Virtual Assisstant

Please login or register as jobseeker to apply for this job.

TYPE OF WORK

Any

SALARY

250

HOURS PER WEEK

50

DATE POSTED

Jan 12, 2025

JOB OVERVIEW

This is a long-term opportunity (can be part-time as well as full-time depending on your time availability and needs) for many months (and hopefully years) to come, so that's why we require you to do a SAMPLE video that you will submit to us as a response to this Job post. The info about the sample can be found under the guidelines.

We will need you to:

1) Generate a script based on a topic WE ASSIGN YOU with a set of prompts via Claude AI (will give you access to the Premium version). You can check out this video tutorial on how to do it:

- https://www.loom.com/share/a3c7e33b873c40eaa4d0af823215e194?sid=9392a131-1646-4b76-96a1-14e6bc2c3c52

And here is the copied/pasted conversation we had with Claude AI, so you can use the prompts and don't have to input them manually:

- https://docs.google.com/document/d/1YRUErjHihrf1RqPAu5dagITsG-BE0Finy4RohgQ9tXg/edit?usp=sharing

Btw, keep in mind that there is no formatting in the Google Docs above, so you might need some time to differentiate between mine and Claude's responses.

It is VERY IMPORTANT that after the initial INTRO (first paragraph that lasts for 20-40 seconds) you input this sentence:

"Before we jump back in, tell us where you're tuning in from, and if this story touches you, make sure you're subscribed—because tomorrow, I've saved something extra special for you!"

And also, we want to put one more CTA (=call to action) somewhere around 1,500 words or so into the script (after the first 10 minutes of the video roughly) that goes like this:

"Preparing and narrating this story took us a lot of time, so if you are enjoying it, subscribe to our channel, it means a lot to us! Now back to the story."

It helps with the engagement and brings back returning viewership.

One last CTA (=call to action) that we will implement is for the end of the script. After the complete end of the story, you will copy/paste this:

"Up next, you’ve got two more standout stories right on your screen. If this one hit the mark, you won’t want to pass these up. Just click and check them out! And don't forget to subscribe and turn on the notification bell, so you don't miss any upload from us!"

You will need an email to log in to Claude AI premium since the free version has limitations when it comes to word count output. We will give it to you along with log in credentials to NordVPN Premium.

2) Generate audio through ElevenLabs or CapCut website version with voice over called "Wright - authoritative and deep" (this goes for ElevenLabs, for CapCut browse version you can use Voice Actor called "Peter). Use DEFAULT settings for the Voice. You can check a simple step-by-step video guide here in this tutorial:

https://drive.google.com/file/d/1hiVKdJamWjGOXOPIN8vyKwPKgXgUHluN/view?usp=sharing

- You will need a premium NordVPN log in credentials which we will give to you.

3) Generate images (around 15-20 images for the 1st minute, each image will be on the screen for roughly 3-5 seconds and then after the first 60 seconds, 1 image for the rest of the video) via LTX Studio Free trial account thanks to the script you have:

https://ltx.studio

Check out a quick tutorial on how to do it here:

https://www.loom.com/share/a08c2f03c48944d4a4846e265f4e9690?sid=d13dcda1-c5d2-4a49-a09b-46f81fa1ecd4

The video should look something like this (at least the Intro), where we want the first 1-2 minutes to be switching between images each lasting around 3-5 seconds and being zoomed in and out and also with the overlay effect.

- https://www.youtube.com/watch?v=kFFsStOukfc&start=24

You can create the zoom in and zoom out + overlay effects like this:

Part 1:

https://www.loom.com/share/6bf225b825224c319c0a4d6159dee6a5?sid=db08cfd2-b34a-4f70-ae4c-45f62ca3fc29

Part 2:

https://www.loom.com/share/4b74e25cf5684c169bc6e3bea6872bc5

And you can download this overlay dust effect to be used in the video:

https://cutestockfootage.com/stock-footage/11862/huge-dust-particles-overlay/

This is a tutorial in Adobe Premiere, but the strategy is same in CapCut and in any other editing software,

4) Put the images along with the generated voice over audio in proper order in CapCut, Adobe Premiere or any other fast-rendering software and add the automatic subtitles generation and add some effect layer (preferably the Dust Effect shared above) and automatic zoom in and zoom out effect. Then render.

