secondly oxygen, you want to burn your fat somehow after all. A birds lungstructure is wildly different from a mammal and more efficient generally but the available oxygen is still limited by lungsurface. For a bigger bird you need bigger lungs and the volume (thus weight of the tissue) goes quicker up then your surface. You also need more muscles (weight again) to move air around all that air.
Speaking of muscles the power of a muscle (25-33 Newtons/cm²) grows with its crosssection-area but thanks to the Square-cube-law if you double the size of your muscle you get only 4 times the strength but 8 (?) times the Mass. So at some point your muscles get to big to move themselves, your legs or wings around.
So a roc shouldnt be able to start at all because its to heavy and not powerful enough to accelerate to a velocity that would generate enough lift.
Mechanical loadsMaterials brake. If you take a long narrow piece of wood it might bend under its own weight if you hold it on one end, wave it around and it might even brake. Same goes for long bones. Or wings on wing on wings in KSP.
First of you want to hold your wings straight since you need that lift carrying you. Keeping your bones straight without flex at a given length means that you need a certain thickness which goes into mass. Even so turning, acceleration and deceleration puts stresses on them that are amplified by the length so at some point of your wing is to long it snaps because you wanted to do a turn.
Worse still then using bones in your wings is the decision of having "joints". At those points you have either to evolve a locking mechanism (which some birds did) or your wing flexes around them. You have to counteract that with muscles, so even in gliding/soaring you are using energy.
MoltingFeathers degrade with time and have to be replaced this costs comparatively huge amounts of energy and time. From one
article i found: "The length of a bird's flight feathers is proportional to its body mass raised to the one-third power, so that feather length roughly doubles with a tenfold increase in the bird's weight. But feather growth rate is proportional to body mass raised only to the one-sixth power."
So if the bird gets bigger you have factor in more time to regrow feathers and how much lift you lose from not having that feather. There are ways around it, stretching molting times or Molting all at once but these come with costs like less efficient flight to being flightless for a time. The bigest bird ever, Argentavis magnificens, with a wingspan of 6 meters and a weight of 70kilograms is theorised to having molted all at once subsiding on accumulated bodyfat for its flightless time.
Anyway my takeaway would be the following: Base the running start distance of birds on its acceleration and wingarea. Subtract jumping heights and existing winds from those starts. A similar thing can be done with landing if you can calculate a birds drag and stalling. This way you would get the right behaviour for all bird at all ages. Thermics would be cool but i guess they will be out for a long time.