Gemini Live Video is incredible.

9to5Google www.youtube.com

Kurze Zusammenfassung

Das Video behandelt die neue visuelle, KI-gestützte Funktion von Gemini, die langsam ausgerollt wird und nun breiter verfügbar ist. Es wird erklärt, wie man darauf zugreift, was man damit machen kann und ob sich die Nutzung lohnt.

  • Gemini Live bietet neue Möglichkeiten zur Interaktion mit der Umgebung und zur Unterstützung bei Aufgaben.
  • Der Zugriff erfordert ein Google One AI Advanced-Abonnement.
  • Die Genauigkeit ist nicht perfekt, und es kann zu Fehlern kommen.

Einführung in Gemini Live

Google rollt die visuelle, KI-gestützte Funktion von Gemini langsam aus, aber sie wird nun breiter verfügbar. Die Kamera- und Bildschirmfreigabefunktionen sind technisch Teil des April 2025 Pixel Drop, funktionieren aber auf fast jedem Android-Telefon. Pixel-Nutzer erhalten in den nächsten Tagen ein Update. Nach dem Start von Gemini wird ein Pop-up angezeigt, das die neuen Funktionen ankündigt.

Zugriff und erste Schritte

Um auf die neuen visuellen Modi zuzugreifen, ist ein Google One AI Advanced-Abonnement erforderlich. Besitzer bestimmter Pixel- und Galaxy-Modelle erhalten eine kostenlose Testphase. Der einfachste Weg, die neuen Steuerelemente zu nutzen, ist über die Gemini-App. Dort gibt es eine Kamera- und eine Präsentationsschaltfläche. Über eine Schaltfläche kann die Bildschirmfreigabe gestartet werden, wobei der gesamte Bildschirm freigegeben werden muss.

Benutzung und Funktionen

Der Start und die Einführung sind sehr schnell. Die Benutzeroberfläche ähnelt der Pixel-Kamera-UI. Es kann zwischen der Haupt- und der Selfie-Kamera gewechselt werden. Der Fokus der Kamera kann etwas ungenau sein. Gemini ist schnell bei der Erkennung von Objekten und der Bereitstellung von Informationen. Es fehlt eine Zoomfunktion. Die besten Anwendungsfälle sind detailliertere Fragen zu Objekten, Orten und Bereichen. Beispiele sind die Bestimmung des Kaloriengehalts von Lebensmitteln oder Allergeninformationen.

Beispiele und Übersetzungen

Es werden Beispiele für die Verwendung von Gemini Live gezeigt, darunter die Frage, ob ein Lippenbalsam für Vegetarier und Veganer geeignet ist, und Ratschläge zur Reinigung eines Kamerasensors. Auch Echtzeitübersetzungen wurden getestet, die recht genau erscheinen. Gemini Live zeigt jedoch keinen Text auf dem Bildschirm an, sondern gibt nur Audio-Antworten. Textbasierte Antworten sind erst nach dem Beenden der Sitzung im Chatprotokoll sichtbar.

Anwendungsfälle und Bildschirmfreigabe

Gemini Live kann als visuelle Lernhilfe oder für Tutorials verwendet werden. Die Videofunktionen funktionieren reibungslos. Die Bildschirmfreigabe wirkt weniger natürlich, da es kein visuelles Feedback gibt. Nützlich ist die Funktion, um Webseiten zusammenzufassen oder weitere Informationen zu erhalten, ohne die Seite zu verlassen. Es wird erwartet, dass zukünftige Integrationen, wie das Hinzufügen von Artikeln zur Einkaufsliste oder Flugdaten zum Kalender, die Funktion verbessern werden.

Genauigkeit und Fehler

Gemini ist nicht perfekt und kann Fehler machen. Bei bestimmten Artikeln konnte Gemini keine genauen Informationen liefern. Es besteht die Gefahr von Halluzinationen, insbesondere bei Dingen, die man nicht kennt. In einem Beispiel wurde ein Game Boy-Spiel falsch identifiziert. Eine klare Sicht auf das Objekt im Sucher hilft, Fehler zu vermeiden. Die Informationen sollten nicht blind vertraut werden.

