SYSTEM NOTICE

Auto translation by AI. Be sure, accuracy, nuances and authorial intent may not be fully reflected.
見出し画像

I tried out Pollo Agent! Being able to create videos while chatting is pretty amazing | Pollo AI

Introduction


Hello, this is Nekoko🐾

This time, I'll be writing about my impressions of using the new feature from Pollo AI, Pollo Agent.

To start with the conclusion,
this is quite interesting.

Up until now, AI video generation has mostly followed a flow like
"first think of a prompt"
or "think of the structure yourself," but

Pollo Agent felt like it
helps you decide the direction of the video while chatting with you.

What's more, it doesn't just make small talk; it helps with:

  • What length it should be

  • What kind of atmosphere it should have

  • What videos might serve as a reference

  • What kind of scene structure would be good

Since it naturally builds these things up,
it seems like a great fit for people who think, "I want to make a video, but the first step feels heavy..."

👉 Pollo AI Official Website


What I tried this time

This time, I tried generating a video using Pollo Agent . Select AI Agent from the side menu👇


AI Agent icon


Then, an input screen for the agent will appear👇

AI Agent window

First, upload an image and start interacting with Pollo Agent. Then, instead of writing a long prompt right away, it starts by asking

  • what kind of atmosphere the video should have

  • You can choose

what kind of video style you want.

When you press Auto in the AI agent window👇

  • Auto

  • Viral Video

  • UGC Video Ad

  • News Video

  • Music Video

  • Story Video

can be selected😊 Also,
Style can also be selected from below😊

  • Auto

  • Realism

  • Ghibli

  • Ukiyo-e

  • Cyberpunk

  • Pixel Art

At this point, I thought it was quite easy to use😊

With normal video generation, you often have to decide and write everything yourself from the start, so being able to proceed while chatting like this is quite easy.

Pollo Agent I tried out Viral Video

This time, I tried creating a video by setting the Pollo Agent window settings to "Viral Video" and "Realism"😊

It properly understands the content of the image

The image I uploaded this time was of
a ginger cat wearing a chef's hat, cutting a cabbage.

Pollo Agent looked at that image and...

  • ginger cat

  • chef's hat

  • kitchen knife

  • cabbage

  • realistic and humorous atmosphere

It correctly interpreted the content, just like that.

This is the first amazing thing.

Even without me explaining in detail,
I felt like it really understood 'what is funny about this image'.

All you have to input into the chat at the start is the reference image, video type, style, and simple instructions😊
This time, in line with the reference image




すごく上手くキャベツの千切りをするネコのコックさんの動画

It even finds reference videos for you

What was even more interesting was that
Pollo Agent found popular videos that could serve as references.

It doesn't just make a video,

'Videos in this direction seem likely to be popular'
'This way of presenting it seems interesting'

Because it suggests things like that,
it felt more like an AI that thinks through video planning with you.

I think this is a major difference from standard video generation.

It even creates the scene composition for you as is

After finding a reference video,
it then used that vibe to create a scene composition
that fits this image.

In this case, it used a cat cooking scene as a base,

  • the initial highlight

  • memorable movements

  • the final wrap-up

and organized a flow like that into a form that works well as a video.

In other words,

the flow of: provide image → talk → find references → create composition → generate video

connects quite naturally.

I felt that this ability to handle the "planning part in the middle" is the strength ofPollo Agent .

The video that was actually created was also quite good.

And, regarding the crucial generation result,
I was able to create a video that was very close to what I expected👇

This time, it turned out to be exactly the kind of video I wanted to see, and I think it clearly demonstrated the convenience of
Pollo Agent .

Of course, I don't think it will always result in a perfect outcome every time.
But at least in this experience,

  • when you're unsure of what to make

  • when it's difficult to construct prompts yourself

  • when you want to explore the direction of a video that is likely to go viral

I felt it was quite helpful in these situations.

"No prompt required" is truly close

People often say "no prompt required," and
Pollo Agent is quite close to that.

Of course, it doesn't mean you don't have to think about anything at all.
But you don't have to write everything out yourself from scratch,

  • Something like this

  • I'd like it to be short

  • I'd like it to be more realistic

  • I want to use this as a base

The fact that you can start with just that kind of conversation is a big deal.

I thought this ease of entry was quite appealing.

