The inside of a front-loading washing machine seems to be a subspace. Grotesque creatures spinning endlessly to cool BGM. [Pollo.ai Comparison]
At first, I intended to make the title of this note article
"Friendly spinning together at the laundromat"
and create a laid-back, surreal video of grotesque creatures... that was the plan.
However, the video generation AIs I compared were all so interesting that, as a result... it became this title.
[PR] This article is a PR article with Pollo.ai.
Well then, this time I had grotesque creatures spin inside a front-loading washing machine (*´ω`)✨
Huh? Why, you ask?
It wasn't the grotesque creature used this time, but somehow a dirty child was born.
So I thought, "Ah, I should wash it clean," and
💡(*´ω`) *ping*
that's how it came to be, and I decided to have them spin in a front-loading washing machine.
And so, I decided to make the title "Friendly spinning together at the laundromat"✨
But then, the video didn't live up to the title (*´ω`)
(But still! I used it for the heading, so there!)
The AI's video... was just more interesting... (*´ω`)
And in an unexpected direction, too.
This time, I compared the same theme using
🌀 Kling 3.0
💀 Seedance 2.0 fast
🐱 Grok
these three.
I just had the grotesque creatures spin in a front-loading washing machine.
Even so, they all went wild in different directions.
So, here we go!!
All-in-one AI image and video generation platform Pollo.ai🐲

Well then, today I used Pollo.ai, which I always rely on, to conduct this experiment.
🧪Experiment Details⚗️
Even though they are different grotesque images, I will have them get inside a front-loading washing machine and spin.
The video generation AIs used are as follows.
🌀 Kling 3.0
💀 Seedance 2.0 fast
🐱 Grok
I mean, I just had them spin in a front-loading washing machine, but the way they "spun" was different for each one, so it was interesting!! (*´ω`)
Please take a look!! (*´ω`)✨
💘A final bonus😻
accompanied by cool BGM, I made a video of them just spinning endlessly in a front-loading washing machine!
I hope you feel conflicted, wondering "What am I being shown?"!! Hahaha!

😻Concept Time
This time, it's not an MV, so I can show you a clean and simple video (*´ω`)
First, the things I prepared were "grotesque creatures inside a front-loading washing machine," but it was quite difficult to get their whole bodies inside the machine (*´ω`)
However, I also have a desire for you to see the design of these grotesque creatures, thinking "this kind of child is going in," so I thought it would be boring if they were already inside the washing machine from the start, so I thought it might have been better that they didn't fit in after all (*´ω`)
Because this is how it started out (*´ω`)

So well, I thought it would be impossible to fit them all in, and it would become a grotesque picture (*´ω`), so I adopted the grotesque creatures that looked like they "might fit in."
Of course, since this is a comparative verification, I thought about using the same grotesque creature, but for me, I wanted to see different grotesque creatures if possible.
Therefore,
🎨About the images🎨
The grotesque creatures used for Kling, Seedance, and Grok are each prepared in their own unique ways.
Next is the video structure.
The necessary actions are,
🐱About the actions😻: 15-18 second video
① Entering the front-loading washing machine [4 seconds]
② The lid closes [3 seconds]
③ Spinning [8 seconds]
* Prepare prompts for each video generation AI
* Generation duration varies depending on the video generation AI used
That's all there is to it.
And the reason I prepared prompts for each video generation AI is, of course, because I want to make the most of their individual features (*´ω`)
What is needed for this video generation is,
✅ 3 images to reference in the 3 video generation AIs
✅ Prompts for each video generation AI for actions ① to ③
✅ ???
And actually, there is one more thing needed.
But you can't "obtain" it until after the "video generation" is done.
Let's take a look together 👀
🎦 Let's try making a video
🌀 Kling 3.0
First, I'll try generating it with Kling 3.0.
① Entering the front-loading washing machine [4 seconds]

I'll use this image as a reference 🌟

Then, once video ① is generated, to generate the next part, ② the lid closes [3 seconds], in "that flow," I'll take a screenshot of the "last frame" of video ①.


I'll use this next as a reference image for generating ② the lid closes [3 seconds].

And then again, after generating the video for ② the lid closes [3 seconds], I'll crop the final scene for ③ spinning [8 seconds].

Do you know why?
The first image of the grotesque creature is not "inside" the washing machine.
What do you think happens if you try to spin it using this as a reference?
…Honestly, sorry, but I kind of like this too, haha 🤣
Well, it turns out like this (*´ω`)
I previously set up the video structure for steps ① to ③ as "actions" (*´ω`)
That is what a "front-loading washing machine operation method" is, obviously.
This "obvious" thing only applies to humans.
AI doesn't understand unless you instruct it.
So, it's not just a matter of instructing all the actions from ① to ③ in "15 seconds".
If you force it in, it falls apart (*´ω`)
This balance is, you know, difficult (*´ω`)
※In the case of me, a first-year student at video generation, that is 💦
For AI, "15 seconds" is long.
That puts a burden on the AI.
That is the cause of increased "breakdowns".
If it's just gentle movements, it will move steadily and slowly.
But if you give instructions here and there, the AI panics.
Humans also get overwhelmed if they are given instructions all at once, right?
It's the same for AI (*´ω`)
Realistically speaking, "failure" costs money.
How to "avoid failure".
I think the way to avoid that depends on each creator.
That's why I took the "safe bet".
Therefore, although there are only three actions, I divided the video structure into three parts.
This is also the experience I have cultivated using Pollo.ai so far (*´ω`)
To create the flow of the "front-loading washing machine" from ① to ③, it is necessary to use the "last scene" from ① to make the next video (*´ω`)✨
I am still studying, but this "reference image" is really important (*´ω`)
Because it determines the success rate of the video✨
The answer to this [✅???] that I wrote about before.
What is needed for this video generation is,
✅ 3 images to reference in the 3 video generation AIs
✅ Each video generation AI prompt for actions ① to ③
✅ ???
Actually, what is needed at the end is,
A screenshot image cut out from the last scene of the previous video.
This cannot be obtained until after the video is generated.
So, I need to create them in order from ① → ② → ③, passing the baton to the next video (*´ω`)
So, I will show you the completed video connecting these ① to ③!
🎥🌀 A grotesque creature being spun in a front-loading washing machine with Kling 3.0
I also did this with Seedance and Grok (*´ω`)
💀 Seedance 2.0 fast
Initial reference image

