Just a couple of thoughts on this (not so recent) question:
The first (selfish) goal is to develop a reputation for being a intelligent, insightful colleague with good comments.
If everyone went into a workshop or conference talk with the primary goal of "developing a reputation for being an intelligent, insightful colleague with good comments", then...yikes. Chaos would ensue. Luckily I've never seen this; the closest I have seen is in faculty meetings, where often I feel that being a good colleague requires me to say something from time to time just to be heard, rather than because I think I am contributing something absolutely essential to the discussion. It is, frankly, one of my worse habits: I shudder to imagine what would happen if all of my colleagues behaved the same way.
You should speak up in a talk because you think that something you say will be directly helpful to the speaker or to others, or because you didn't fully understand something the speaker said and you have reason to believe that a quick question and answer could set you back on track. I don't think that you should aim to make yourself heard specifically to impress your colleagues: that seems too likely to backfire, either because what you say is not as insightful as you think or because the speaker and the audience will not appreciate your commandeering more than your fair share of the speaker's time.
The second goal is to effectively communicate my concerns to help others improve their research.
Yes, that's a good reason to speak up. What you say about "too soft" questions getting brushed off by speakers sounds like a "local phenomenon" rather than a general truth about such talks. Where I come from, too-pointed questioning is more likely not to have the desired effect: a speaker either becomes flustered and the entire talk becomes (at least!) temporarily derailed, or to avoid that they say "I don't know" and move on without really thinking about the question. For several years this type of thing happened often in response to questions of mine that I didn't even realize were so pointed, so I have (somewhat) learned to correct in the opposite direction: by trying to ensure that my questions are friendly and do not come off as quizzing or challenging the speaker. But mine is not a universal truth either: I think it just depends on the local culture involved, maybe even the specific group of people in the room.
I gently communicate the problems I see with the work.
If you feel like you see a problem -- rather than just asking a question about something that you may not properly understand -- then pointing this out during the talk itself may not be ideal. The following is in my experience a universal truth: people don't like being told they're wrong. Do you want them to stop short and try to "fix" their work on the spot? I think that if you really have something to say which vitiates a substantial part of the talk, it will be better for all involved if you wait until after the talk and speak solely to the speaker about it. (Admittedly, j'accuse moments in talks can be pretty entertaining for the spectators: see e.g. the beginning of this wonderful story. Spoiler alert: as amazing as this beginning event is, yet more amazing is that it is not what the story is really about at all.)
I might go so far as to say that it is not really "fair" to raise a serious objection during a fixed-duration talk. The speaker has other things on her mind and plate besides understanding and addressing your specific concern. Often in a talk it turns out that the brilliant professor's lightning-quick refutation of the speaker's work is not actually a refutation of the speaker's work: it was based on some kind of misunderstanding or miscommunication (e.g. people using terminology in subtly different ways or elided technical hypotheses). When this happens the majority of the audience cannot really follow what happened and goes away with a vague impression that something was wrong with the speaker's work. That's not really fair.