The function "fetch a snapshot out of a video stream" currently requires an unproportionally huge knowledge of gstreamer inner workings to accomplish.

The first problem is to fetch a frame (a buffer) out of the video stream. This can be trivially accomplished if playbin is used, with the "frame" property. If playbin is not used, however, it is not straightforward nor intuitive to fetch the buffer.

Next, the buffer must be run through a special pipeline. This poses two difficult problems:

1. To "manually" feed a single buffer into a pipeline is very difficult, and various tricks exists that can solve the problem. None is straightforward nor intuitive.

2. The special pipeline must be created so that the buffer is properly converted into a gtk pixbuf. This is not straightforward nor intuitive.

All the things outlined above are not of interest to the application programmer trying to solve his problem; they are irritating obstacles. My personal experience is that this obstacle was so huge I considered not using gstreamer at all for the application I'm developing. I have also noticed that the question "how do I get a snapshot from a video stream?" regularly appears on the mailing list and in the IRC channel.

Conclusion; I suggest that a snapshot element (or some other, equivalent method) is introduced into gstreamer to hide this mess from the application programmer.