Honor’s new Image to Video service, powered by Google, may change the way we view photographs forever
Can we actually believe anything we see from now on?
The video clip below is of the launch of the Honor 400 5G and Honor 400 Pro 5G devices in London last week.
Except it isn’t.
The video is entirely AI.
The person is real, but the obligatory on-stage presentation of new devices to an audience was entirely AI-generated.
Sure, the presenter did turn and show off the devices much like you can see above, but I didn’t actually capture this moment.
I was only in camera mode at the time (using an Honor 400 Pro 5G), and only got to take a static photo – which you can see below.
Everything you saw in the video is down to the Image to Video feature, produced in conjunction with Google (using Google’s Veo 2 model), and exclusive to Honor’s new devices for a two-month period.
The feature can be used to bring almost anything to life and will be seen by many as a game changer for AI.
It did a fantastic job of creating a video that anyone would believe to be genuine, and it’s one of many videos I’ve created – some of which you can see below.
The service is limited to ten videos created per day, for a period of two months. After that time it’s not clear if there will still be an ability to create some videos for free, or whether there will be a need for a subscription.
It’s possible that Google might opt to include an allowance for Google One account holders, but pricing will be revealed in the future. As will information on what other devices will get access to the feature (it is almost certainly going to be added to other Honor devices as the heavy lifting is done in the cloud, not on the phone itself).
Ten videos per day should be more than enough for most people – especially when the novelty invariably wears off in the future, and users will more likely settle down to creating only a handful of videos every month.
Never the same video twice
What you need to be aware of is that there’s no way to direct the service to create any particular video. It isn’t a replacement to commercial AI software that can be given complex instructions – but rather a way for you to quickly create results.
The AI service analyses your image and then determines what video to produce, and every time you run it on the same image you’ll get a totally different result.
A demonstrator at the launch event showed me a photo of a scene where the first creation was of someone moving around, the second of a dog jumping into the frame, and the third of a weird hand coming out to cover everyone.
Some customers have reported that they don’t even have the feature on their purchased device yet. O2 users in particular appear to be still waiting.
The firmware release that includes Image to Video (and other AI features) is 9.0.0.121 and should be pushed to all devices soon. You can also try and speed up the process by going into Settings/System & updates/Software update, then ‘Tryout version’ in the top-right menu.
A question of ethics?
There’s currently a lot of concern over generative AI, with a debate flaring up just this week over companies being made to ask for permission to use works created by individuals and how it will impact AI companies if they have to do so, rather than just scrape the Internet for material.
My opinion is that everyone should retain copyright on their own generated material unless they implicitly give permission to share it for use by any AI service.
In many cases people will willingly give permission anyway, if it allows them access to extra features.
It must be the choice of the individual, not a company that simply wants free rein to do as it pleases, with the intention of taking existing content to create new content that it can monetise itself.
We’ve likely all seen content on social media and YouTube with AI generated voices, and more recently, AI generated presenters that could easily lead to roles like influencers being rendered obsolete.
Why use a real person when AI can do all the heavy-lifting and promote any product or service with minimal effort (and nobody that needs to be paid, or might have boundaries they’re not willing to cross)?
Google is promising that all of the generated videos with Image to Video will be watermarked to make it clear that the video was AI generated, and there will be a process to block certain images from being turned into video (I am not sure yet how this will work, but assume it will be similar to requesting to have your home removed from Google Maps, or other personal data removed from the Internet).
Google also employs other processes to ensure videos can not be created from photos of certain things, such as any scene containing nudity.
Users trying to make a video of ineligible photos will be told a video cannot be made, although it won’t necessarily give a reason as to why.
Have fun with it!
Image to Video is a great fun feature, that’s sure to create plenty of fantastic shareable content, and it’s reassuring that Google is putting some some guard rails in place.
I look forward to experimenting more with the feature for my full handset review, and of course I’ll be sharing the results on here.
Finally, if you have bought either phone and can use the service, please do share your creations in my shared Google Photos album (see link below) so we can all enjoy them!
Disclosure: The featured image was AI generated using WordPress’ own AI tools. All of the article text was written by myself with no AI assistance at all because I actually like to write using my own words!