Fazit und Ausblick

Android XR- und AR-Plattformen werden stark auf Gemini angewiesen sein. Gemini Live ist ein erster Schritt in diese Richtung. Es ist ein guter Anfang, aber man sollte den Informationen nicht blind vertrauen. Die Funktion wird sich mit der Zeit verbessern. Die Bildschirmfreigabe ist noch begrenzt und wirkt wie ein "Showeffekt". Es wird gehofft, dass sich die Funktion weiterentwickelt.

Fasse jedes YouTube-Video in Sekunden zusammen

Hol dir Zusammenfassungen wie diese für jedes Video, in deiner Sprache, mit der kostenlosen Summ-App.

Laden im App Store
Jetzt bei Google Play

Transkript

Vollständiges Transkript von „Gemini Live Video is incredible.“. Klicke auf einen Zeitstempel, um das Video an dieser Stelle zu öffnen. Du brauchst ein anderes Video? Nutze den YouTube-Transkript-Generator.

0:00 So, Google has been rolling out Gemini's live visual astro powered function super slowly over the past couple of weeks, but it's starting to expand. It's rolling out pretty widely as of today. So, here's what it's like to use and what you need to know and if it's worth even giving a go. But first, if you want more Gemini and Android content, well, you know what to do by now. Hit that subscribe button. It helps me, it helps the channel, and you get some help along the way. So, cheers, and let's get back to it. So, Google is suggesting that the arrival of the camera and screen sharing controls are actually or technically part of the April 2025 Pixel Drop, but this has been non-scheduled. It's a set of features that definitely feels odd given the function will work on practically any Android phone. That said, it is technically tied to that. So, yeah, if you have a Pixel phone, you should get an update over the next couple of days. When you do launch

0:45 Gemini on your Android phone, you will get a pop-up to indicate that Gemini Live's Astropowered features are ready for you to test out. The mini pop-up says that you can talk through ideas, learn about your environment, or get help with what's on your screen. The latter refers to the screen sharing function, which is actually just like a souped-up version of Circle to search. So, you're probably wondering, how can I access this? Well, to access any of the new visual modes, you will need a Google 1 AI advanced subscription plan. If you happen to have a Google Pixel 9 Pro, 9 Pro XL, Pixel Fold, or even some Galaxy S25 models, you will have received a substantial free trial of this paid tier. The trial is 12 months on the Pixel 9 Pro models that was from launch and 6 months on some S25 devices. Google has given us a couple of ways to access these new controls as well. I think the

1:31 easiest is via the dedicated Gemini app. So, when you launch the Gemini Live application or Gemini Live from within the application, that usual call style UI will gain an extra couple of buttons. This is probably the dead giveaway that you have something new. There's a camera button and there's a presentation icon. So, you can use your voice with that usual wake word. I'm not going to say that out loud and activate devices in my office here or at your home as well. Basically, once you've done this, above that compact Gemini pop-up bar, you'll also see a share screen with live. That's just a tapable button that will start that AI powered screen sharing. You're not able to share or select a single app at this stage. Instead, you will have to ensure share your entire screen. This might be intrusive for some people. So, yeah, give that one a miss if you're worried about that. When it is active, a status bar chip will indicate

2:17 that this is live and running. Tapping this just allows you to quickly end or close your session. So, it's super simple, super easy to activate no matter which option you do choose. So, what is it like to use? Well, the initial startup in the introduction is very, very fast. But the viewfinder opens, especially when you're using that video mode, and it's really, really easy to decipher. It's actually somewhat mimics the Pixel camera UI, so it's going to be instantly familiar to you if you've ever used Google Pixel phone or if you've used a Google phone from a while back. You don't have to use that rear camera as well. There is a toggle to switch to the selfie camera if you want to put yourself or your background in the frame or ask questions or advice about whatever is in the frame. And that's basically what this feature is doing. When you're using the live astro mode or when I've used it, I've actually found that the camera focus can be a little

