What is the AI video creator feature?
The Colossyan AI Video Creator is the best way to create a draft from a prompt or document
(and preferably, both at the same time!).
Essentially, it will use the content and prompt that you provide, interpret it, and generate a draft.
For our Enterprise customers, brand kits support the AI video creator, both using Document-to-video and Prompt-to-video.
We offer creators the ability to start with just a prompt, instead of a document.
Document-to-video works well with other Colossyan features like interactive video and conversation scenes.
The AI video creator has two features included:
⌨️ Prompt-to-video: create complete videos out of just a few words or a sentence!
Your prompt should be at least 25 characters.
📄 Document-to-Video is a way to create high-quality videos using your existing documents.
It can be done in seconds, with very little editing required.
Using the feature, you can upload a document, add a few specifications, and select a template. After this, sit back for a moment, and a draft will be generated. Please note that the language of the created draft will match the language of the uploaded document, but you can add other language variants before generating.
How do I use the AI video creator?
Step 1
Select the "AI video creator" button from the home screen
Step 2
In the case of ⌨️ Prompt-to-video:
Write your prompt, which is at least 25 characters.
Set the topic, the tone, the language, and any extra details you prefer with this feature.
For the 📄 Document-to-Video feature:
- Click on the "Upload document" button. We currently support DOCX, TXT, and PDF.
- Give an additional prompt
- Select your desired video duration
- Decide if you want a standard video or an interactive video. If you opt for interactive options, you can have the AI Video Creator add:
- Branching, to create a more dynamic, choice-driven experience
- Quizzes, to test the viewer's knowledge
- Both branching and quizzes
After these steps, click on "Continue" and select a template on the next screen:
For our enterprise customers, you can also apply your brand kit to any of these templates at this point.
Step 3
Hang tight for a moment. Your draft is being generated. This takes between 30-60 seconds normally.
Step 4
🎬 Start editing! You should now have a pre-made draft, with a voiceover, new text and media items, and pre-placed animation markers in the script.
Give it a preview, make any changes you need, and generate your video!
Best practices for the Colossyan AI Video Creator
Provide a document AND a prompt
The best way to think of the document you provide is as ‘context’ for the AI to use as ‘core information’ or as a ‘knowledge base’, while the prompt is the ‘direction’ or ‘instruction’. Providing both is by far and away the best way to get the most out of the Colossyan AI Video Creator.
Through extensive testing, we have found that the prompt, coupled with the uploaded document, creates much better results than just a prompt or just an uploaded document.
A detailed prompt
Take a few minutes to focus on the following things:
- Who the video is for
- The style of the video
- The core concepts that are critical for the video to discuss
| Examples | Bad prompt ❌ | Good prompt ✅ |
|---|---|---|
| Example 1 | No prompt (prompt field left empty!) | I need a short explainer video to help my colleagues in the sales team understand our new sales methodology. It is critical that all core topics covered in the provided context are included, including topics X, Y, and Z. |
| Example 2 | I want a tutorial. | Create a tutorial that explains the core updates to our product to help new and existing customers use the new feature X. It’s critical that the tutorial covers A, B, and C. Ignore X, Y, and Z that were also included in the provided context. |
How will this feature be improved in the future?
We are currently reviewing all the following and are aiming to update the AI Video Creator with these capabilities in future updates. This list is based largely on community feedback, and we are very excited about how we can improve this powerful feature in the upcoming months.
Script to video
Currently, if you write your own script, it will be treated as context to be interpreted, and it will not use your script exactly as it has been written. Script to video will allow you to write your own script and have it imported as is, while we handle the draft creation.
URL to video and more context options
Currently, if you supply a URL and ask the model to use it as context, it will not work as intended. With this update, you’ll be able to provide a URL instead of a document, and we’ll generate a video based on the prompt you give and the contents of the website. We also want to allow users to upload several documents at once, to provide even more context to the model.
Using the media you’ve provided
Say you’ve included screenshots or certain media within the context document you’ve provided. We are not able to use that media in the draft that is generated. There are significant challenges here for us to figure out, but the plan is to assign semantic value to each image or video included in the upload, and then have the video creator dynamically place these into a scene that will work with them.
Specifying visual design
For instance, if you ask the model not to include avatars or to use a certain voice, this isn’t something that will work right now. The model uses the specific scenes that are in a template and will always do so currently. We are planning on overhauling how templates are handled, to give creators more control over scene types and scene design, as well as give you more options to specify things like if you want avatars included and which avatar and voice you want to use.
Custom templates
Currently, you cannot create your own template for the AI Video Creator to use. Our plans for specifying visual design (see above) are linked with custom templates, and we will ensure that creators are able to have complete control over the outcome of the prompting flow.
Storyboard
We’ll be adding a detailed storyboard into the flow, which will allow you to customise the scene types, scripts and scene layouts before you get into the editor, giving you more control than ever!
If you have any ideas about how the feature could be improved, please let us know.