Surprisingly, the maximum intensity one can obtain from a light source does not depend on the distance to the light source nor on the distance between the light source and the imaging equipment. Instead, it depends on the distance between the imaging equipment and the object being heated (assuming that the imaging system is designed to be the best possible for these distances). What you want to do is "refract as close as possible" to the object being heated.
By thermodynamics, it's not possible for the light emitted by an object to heat up another that is already hotter than the first. For example, this means that we can't use an imaging system (or a non imaging power concentrating system such as a collection of mirrors) to use sunlight to heat a spot on the earth to a temperature higher than that of the sun.
So when you use an imaging system to magnify the effect of the sun, you do not get an image that is hotter than the sun. The best you can do is to make the image of the sun appear larger. And indeed, this is what a magnifying glass does. If you look through one (from the distance appropriate to frying an ant), you will see that objects in the direction of the image appear larger than they actually are. So when you apply a magnifying glass to an ant, what the ant sees is that the sun has increased in size.
The best you can do is to make the sun appear over all the sky, plus, with a mirror, to appear in the direction of the ground as well. Alternatively, you can arrange for the bottom side of the ant to be an insulator. (Uh, even concrete is a pretty good insulator if you try this at home.) This will heat the ant up until it approaches the temperature of the sun (with optimum optics and insulation). At that time the ant will radiate as much heat back to the sun as the sun radiates to the ant. And so the heat flows will balance and the ant will cease heating up. (All this assumes black body type radiation which is generally an okay approximation for this sort of thing.)