3:04 bit finicky. Um, all the lenses themselves will switch automatically, which can be annoying when you get things up close to the camera itself. It advises you to stay still or keep the subject as still as possible. I can attest to this, but despite some subjects not being in focus, I do think Gemini is actually quick to determine what something is or produce information without too much difficulty, not being able to zoom in or even even with devices with dedicated telephoto lenses does feel a little bit shortsighted. So, if you want to point something out in your world in the distance, you're going to have to get up very very close. I hope that Gemini can work out or pick out what you're pointing the camera at. I do think for simple or basic queries, it is perfectly adequate. However, I think the best use cases for deeper questions like help or advice on objects, areas, and locations rather

3:50 than going out and just simply asking things or basically the kind of things you would just search for yourself. I think a prime example is getting the calorific uh content on food items or even the allergen information. And just to give you an example, here's how a few of these queries have gone. I've talked through these, you'll hear my responses, and you can decipher for yourself what you think is well worth it. Is my lip balm suitable for vegetarians and vegans? Okay, I've had a look at the ingredients of your Blissex Agave Rescue Lip Balm, and it appears to be suitable for vegetarians, but not vegans. It contains beeswax, which is an animal product. Would you like me to suggest some vegan friendly alternatives? Yes, if that's possible.

4:36 Sure, you can find her. Moonlight Lip Balm at Holland and Barrett for £4.99 or on Amazon, though it's currently out of stock there. For ELF Holy Hydration Lip Mask, it's available at Superdrug for £6. I have a little mark on my camera sensor here. What's the best way to go about cleaning it? Can you see the mark on that? Yep, I see the mark on your Lumix S1R2 sensor. For cleaning, you'll need a blower, sensor swabs, and sensor cleaning solution. It's a delicate process, so take your time. Do you want me to walk you through the steps? Yeah, that would be great. And also, could you give me some advice on where to get sensor cleaning fluid and the tools needed to actually clean the sensor?

5:27 Sure, you can find sensor cleaning kits with fluid, swabs, and a blower at camera stores or online. Major retailers like Amazon, Jessups, and Curries usually carry them. Do you want me to find a specific kit for you? If that's humanly possible, yeah, that'd be great. So, there's a couple of examples, but I also tried some real-time translations. I do think they seem to be fairly accurate. Gemini Live doesn't give you any or show you any on-creen text, so you only get audio cues and answers. I do think Google Lens overlays with the translated text might be better or a better option for translating signs and text on while whilst you're traveling. At the moment, you do have to wait to close a session before you get to see some of the textbased responses. You will get that chat log like you do on a regular Gemini text chat. And I think it's probably going to be for the best

6:13 if you do use this a little bit a little bit more often. I do think this probably could have been a little bit better implemented. But for n for now, it's absolutely fine in tandem with other applications like Google Maps and various other services. You could probably use this as a visual learning aid or a tutorial option. maybe just get some extra advice from something you haven't been able to search correctly. I also tested Gemini Live's Astra style video functions on the Pixel 9 Pro XL as you can see here. And although it wasn't always instant, it felt pretty smooth and responses were almost always fast enough to feel fluid and there is a little bit of a conversational element there. Even though Gemini does have to react to your questions and your queries, I think screen sharing, this is one of those features that feels a little less natural than using the camera and pointing at things because as

6:58 I say, you get no real visual feedback at all. Merely audio cues and information doesn't feel quite as integrated in the same way as the camera itself. I think doing things like getting webpage summaries about is about the extent of the features real usability or getting further information without leaving a page or screen that might be really useful to you. I'd wager this will change as we get more integrations. I'm thinking things like the ability to add things to your shopping list with URLs maybe flight information adding that to your calendar when you're making travel plans. We can't do any of that yet at this stage but there is an option here to share your screen and get information and context. I do think it's yeah a little bit disappointing. This is the one disappointing option here, but the video modes seem pretty good so far. So for anyone wondering about accuracy and any other considerations you might want to

