Clipmaker FAQ

How do I generate a video?

You need a script prepared for bot input. A prepared script has non-voiceover content between curly brackets. For example:

{{INTRO}}
{{Insert Clip 0:39 to 0:42 in the background}} Polo G is in big trouble, and he's giving up his own mom to the feds to get out of it.
            

To generate a video:

1. Navigate to the Video dashboard in Clipmaker.
2. Click the "plus" icon in the top left corner.
3. Enter a name for the script.
4. Copy the whole script including info in curly brackets and paste to the field on the left.
5. Select a voice to generate the voiceover (for rap channels the voice name is San, as of June 2024).
6. Click "Submit".
7. On the Video dashboard, click the "gears" icon on the new card to start generating the script.

You can track its generation on the progress bar at the bottom of your screen.

The visualization of the instruction is available in the loom created by our team.

What actions should I take before creating a video to make sure it's good quality?

Quality of the bot's output is closely tied with the database. If you want guaranteed quality, it's highly recommended to take the following steps prior to video generation:

1. Look through your script and detect its main topics and characters.
2. Check the faces database and the names database. Searching the face database for a specific person will show you how many times a person is mentioned.
3. If a character in your script is not mentioned at all, add them to the database of names (aliases) and faces. Then, upload at least some videos about the character via the interface for video upload.
4. If a person is mentioned less than 200 times, it's highly likely that the quality of the output will not be sufficient. Add more videos about this person to the database.
5. We recommend achieving at least 400 mentions of a character in the database to have a helpful number of relevant footage.

How do I download a file with the Premier timeline footage info (a .json file)?

Click the "diskette" icon on a generated script card on the Videos Dashboard tab. The download starts right after your click. The file has the following name format: script-XXX.json where XXX is your script's serial number.

Can I download the voice track (voiceover) separately?

Yes, you can. Click the "details" icon on a generated script card to access the selected script tab. Now, click the "music" icon to download the voice track.

How can I track the status of my video?

There are 4 statuses that your uploaded script can take:

• new - newly added video script.
• in process - script is being processed (relevant footage search and collection)
• done - video generation is complete, voiceover and .json file are ready to download.
• error - an error occurred while generating the script.

The status can be seen when you hover your cursor over the indicator, and is displayed in the video card that is accessible via the "details" icon. The color of an indicator also gives a hint about the status: empty stands for "new", blinking green stands for "in process", green for "done" and red for "error".

Can I see a preview of a generated video?

Yes, you can. The preview option is available only for the videos that were compiled. Only administrators have access to video compilation. If you have administrator rights, you can compile a video by clicking the "camera" icon and confirming compilation in a pop-up window. The video status changes to "in progress" and you will see a progress bar at the bottom of the page. When the compilation is ready, the video preview becomes accessible in a built-in player on the script page that can be accessed by clicking the "details" icon.

How long does it take to generate a video?

Generation of a 25-minute video takes 50 - 90 minutes depending on a topic and contents of the database.

Can I generate more than 1 video at once?

Yes, the tool can handle up to 5 concurrent generations. The scripts that are added above this limit are waiting in a queue. The tool is capable of handling more than 5 concurrent generations, but the current limit of 5 is tested and proven to be stable for the system.

Can I generate one video in batches?

It is possible but not recommended. This approach can lead to more duplicates and affect uniformity of a generated voiceover.

Is there any difference between the 3 video tracks (timelines) in Premiere?

The first and second video tracks are unique, footage never repeats within these two. This algorithmic rule is not applied to the third track, so videos that appear there can appear further on any of the three tracks.

How is Elevenlabs voiceover generated by the bot preprocessed?

The voiceover is preprocessed by automatically removing pauses (silence) between words and sentences. (This is subject to change in future releases).

What kind of video can I add to the database?

The database supports 2 basic types of videos: faceless videos, and news reports. You need to specify the video type in the upload interface.

What kind of video can't I add to the database?

It is not advised to add too long videos to the database, longer interviews, music videos, or face-centered videos (narrative videos that feature the creator's face, making their presence a key element of the content).

How do I know the video was successfully added to the database?

The video was successfully added to the database if it doesn't show up in the queue (i.e. has no error and is not processing).

How long does it take the algorithm to add a video to the database?

For a standard faceless video of 25 min length, the procedure takes around 90 minutes. One video processing unit can process up to 5 videos concurrently. Still, the time may vary depending on a multitude of factors. It is necessary to add relevant videos in advance.

What does the "priority" parameter mean in the video adding interface?

The "priority" parameter impacts video processing order in the queue. The default priority is "medium". If set on "high", the video will be taken into processing by a first available processing unit. The "priority" parameter makes a difference if the video processing queue is full (more than 5 videos being processed concurrently).

What is the "type" parameter?

The "type" refers to a predefined set of features of a video that is being uploaded to the database. It affects the way scenes from the video are being processed. Currently, the database supports two video types: faceless videos, and news reports.