The line with 1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 Bf5 is generally known as the Gunderam Defence, in which Black tries to set up a Caro-Kann Defence type formation, where White has a move and a half-open f-file in return for the pawn.
Gunderam analysed many of the important lines of this defence, although the most critical line may well be one that, to my knowledge, he didn't analyse extensively. White's most popular reply is 6.Ne5, preparing to expand on the kingside with g2-g4, and White can get sufficient compensation for the pawn if Black responds with the standard 6...e6 7.g4 Bg6 (after which White continues with Bg2 and h4, often transposing into 5...Bg4 lines), while the complications after 7...Be4 are also OK for White. However, 6...c6 7.g4 Be6!, instead of ...Bg6, is hard for White to generate full compensation for the pawn against. It looks odd, blocking in the e7-pawn, but from e6 the bishop is not vulnerable to pawn attacks and reinforces Black's control of the d5-square.
Since White gets nowhere after 8.g5 Nd5, I think 8.Bc4 is White's best bet, trading off the troublesome bishop. Then one line which I feel may be critical is 8...Bxc4 9.Nxc4 e6 10.g5 Nd5 11.0-0 Be7 12.Qh5 0-0.
At club level I would be happy to take White in this position, as there are clearly practical chances here, but objectively I am not sure that White has enough of an attack for the pawn. In particular the kingside is somewhat draughty. The problem with castling queenside, which rather appeals to me in many analogous variations, is that Black gets various threats of ...Nd5xc3, ...Bb4 etc. if White tries to castle to that side.
Therefore in a discussion at the Chesspublishing.com forum I advocated the rare 6.Bd3, inspired by an idea of Stefan Bücker in the Soller Gambit (1.d4 e5 2.dxe5 Nc6 3.Nf3 f6 4.exf6 Nxf6 5.Bf4 Bd6). After all, it can't be too bad with an extra tempo, and White can usually castle safely to the queenside in this line, which leaves White freer to push the h-pawn. Stefan himself was quite keen on the idea as well, but upon reflection I am not convinced that White can get full compensation for the pawn here either. Black continues 6...Bxd3 7.Qxd3 c6 intending ...Nbd7, ...e6 and ...Be7 with a solid defensive formation. One important continuation is then 8.Bf4 e6 9.0-0-0 Nbd7 10.h4 Be7 11.Rhf1 Qa5 12.Kb1.
Black must be careful not to castle kingside too quickly in this variation since White has a ready-made kingside attack, by advancing the g and h-pawns, but if Black keeps White guessing for a while as to where Black puts the king, then it will be difficult for White to make much of having the superior piece activity.
The line with 5...Bf5 is not a strong reason for club players to give up the Blackmar-Diemer, for even if, as I suspect, White's compensation for the pawn isn't quite enough, White's practical chances are still favourable enough to give good results and entertaining games. However, it is one reason why the BDG is not likely to be as reliable at high levels of play as, say, the Evans or King's Gambits.