5) Once it's rendered - Upload everything on YT Studio with title same as assigned and thumbnail the one you generated. Here is a tutorial on how to do it:

https://www.loom.com/share/b493c62b9ec3492b84a170002253d768?sid=ad945792-7e92-4173-962a-b96130860b39

Make sure you properly generate the image via Mid-Journey or LTX Studio and then edit it in Adobe Photoshop according to this short tutorial:

https://youtu.be/J_IOtqukBMA

I found another very helpful and free AI-generation image tool, possibly even better than Mid-Journey:

https://ideogram.ai

We want to use these 2 colors for circles/arrows:

- Yellow code = #feff00 (text has to be strictly yellow with black edges as you can see on inspirational channels)
- Red code = #fd0100 (circles and arrows can be red and sometimes even yellow)

One helpful tip: If you are struggling generating a high-quality and catchy thumbnail with your prompts, then ask Chat-GPT to generate a prompt for you that shows what you want according to the software that you will be using (either Mid-Journey, Freepik AI or LTX Studio).

Make sure you paste the first few (5-8) paragraphs from the script in the Video description. This helps with the SEO and general video performance.

Also, from now onwards, we want to mark every video "YES" under the ALTERED CONTENT within the channel upload. You will find it by scrolling lower when uploading it.

One important thing I didn't mentioned there, because that channel isn't monetized is to include advertisement for every 1 minute of the video. Meaning that if video is 80 minutes long, it will have 80 ad slots (at 01:00, 02:00, 03:00, etc.). You can automate this process by this smart browser extension:

https://chromewebstore.google.com/detail/nexlev-youtube-research-a/ddlgficaecdmmddmjmnhcmlidjokcijd

It will allow you to select the timeframe between the ads and then automatically inserting the advertisements in the video, saving you time and making the process more efficient.

Keep in mind that for the first 5 videos, we will work more closely together and make sure the thumbnails and everything is approved. After that, the work will become more automated including topic generations (we will eventually show you how to do that as well).

IMPORTANT TO READ - MOST FREQUENT MISTAKES:

- make sure you pick the right Voice Actor in ElevenLabs from the Voice Library (his name is "Wright - authoritative and deep"), not the one mentioned in the Loom tutorial which was mentioned just for testing purposes and general tutorial steps

- when hit with "unusual activity detected" in ElevenLabs, make sure that you set your browser to clear cookies & history on exit, reconnect to a different P2P option within NordVPN and then open the browser again (it needs to be first closed, the VPN re-connected and only after that opened)

- when hit with "Please use a valid, non-disposable email address" in ElevenLabs, just Google "temporary email address" and go for a different provider, one of those will surely work out

- make sure the last image that's on the screen for the rest of the video is zooming in and out properly (not as fast as the images in the Intro though)

- if you get limited within Claude AI, then it's because their systems are overused and you need to wait till the time it shows you and start FROM SCRATCH, because Claude AI isn't able to continue with the script where it left before.

- it's also beneficial if you have multiple browsers. For example 1 browser that's dedicated separately for Voice over generating (in order to not clear history and cookies on your important browser) and other browser for generating scripts, etc.

- make sure you are saving all the materials to your PC drive, so whenever we need it, you can just let us know

- we want you to save up all the scripts, images used within the video and thumbnails (also thumbnails without the arrows/circles and texts). There is no need to save voice over or videos itself (but once you load them up to YT Studio, however, keep videos on your device for at least 1 month after posting in case of channel termination.


INFO ABOUT THE SAMPLE TASK THAT'S NECESSARY TO BE HIRED:

Topic:

Bikers Mess with the WRONG Soldier on Veterans Day

Inspirational video:

Just pick any video from this channel:

https://www.youtube.com/@StarZemStories/videos

Ideally a video with the closest topic. Download the videos text transcript which you will paste into Claude AI and ask him to create the new story above in parts according to the requirements. I think it's best that you ask Claude AI to split it into 12-15 parts because of the word count limit and the target of 12-15K words per script/video.

Normally, you will paste it into Claude AI, but since Claude AI free account is very limited in terms of character output, you will simply use Chat-GPT for the sample task.

For generating the voice over audio via ElevenLabs with multiple Free Trial accounts, you will instead of NordVPN use multiple Google Chrome Anonymous Tabs, and make sure you close them once you utilize all the free characters on one of the accounts. This should work, at least for the trial video.

Once you have the trial video, upload it on Google Drive, share the link with us and MAKE SURE THAT PEOPLE WITH LINK CAN VIEW IT.

SKILL REQUIREMENT
VIEW OTHER JOB POSTS FROM:
SHARE THIS POST
facebook linkedin