While waiting for the video to finish,

I thought of the next variation in the meantime💡

Pollo Agent also makes suggestions like this😊

🐱 Next variation ideas you can create:

Speed-up version — The knife skills become even faster, and the BGM tempo increases. Maximum viral potential🔥

Full cooking version (longer, 30 seconds) — Chef Cat shows the full process from julienning to stir-frying to plating

Rival appearance version — A human chef and a cat chef side-by-side in a julienning showdown! Includes a shocking ending😂

Even at this time, you can just chat with 'Speed-up version' and it will generate the video👇

The speed didn't really change much😭
But I was able to make a different version of the video!

I also tried a story video as an addition

This is also quite interesting!

Pollo Agent using
I tried one more as an addition.
This time, it's a story video.

This time, without uploading an image,

'In Japan, it's the season for graduation and school entrance. Please create a story video based on students graduating and then entering school.'

I asked it in this way.

Then, instead of making a video right from the start,
it first proposed the direction of the story.

This part was interesting first of all😊
For viral videos, it felt like building based on a 'template likely to grow,' but
for story videos, it felt like it thought about the flow of emotions and the structure of the story first.

It also handles mid-process revisions through conversation

In the middle of our interaction, when I

'In Japan, middle school is 3 years and high school is 3 years'

added this supplement,
it properly reflected that content and adjusted the settings.

This aspect is great because it's not just about generating, but
being able to adjust the direction while chatting.

It wasn't just using the initial proposal as is,
but when I thought 'that's a bit different,' it felt like it properly understood the human's intent.

It's reassuring to proceed while checking properly

After that, before generating the video,

  • video length

  • aspect ratio

  • visual atmosphere

  • narration language

and so on were also confirmed.

With a story video like this one,
it feels more reassuring to have these checks in between rather than having it generated all at once.

Rather than 'realizing a completely different video was made,' it feels like
proceeding while aligning the direction little by little.

Here is the video that was actually created👇

The movement of the hand opening the door became a strange movement and it was funny.
More like a trembling hand...😆
Please take a look and see for yourself😅

There is a merit different from viral videos

The 'viral video' I tried first was,

  • It finds videos that might be useful as references

  • It aligns with formats that are likely to go viral

  • It makes it easy to create short, fast-paced videos

It felt very intuitive and convenient in that sense.

On the other hand, for this story video,

  • creating the flow of the story

  • thinking about scene composition

  • assembling it including narration and atmosphere

I was working in that direction, and I realized that even with the same
Pollo Agent the process and the suggestions vary significantly depending on the type of video you want to create.

This is what I think is interesting about Pollo Agent .

It doesn't just generate videos; the Agent's behavior changes depending on 'what you want to create.'

Impressions after actually using it

What I felt after trying these additional examples is that
Pollo Agent is not just about 'creating videos while chatting,' but also

  • thinking about how to present it like a viral video

  • thinking about the structure of a story video

  • incorporating revisions along the way

it also plays a role closer to planning.

This is a slightly different experience from standard video generation.

It's very convenient to be able to consult not just on 'wanting to make one short video,' but also on 'what form it should take.'

Points I was a little concerned about

While it was convenient, it did consume a fair amount of credits.
However, this seems to be influenced not just by the difference in features, but also by the difference in video length.

The 'viral video' I tried this time was about 10 seconds long,
while the 'story video' was about 30 seconds long,
so I felt that costs are likely to vary depending on the video being created.

That said,

  • story ideas

  • scene composition

  • narration

  • video generation

considering it handles everything up to this point, it feels reasonable.

It seems like a great fit when you want to 'get one video properly shaped first'.

Summary

Pollo Agent The most impressive thing about using it was the feeling that it wasn't just generating a video, but actually helping me build the video concept together.

  • It understands images

  • You can decide the direction through conversation

  • It finds reference videos

  • It even thinks of the scene composition

  • You can proceed directly to generation

This flow was very natural.

For those who 'aren't sure what to make' or 'often get stuck at the prompt-writing stage',
I think this is a very easy-to-use feature 😊

I'd like to try it out with some other subjects in the future!

Bonus

This time, the bonus isn't a video effect, but the new feature in Kling 3.0 that I introduced in the previous article: Motion Control.
You can upload an image of a character or person you want to animate, upload a video of the movement, and it will create a video that matches that movement.
It's a feature like pseudo-motion capture.
Since there are also presets available for the movement videos, you can easily create dance videos and more😊👇

It's just too amazing🐾

Previous article👇

Links and Related Articles

👉 Pollo AI Official Website

🐾 Please check out these related articles too!

Here is the link to the note magazine containing articles that have introduced Pollo AI so far👇

Thank you for reading this far🐾



いいなと思ったら応援しよう!

ネココ よろしければ応援お願いいたします! 絵本を描く活動費に使わせていただきます!