7:44 have. I think like almost all AI platforms, Gemini is definitely not a perfect system. In fact, it can get things wrong in lots of scenarios and it does well here with information recall in the most part. I found that certain items did cause issues where Gemini wasn't able to give me that accurate information. As I noted, this is going to be fine for things you know about and you can effectively interrupt and course correct, at least course correct that way with AI. The problem here is that hallucinations about things you might not know about could lead to problems in future and lead to problems when you do start doing that in air quotes research. For instance, I asked about a Game Boy game cartridge, as you can see on screen now, and Gemini misidentified the car and gave me some wildly incorrect information about the title and gameplay. Have a listen for yourself and you tell me what you think. Okay, you're

8:30 playing Pokémon Crystal. A lot of players recommend up a That's not That's not Pokémon Crystal. That's Pokémon Yellow. So, what starter Pokémon are going to use in Pokémon Yellow? Ah, my mistake. Since you're playing Pokémon Yellow, you'll start with Pikachu. Okay, those are the GeekW World GK3 in-ear monitors. They're a budget friendly option known for having a balanced sound. Would you like me to

9:16 find that adapter for you now? No, I think you're incorrect. That's the model on the side there. So, if you could find me the exact adapter for these IM, that would be great. Oh, you're right. Those are the Rapto XHBB Hook X. To find the right adapter, could you tell me what device you're plugging them into? I'm going to plug them into the front of my laptop, and I believe that's 3.5 mm port. So, yeah, whatever would work with that would be great. Okay, the RaptG go XHBB Hookex earbuds use a two pin connector, so you'll need a two pin to 3.5 mm adapter to plug them into your laptop. You can find them online for around £10 to £15. Would you

10:02 like me to find some options for you? That contradicts what you've just told me with this Gemini. So, what do I need for this particular end of the connector for those wrapped Go IM? My bad. That's a four pin mini den connector. You'll need a four pin mini den to 3.5 mm headphone jack adapter. They're readily available online from places like Amazon or electronic stores. I also want to say making sure you have a clear view of something within the viewfinder does help mitigate this. I just adjusted where I filmed and instantly resolved this problem without much intervention. The thing is you simply can't trust the information blindly, at least at this stage. Gemini is still prone to errors and they range from minor in this case to major in

10:48 equal measure. So what do I think so far having used this for a few hours since it's been available? Well, Android XR and AR those platforms I think are going to rely heavily on Gemini and as a means to interact and get information about the world around us. Gemini Live utilizing these Astro style functions I think is that first step or the first realization of that end goal. We might be a few years away from Google made ARI glasses or even third-party AI glasses powered by Google's platforms, but the demos shown late last year felt like the the way that this is probably going to head. It ain't a bad way so far to kick things off. That's all I'm trying to say. Like any AI product, I definitely would say be careful putting 100% trust in the information that is spewed out to you. That said, I do think it feels like a great start and a really solid way to

11:34 interact with the world around you or help you get information and get things when you're stuck or having problems. I do it will improve over time and it will get better as more data points are introduced. That means more people using it. So, at least in theory, it's probably the worst that Gemini live video or these mid modes will ever be, which I think isn't all that bad to begin with. Sharing your screen though, now this is the one area I'm still not sold on. I think it's very limited at this stage and while it's a nice secondary option, I do think it's nice that they added this period, it's even more limited. It if it can develop to play nicely with more of your other services, I think it will be a useful tool. Right now, it just feels to me a bit like a parlor trick that does really little more than the regular Gemini Live conversations that you'll have with that service as is. Let me know if you've

12:21 managed to try this for yourself. What do you think? And tell me what you thought of the demos that we've just shown in this video. I'm really hoping this evolves as it's a good basis for more development. That said, hopefully it gives you an idea of what to expect. Cheers for watching though and I will speak to you soon.

Share

Weitere Video-Zusammenfassungen