Log in to follow creators, like videos, and view comments.


Create effects

See more
© 2023 TikTok




Documentary filmmakers & multimedia storytellers

AI Video experiment | Week 5 | 18th May 2023 | AI software used: Deepbrain

This will be our final output for these AI video experiments as we feel 5 outputs is enough to show the breadth of what the AI software available to us can do at this stage. It's also becoming tricky to find software that doesn't offer a similar service at this point.

As a reminder this was an opportunity to be creative and to have licence to experiment with AI programs available to the public. We can learn what tools are best to use, which programs work for specific effects desired. We wanted to understand this new technology and how we can best use it for our clients. The task we set ourselves was to create multiple videos from a singular script each week using different AI tools available.

Overall each software had its strengths and weaknesses. Where they fail is when they try to offer all of the services! The programs that worked the best are the ones that focussed on their strengths. These would become part of an already existing editing workflow. For example you could start in Chat GPT to form your script, then bring that into Steve.ai to create a short and snappy animation, then over to Murf.ai for the Voice over. And bring it all together in something with a bit more manual, human control like premiere. The main downfall of these programs is the un-adaptability to tone/branding, often basing the content from templates, it gets really fiddly and limiting to try and input your own style which is so important for consistent engaging content. That’s something we’d like to see more of, a program that gets to know your company and can output/suggest accordingly, we think that’s when we’ll see a big change in the industry for video editors.

We thank you for staying tuned with us over the last five weeks and welcome any feedback or questions you may have on any of the videos we’ve published.
AI Video experiment | Week 4 | 9th May 2023 | AI software used: Murf.ai

The main takeaway is that this is a great program to get a quick V.O sorted but would recommend another software to edit your visuals with.

- The text-to-v.o function works really well! The range of voices is great, giving us many more options on tone of voice, style of voice, accents, gender, age, etc - overall I’m impressed with this function. The two real voices we chose for the final output are much more emotional and much less robotic than the other programs have allowed. The effect is as though Fred and I were narrating the video ourselves giving it a personal feel.
- The visuals are very static, it only allows a drag and drop which doesn’t allow for a lot of dynamism which is much needed for an output like this.
- There are too few images being used by the program so it feels slow and clunky.
- The images used are not illustrative of what’s being said.
- Tonally, the music doesn’t suit the story being told.

#AI #ArtificialIntelligence #Experiment #MurfAI #AISoftware
AI Video experiment | Week 2 | 25th April 2023 | AI software used: Pictory

The issues I have with Pictory are very similar to last week’s software used (InVideo) although I do rate Pictory slightly higher because it takes less time to create a video which is better for the company. That said, the limitations of the software that make this time saving possible means the level of creative input from our end is vastly reduced.

- After the achingly long first output, we asked the prog to keep the video under 50 seconds which has improved the pacing but is still
#by no means perfect. Still, it’s nice to have this function.
- Unfortunately, there is no option to add atmos or SFX.
- It is possible to add music but only from their own stock libraries which are very corporate and don't have many options.
- There’s no eye for how the images sit together visually to tell a story.
- Colour and movement has not been taken into account either.
- The VO lacks personality. There is an option to add our own VO but there’s less options to time assets as there's no timeline view in Pictory which could get very fiddly unless you recorded the audio to match the video after creating it.
- As Charlie mentions the auto-suggested footage is better/more on point than invideo's but it still doesn’t seem to understand dates and the chronology of the timeline - and unfortunately this software doesn't even give you an option to tag. You can only upload images and drag and drop them.
- A positive of this software is that it does allow you to drag the images within the frame so that all information is included and no information is lost in a crop.
- The text on screen is slightly more exciting than InVideo’s due to the two-tone colour it automatically generated. But it's not responsive to the images/footage in the background so you have to change each as you go to make sure the overlaps look correct.

#aisoftware #ArtificialIntelligence #Experiment #Pictory #AISoftware
Fat Rat's AI Video Experiment | Week 1 | 18th April 2023 | AI software used: InVideo

*Drum roll please* Here is our first foray into the AI experiments.

Although suitably impressed with this final output, it's not content which would pass our quality controls. We've gathered together further thoughts below:

- The pacing isn't right. From the speed of the robotic VO to the choice of the number of images used to illustrate points within the script makes it feel sluggish and quite frankly boring.
- Bridging between one section to another is lazy.
- The tone and speed of the VO is entirely wrong for the style of the output. It should be fun, pre-recorded VO but this is much more time consuming and doesn't fall within the parameters of the rules we set ourselves.
- It is humourless. Despite the script being in 'our voice' the software doesn't seem to understand where to pause for effect or to add a cheekier tone of voice.
- The software doesn't seem to understand dates, countries and nuance resulting in the wrong images being used to illustrate specific times and places and unfortunately this software doesn't even give you an option to tag.
- Some images have been cropped meaning the main information within that image has been cut out.
- The text on screen is good but again far too slow for a quick output like this. I feel this video could be 30 seconds or less if treated right.

Whether you're a filmmaker or a client, how do you think AI-generated video might impact the creative industry? Let's hear your thoughts below.

#AI #ArtificialIntelligence #AISoftware #Experiment #invideo
Get TikTok App