We know that AI Tech is progressing at a rapid rate, and the recent unveiling of Google’s Veo 3 video generator once again highlighted it. The VO3 is a step in terms of realism, and adds audio for the first time. Most of the time, outputs from VO3 are much more unmatched than the actual video.
This is not yet correct, but the VO3 is taking the internet from the storm, covering everything with viral clips Street interview To Disabled stormPainting the soldiers of the Gangeya Empire is a way to ensure stability between the least clips, as they definitely look the same.
Google has provided a display of impressive looking video clips, which also includes A sailor in the sea And one Classical violinistIt is very difficult to tell you that these videos are AI, and even then it is not always possible.
But they are the end results. What about the construction process? If you pay Google for an AI membership, you can produce your own 3 videos, and there are some ways to go about it, which I will reach here.
All this requires keeping AI video construction in reference every time: there are question marks around technology in terms of energy use, copyright violations, threat to creative industries and misinformation spread, all of which we have written on a large scale first.
Video making with VO3
If you pay $ 20 per month for Google AI Pro Plan, you get three VEO 3 video generations per day in the Gemini app, using a sharp, low quality Veo 3 fast model. If you have grown up with a $ 250-maheen Google Alla Ultra Plan, you get a “highest range” for full Veo 3 access-Google does not determine it at all, so there can be no difficult roofs, and it can lead to well-downwards depending on demand. Each video is fixed for eight seconds longer.
If you are using flow and whisk tools for your video production instead of Gemini Chatbot, it is a bit different: you get 1,000 AI credits per month and 12,500 credits on ultra plan per month. A standard Veo 3 video will return you 100 credits, and a VEO 3 fast video is going to spend you 20 credit – and in these tools, the resolution can be resolved on 1080p (if you are using the Gemini app then it is 720p).
Making videos in Mithun app.
Credit: Lifehacker
According to google Josh woodward, Accession, It seems that some textures, light and details are not as good. Inside the flow app, there is a label that is saying that the quality of VO3 Fast VO3 is fifth.
To start making videos, if you are a Google AI Pro subscriber like me, you need to head Gemini app On the web (mobile video construction is limited to Google Ai Ultra customers for now). Click the model picker in the top left corner, then choose 2.5 Pro (Preview) Or whatever the latest model is till the time you are reading it: you can choose then Video In the text input box and you are ready to do some promotion.
First, I used VO2 to try and make it again Old Sony TV AdvertisementWhere thousands of colorful bouncing balls are thrown on the streets of San Francisco. Result Were not greatSo I gave the same challenge to VO3. As you can see below, I got a better video back. It can pass almost as that was filmed in real life (the sun is great through trees), but it still ignores most of my quick instructions, and is not as good as Sony’s advertisement.
This brings us back to the nature of liberal AI, which is to copy what has been seen earlier. I am speculating that VO3 has been trained on a lot of and lots of street interviews Vax pops, not many advertisements where the bouncy balls are cascaded under the hills. It also throws light on the fact that you can take a lot to get what you want, and during my AI video tests, obtaining equipment to follow the signals is an ongoing challenge.
With only two VO3 generations for the day, I assigned the VO3 to the classic “Welcome to Jurassic Park” scene again In Spielberg’s filmThen, it’s better than it VO2 effortBut there are problems with quick adherence, and there are many fossilists. Dinosaur (and dinosaur sound) is done well, though.
What do you think so far?
Using flow to make long movies
Google also provides Slow For animation and Flow For long -term movie projects, using the same AI model as you will find in the Gemini app, according to the plan you are subscribed. If you do not want to make your own videos, or you are a free Google Gemini user, you can see what other people have made through Flow TV.
Once you get into the flow in your web browser, click new project To start Then you can start promoting, using the Settings button on the top right of the prompt box to choose the model you want to use – you will see how many credits the generation is going to use, before you render.
Making videos in flow.
Credit: Lifehacker
I decided to publish a jurassic park scene completely preparing and making my AI film out of the field to publish a proper, full VO3 clip on 100 credits. I added some dialogues with the prompt in a little more detail, and what came from the other end was equal with the VO2 output. You really get two generations, between which you can see Here And Here,
Then, we have common problems, in which the AI generator does not really know what it is going to do here, or what is seen in other videos to build a scene beyond. Our fearless courageous people are looking in the wrong direction, when one of them will look at the “wow … you” line, and everything from dinosaurs to trees looks normal.
The difference with flow and expanded video is that you can click Add to the scene Start construction on any of these generated videos and for a long time, made of a piece of eight-second. The scenes can then be extended and arranged as required, moved from one clip to another with the same character and environment.
I try to try John hamond Didn’t really work to enter the scene. The original characters remained quite well in place, but our new character appeared outside from anywhere and all the sound was suddenly cut because the flow somehow turned me back to VO2. We also found a freak camera shake half way. It is clear that I am going to be able to switch to AI film direction from technical journalism anytime, especially with just 1,000 credits per month.
The Veo 3 is still in an early stage, and Google has put the “experimental” label on all and the flow interface. However, at this time you are going to spend a lot of credit and there is something to work on a long time that is consistent and realistic. It is likely that hours of efforts and test run polish AI video you watched by populating your social media feed.
