What do you like best about OBS Studio?
The scene and source model is the reason OBS has survived every change in how I produce video. A scene is a layout, a source is anything that can appear inside it, and everything I make is some arrangement of the two. I keep one scene collection per format: a webinar layout with slides, my camera in the corner, and a lower third; an interview layout with two camera frames side by side; a bare screen-capture layout for tutorial recordings. Sources are shared across scenes, so the color correction and noise filters on my camera live in one place and follow it everywhere it appears. When a new format comes up, I duplicate the closest existing scene, swap two sources, and I am rehearsing within minutes instead of rebuilding from a blank canvas.
Studio Mode is what keeps live sessions calm. The preview pane holds the scene I am preparing, the program pane holds what the audience currently sees, and nothing moves between them until I press the transition. During a live product demo I line up the next browser tab in preview, confirm the right page is loaded and the zoom level is sane, and only then take it to program. The audience never sees me fumbling for a window. Before I worked this way every scene change was a small gamble, and the difference in how relaxed I am on air is noticeable to me and, I suspect, to viewers.
Audio runs deeper than the mixer panel suggests at first glance. Each source carries its own filter chain, so my microphone gets noise suppression, a compressor, and a touch of gain before anyone hears it, while desktop audio stays untouched. Tracks can be split, which means my voice records separately from system sound, and a session that aired with the music slightly hot can be rebalanced in the edit instead of being redone. Audio monitoring lets me hear exactly what is going out while it goes out. A recent release reworked the mixer itself, and it shows: sources can be pinned and ordered sensibly now, and the panel finally behaves when the source list gets long. For a free tool this part is surprisingly hard to outgrow.
Hardware encoding is what makes recording on a working machine practical. With NVENC handling the encode, I capture 1080p60 product walkthroughs while the product itself runs on the same laptop, and the fans stay quiet enough that the microphone does not pick them up. x264 is there for the rare case I want it. Mostly I do not. The replay buffer rides along in the background holding the last few minutes in memory, so when something unrepeatable happens during a test session, one hotkey saves the clip after the fact instead of me wishing I had been recording.
The virtual camera quietly changed how I show up in ordinary meetings. It presents the full OBS composition to Zoom, Meet, and Teams as if it were a webcam, so a client call gets the same framed camera, the same name strap, and the same picture-in-picture screen share as a public stream. I control what is shared and when, from my own keys, instead of wrestling with the platform's share dialog. People comment on it, which is a strange thing to say about a video call, but they do.
Browser sources cover a surprising share of what makes a stream look produced. Anything that renders in a web page can live inside a scene: countdown timers, alert widgets, a shared agenda, an internal dashboard during a data walkthrough. Instead of waiting for a dedicated integration, I point a source at a URL and position it like any other layer. Half the custom touches on my streams are small web pages doing one job each, and that flexibility means the tool rarely says no to an idea.
Plugins are the multiplier on everything above. The websocket server ships in the box, so a Stream Deck on my desk switches scenes and toggles the microphone, and recordings start or stop without the OBS window ever needing focus. Community plugins cover transitions, per-source recording, and automation I would not expect around a free tool. Quality varies between plugins, and the good ones are very good. A built-in plugin manager arrived recently and has started to replace the old routine of hunting installers down from forum threads, which was the least pleasant part of extending the program.
Hotkeys and projectors are the unglamorous features I would miss within the first ten minutes. Nearly any action can be bound to a key, so a scene change or a mute happens from muscle memory while my eyes stay on the content, and the replay save lives under my left hand. A fullscreen projector mirrors the program output to my second monitor as a confidence view, which is how I once caught a frozen camera before the audience could. Neither feature shows up in screenshots. Both are why live sessions feel manageable.
What ties it together is that all of this costs nothing and asks for nothing. No account, no watermark, no recording length cap, no export tier. The project moves at a real pace too: recordings now survive an application crash because of how the file is written to disk, the output side speaks more than plain RTMP these days, and meaningful releases keep landing rather than the codebase coasting on its reputation. I have built paid work on top of free software before and regretted it. This is the exception. Review collected by and hosted on G2.com.
What do you dislike about OBS Studio?
The settings window is where new users go to get lost. The auto-configuration wizard will get a first stream working, but past that point you are facing output modes, encoder menus, color formats, sample rates, and bitrate choices with no guidance about which of them matter for your situation. I have onboarded colleagues who produce video professionally, and even they needed a sit-down session before they trusted themselves in there. My workaround is to stop explaining settings at all: I export a known-good profile and scene collection, hand both files over, and let people start from a working state. It is effective, and it is also an admission that the first-run experience expects more from newcomers than it should.
Native output goes to one destination at a time, and that remains the first wall anyone publishing on multiple channels hits. Streaming the same session to two platforms simultaneously means installing a plugin or paying a restreaming service to fan the signal out. There are narrower built-in cases where multiple renditions go out to a single service, but a plain checkbox that sends one stream to platform A and platform B does not exist. Given how normal multi-platform publishing has become, the absence stands out a little more each year.
Major releases and the plugin ecosystem move on different clocks. After a big version ships, the heavyweight plugins can take weeks to catch up, and an eager update can leave a production setup half working on a show day. I learned to hold updates until the specific plugins I depend on are confirmed compatible, and I keep the previous installer around in case a rollback is needed. Safe Mode, which launches the program with plugins disabled, is the right first diagnostic when something misbehaves after an update. The plugin manager should narrow this gap over time, but today the cadence mismatch is real and managing it falls on the user.
Recordings always leave for an editor, because nothing inside the program can trim them. Even a clean session needs its top and tail cut, so every capture makes a round trip through separate software before anyone sees it. I understand the scope argument, a broadcaster is not an editor, but a basic trim on the recording output would shorten the path for the simple cases that make up most of my week. Chapter markers dropped with a hotkey while recording help the edit go faster. They do not remove the edit.
Capture has sharp edges on fresh hardware. Dual-GPU laptops are the classic case: a capture source shows black until you force the program onto the correct graphics adapter, and nothing in the interface tells you that is the problem. Took me an embarrassing amount of searching to fix the first time, honestly. Some applications refuse to appear in window capture and need display capture as the fallback, and macOS adds its own round of permission prompts on first run. None of it recurs once a machine is set up, but the first hour on a new laptop can feel like a scavenger hunt. Review collected by and hosted on G2.com.
