Cutting at specific frames and cutting at I frames are very different operations. With the current consumer video formats it is actually impossible to cut at a specific frame without losing quality, at least in a few frames near the cut.
Because frames are dependent on the previous frames, you need all the frames up to the last I frame in order to decode a later frame. This is why you sometimes see digital video go into that blocky smeary mess when the connection is poor. The decoder lost the I frame or a previous P frame and is decoding P or B frames without the original reference blocks.
This means frame accurate cutting of video requires decoding and re-encoding of the entire video or only those frames missing reference blocks, which is sometimes called "Smart Rendering".
How to cut video frame accurate without encoding?