There isn't any, one correct way.
If you have ever played any of Pusooy's games, many of his animations are done with partial images. For instance, he'll render the image of a left arm, one of a right arm, and one of the chest. His animations are then made by displaying those same body part images in slightly rotated and varied positions relative to each other.
Other people use the still-frame full photos that combine to make an animation.
There are also hybrids of the two. For instance you could completely still-frame the man's actions. Then do the same with the woman's. Then fuse them together in the final product.
I'm curious if you are using the DAZ timeline functionality currently. Using it, it will auto-generate intermediate positions (Meaning if a man's arm is straight down in the first frame, then in the 10th frame you set is at 90 degrees straight forward, the timeline will show him slowly raising his arm a bit more in every frame, without you needing to edit each frame manually). I will warn that the timeline can create some very, very unexpected results at times. Sometimes I certainly debate whether it is helping the process in the slightest or destroying it. Still, at times it's helpful.
As to the actual animation offered, it's certainly not how 2 real people would move. It's not static, and certainly gets the point across, so it has its merits. In real life, everything is always in motion, pivoting and shifting, all in different directions (actually, which things should not move is as important as which things should). It's far too complicated to be worth accurately reproducing.
One trick is to space out the movement that you do add. Choose 3 or so different areas within the image where variation is going to occur (the hips, the man's face, the woman's face). The specific choices can matter some (people are naturally curious and observant of other people's faces for example), but more so is that they are spread out. As a viewers eyes take in the final animation, and move about the image, you don't want large stagnant areas that should be moving. When their eyes pass over the right side of the image, something is moving over there and draws in the eyes. When they are on the left, their eyes are drawn to something else. The key thought here is, wherever they look, there is something close enough to their vantage point to draw the eyes to it, so they don't really notice the other parts nearby that aren't moving.
Another trick... change the camera angle and zoom for your animations. In your existing image you have the entire body of both the man and the woman. Those are a lot of body parts to correctly position and manipulate to make it look realistic. The less that makes it into the picture, the easier it is to make it more believable.
About 1 minute spent grabbing your image, cropping it to show less, then cutting out the right half so that the guy's arm didn't move any longer.
http://postimg.org/image/n6ta9zbmv/I realize this doesn't show everything you had in the original. But in a game if this was the animation, then clicking it to advance showed either a zoom of the woman's face as you have it now (appearing to orgasm), or even the zoomed out still-frame of the whole image as you have now... you gain an easier to produce animation, and it still tells the same tale.