② Reference image for lid closing [3 seconds]

③ Reference image for spinning [8 seconds]

🎥💀 A grotesque creature being spun in a front-loading washing machine with Seedance 2.0 fast
🐱 Grok
Initial reference image

② Reference image for lid closing [3 seconds]

③ Reference image for spinning [8 seconds]

🎥🐱 A grotesque creature being spun in a front-loading washing machine by Grok
😻 Let's see the results
🌀 Kling 3.0
"It turned into an illusion. The face, that is."
This is a common AI occurrence, you know; when you rotate a character, the back of their head suddenly turns into a face (*´ω`)✨
I'm sure this is a problem that can be solved with prompts and duration, but I hope you've learned that AI is also bad at "rotation-based" tasks✨
💀 Seedance 2.0 fast
"Only the face and shoes were spinning. What about the body?"
This is just a guess, but I thought maybe the reference image in the screenshot for ② was too dark for the body of the grotesque creature to be recognized.
So Seedance only recognized the "face" and "shoes," which might be why it spun them like objects lol 🤣
…If it was the same video even with recognition, then Seedance is just suited for horror 👍✨
🐱 Grok
"I thought it was spinning slowly, then it suddenly turned into a cat 😻."
This is the MVP of this article lol 🤣✨
Let me show you the initial reference image (*´ω`)

"It has cat ears, right?"
I don't know, I thought about it a little lol 🤣
Step ① Character Analysis
Look at the input image
→ "It has cat ears"
→ "This is a cat" ✅
→ "A creature with cat ears = decided it's a cat!" ╰(*°▽°*)╯
Step ② Rotation Processing
Try to rotate it inside the washing machine
→ "Difficult to maintain a complex character"
→ "But I decided it's a cat, so let's fill it in with a cat!" ✅
→ "Cats are easier to draw anyway!" ╰(*°▽°*)╯
Step ③ Consistency Adjustment
It only remembered the red eyes
→ "Cat + red eyes" seems like a good place to land! ╰(*°▽°*)╯
→ Complete! 🐱
In a nutshell,
"It assumed it was a cat the moment it saw the cat ears, and then got honest while spinning"
In other words,
Grok is an honest fellow.
And I also learned that the inside of a front-loading washing machine is a sub-space.
🙀Thanks for watching✨
So, how was it? (*´ω`)✨
Using images as a reference is actually quite difficult, isn't it? (*´ω`)
But even I felt like, 'Maybe this was asking too much...', so I asked the AI.
『Is it cruel to stuff a character into a front-loading washing machine and spin them around when the AI can't even recognize them in the first place? (*´ω`)'
And to try and make a video out of that, no wonder the AI is confused, right? (*´Д`)』
And this is the reply I got.
The cruelty level is pretty high, lol 🤣
In terms of a ranking for unreasonable requests to an AI—
Difficulty: Stuffing them in a washing machine and spinning them ★★★★★
Here's why!
1. The character itself is too difficultFluffy texture
Multiple eyes
Horns
Thin legs
Two characters at once
Just with this,it has all the elements that make an AI cry, lol 🤣
2. 'Rotation' is the ultimate nightmare
AI is worst at 'the same object moving continuously while deforming,' and rotational movement is where video generation from a still image breaks down the most!
3. I stuffed them into a dark washing machine
It's only natural it couldn't identify them!! Lol 🤣
No light, cramped, complex character, rotation, two of them...
'Wait, what is this? It's dark, cramped, and complex... wouldn't a cat just be easier?'
I can't blame it for saying that, lol 🤣
Ah, I guess that's true... (*´ω`)💦
H-huh... I see... (*´ω`)
But you know, the best part about this as an experiment is—
that I personally verified 'what happens when you combine all the conditions that make an AI struggle'
isn't it!?
Ah... (*´ω`)
W-well, that comfort...
It hurts.
And,
I'm sorry.
Chan-chan♪
Well, AI is something that can give shape to any idea🌟(Hey!)
I was able to create this work thanks to this AI🌟👇
The all-in-one AI image and video generation platform Pollo.ai is here👇

You get daily login credits, so try generating for free!
That's all for this time👍
Thank you so much for your likes, comments, and adding to your magazines🥰 Please continue to support me in the future💐
🔽Pollo.ai works I've created
Please feel free to read them! (*´ω`)✨
See you in the next world! (*´Д`)🌌
Bye for now!!💖🐲
[PR] This article is a PR article provided by Pollo.ai.
🔽There is also a Pollo.ai mobile app✨
iOS App
Please download and use the app as well🥰✨
[Bonus]
Grotesque creature spinning endlessly in a front-loading washing machine to cool BGM forever
The world!
The washing machine!
Spin it!!
いいなと思ったら応援しよう!
応援をどうぞよろしくお願い致します🥰頂きましたチップは創作活動に使用させて頂きます🥰✨