This isn't because of wolfcam, it's due to a number of reasons...
it's because demos, especially non-POV ones don't accurately show the backwards hit reconciliation from the unlagged netcode working, ( Sorry if I can't articulate this accurately right now ) the enemy model in demos will probably be slightly ahead of where you're tracking. AFAIK there isn't much that can be done about this.
To get results like those seen in the video, either you or your opponent is probably playing on a real high ping, or with an especially laggy connection.
Are you sure this is happening in all your demos? Against all different opponents? I know from my work in wolfcam that there's a few players in particular whose models are way off their hitbox in demos. I've even seen middies hit in wolfcam a full model's distance away from the model haha.

as it is a demo and it has already happened it should not be too complex to fix (play the hitsounds <ping> milliseconds earlier, need to read ahead in the demofile for this probably)

so, make the sound out of sync with what's actually going on in an attempt to make it look like my shots hit sooner than they do?

noodle shaft is still noodly and not anywhere near the model on any of the hits =/

strenx's montage shows him LGing a guy on bridge on DM6 and it looks like he hits 100 LG, but sounds like 40-50 T_T

It's going to be something to do with lag prediction. From what I know, basically what's happening is you're using truelightning or some sort of lag-prediction WITHOUT truelightning. This will force the beam to appear somewhere that it isn't; the LG beam is hitting the guy, and in-game, it LOOKS like it's hitting him because of the ACTIVE lag prediction.

But in the demo, there is no active lag prediction, so there are cvars to enable the visuals of the lag prediction. I just forget what they are...

I'm probably wrong, but this is what I remember from what I learned back in my heyday. The visual interpretation of most entities are different from what they should be: they are using a lot of interpolation and other prediction methods.

Because I'm too lazy to rewrite this post, here's a link to a video where I fixed my lagged projectile issue: https://www.youtube.com/watch?v=pFtqoz1IJao

interp is the reason POV demos from LAN in CSS makes the player look like he prefires his opponents around corners, so it would stand to reason that interp could make player models in quake appear somewhere different in demos as opposed to real time in game.

the question is, how to make it go back? >:(

Look at the description of the video I posted. Those are the cvars that, back in "teh day", caused the lagged projectiles bug.

its been happening in all of the demos that ive looked at, going all the way back to january of this year.

both of us had 50 ping in this particular video, and i have many more on youtube and on my hard drive that have the same thing.