A job that used to need three people
To make a video ad for one product, you used to need: one person to write the script, one to stand in front of the camera, and one to edit.
For a small shop this was almost impossible, so most shops stopped at the product photo.
Reel Studio brings four steps into one place: script, scene images, voice, then video. At each step you look, change it if you do not like it, and then go to the next.
What you will need
Before you start, have these in hand:
- One photo of the product, taken in window light, on a clean background
- The name of the shop
- The offer in one line, like the price and delivery
- Who you are speaking to, in one line
People skip the last one, and it changes the quality of the script the most. "College students in Dhaka" and "a housewife in a village" need different words.
Step one: the script
If you ask for a script with these details, a script is made in Bangla: the hook for the first three seconds, then two to four scenes, what will be said in each and what text will be on the screen, and at the end what the viewer is asked to do.
Several other hooks come along too, and you can choose which one it starts with. Which hook works when is written separately in the post on hooks and ad formats.
Every box can be changed by hand. If the price is wrong, fix it, and if your shop would say it differently, write it your way. There is no hurry at this step, because the video has not started being made yet.
Step two: the presenter image
An image of the presenter is made, and it is shown to you before the video is made. Ads with a presenter have this approval step. In a scenes-only video you can make each scene's image first, for free.
This pause is on purpose. If the image is wrong, the video made from it will be wrong too, and making a video costs much more. So first the image, then approval, then the video.
If you do not like the image, cancel, and the Token for the remaining steps comes back. The cost of the image that has already been made stays, because it was really made and you saw it.
Step three: the voice
Two paths.
AI voice. Choose from a few voices. It is quick, and sounds the same every time.
Your own voice. Record it in the browser itself, 3 to 30 seconds. The presenter's lips will match your voice.
For a small shop, your own voice almost always works well, because a local accent, and the owner saying the shop's own name, sounds believable.
When recording, close the door of the room, hold the phone about a hand's span from your mouth, and keep the fan off. These three are enough.
Step four: the video
Now the whole video is made: the presenter, the talk, the music and the Bangla text on the screen.
The video has two to four scenes, that is 10 to 20 seconds. This is the right size for Facebook and Instagram reels.
Along with it you get the text for the post, a message you can send on WhatsApp, and hashtags. So you can download the video and post it right away.
About the presenter
There are six drawn faces, and they are drawn, not real people. You can choose any of them.
If you give your own photo, it has to be your own face, and you have to give consent for it separately. Making an ad with someone else's photo is not allowed here, and the consent box is exactly the way that is stopped.
Price and refund
The price is shown before making, so the cost never surprises you. There are two qualities, economy and better, and the two have different prices.
If any step fails, the full Token amount comes back, not just for that step. Keeping something after saying "we could not make it" is not an honest answer.
And if the image sits waiting for approval for 24 hours, the remaining Token goes back, and only the cost of the image that was made stays. So if you forget it, your money is not stuck.
After the video is done
One thing: the first video will not be the best, and that is normal.
What you should do is run it for a week and see how many get past the first three seconds. Then make another version for the same product, changing only the hook. Not a new product, a new hook.
You can make this new version cheaply from a finished ad with a presenter. It does not work with an ad recorded in your own voice, because the voice would be needed again.
There is a box to tell us whether the video helped, and it is used in the next script.
What kind of video really works is written separately in the post on UGC ads.
If you want to make one yourself, open the page for making an AI video from photos.
Questions and answers
How long does it take to make a video? The script takes seconds, the image a few minutes, and the full video longer than that. The work goes step by step, so you can do other things in between, but it is better to keep the tab open at the video step.
Does the Bangla text on the screen come out right? In a reel, the text is placed on top of the video and is not drawn inside the image, so the conjunct letters do not break. If you try to have an image-making model write Bangla, it breaks, and that is exactly why it is done this way.
Will it work if I do not give my own voice? Yes, there are a few ready voices. But your own voice is more believable, especially for local customers. You can try an AI voice in the first video and your own later.
Can I give someone else's photo as the presenter? No. If you give your own photo you have to give consent separately, and that is there for this reason. Making an ad with someone else's face is a problem from both the legal and the ethical side.
Will I get my money back if the video is not made? If any step fails, the full Token amount comes back. If the image sits waiting for approval for 24 hours, the rest goes back and the cost of the image that was made stays. The same rule applies if you cancel the image yourself.
How many seconds long is the video? Two to four scenes, that is 10 to 20 seconds. But if you give a recording of your own voice, the video is as long as your recording, so if you record 30 seconds, the video is 30 seconds too.
Can the same video go on both Facebook and Instagram? Yes, the size is right for both. It is better to adjust the post text a little for each, and that comes along too.
What kind of product photo is good? In window light, on a clean background, with the product filling the frame. If you give a blurry or dark photo, that weakness is carried through every later step, so it pays to spend two minutes more at this one step.
In short
- Four steps: script, presenter image, voice, then video
- The image approval step is on purpose, because a wrong image makes a wrong video
- The presenter is one of six drawn faces, or your own photo with consent
- The voice is AI or your own, recorded for 3 to 30 seconds
- The video is 10 to 20 seconds, with post text and hashtags
- The price is shown first, and the full Token amount comes back if a step fails
- The first video will not be the best, so change the hook and make a second