Abyssal

ai era || ai – is a tool for creating crazy things || DM open

[Abyssal @abyssallD]

Connector Blender MCP + Claude

It gives the model access to an open stage and during the conversation you can sketch the necessary primitives without modeling in the classic way. Camera, composition, arrangement of objects - explain in natural language

https://t.co/2HYJiWqOh7

[Abyssal @abyssallD]

SEEDANCE 2.0 JUST TOOK A BLENDER BLOCKOUT AND TURNED IT INTO A FULL CINEMATIC SCENE

Top: simple colored boxes in Blender.

Bottom: real actors, lighting, smoke, tension.

Same camera moves. Same blocking. Same composition.

Only now it actually looks like real cinema.

This is the kind of control people have been waiting for — precise 3D reference without losing image quality.

Seedance 2.0 isn’t guessing anymore. It’s listening.

You can quickly build the blockout using a connector that plugs into Claude. I’ll drop the link in the comments.

[Abyssal @abyssallD]

This tweet started it all

November 30, 2022. ChatGPT launches.

One click — and the world changed forever.

What was your very first prompt? 😂

[Sam Altman @sama]

today we launched ChatGPT. try talking with it here:

https://t.co/uWra8LKFMN

[Abyssal @abyssallD]

A GUY BUILT AN APP FOR AR FILMING IN VIRTUAL SPACE

You create a 3D scene, move the phone like a real camera operator, choose angles, and switch between characters.

Then it goes into Seedance for processing.

Link app: https://t.co/eM7InkHTlO

There’s another way.

You can just film real people doing the movements, then use BEEBLE AI to put any character in any clothes on them.

Real actors almost always perform the scene better than what you get from prompts.

Your choice.

[Abyssal @abyssallD]

@itsshara_ai @img1 epic entrance🔥 portal directly to the party - it was powerful

[Abyssal @abyssallD]

@misat0x great breakdown! consistency is what actually turns AI generations into a real story instead of random beautiful shots. reference sheets + storyboard are the real game-changer.

[Abyssal @abyssallD]

@misat0x i agree, the timeline is really wild 🤯

[Abyssal @abyssallD]

🚨PEOPLE ARE ALREADY BETTING MONEY ON WHETHER ANTHROPIC WILL EXTEND ACCESS TO FABLE 5

Not on elections. Not on sports.

On whether developers get a little more time with the model.

Right now the odds sit at 69%.

The guy running a 100-app challenge is currently on number 63.

He had to slow down because of burnout.

If they extend access, he says he’ll finish the hundred.

At the same time Kimi 3 launched today and GPT-6 is already on the horizon.

So the real question is simple:

Will Anthropic actually release the version of Fable 5 that a lot of people think they’re still holding back, or will they just quietly pull it because everyone is going to switch to GPT 5.6, GPT-6 and Kimi anyway.

[Abyssal @abyssallD]

@itsshara_ai Agreed! One instruction and it completely transforms the onboarding into something that actually feels like a game. Strong test indeed — and it passed

[Abyssal @abyssallD]

A GUY DECIDED TO PUSH GPT 5.6 SOL TO THE LIMIT

He already had his own Poster Act built with Claude Fable 5.

He asked GPT 5.6 Sol to make a better version.

Then he told it to completely rebuild the onboarding so that instead of a normal landing page the user would feel an instant wow effect.

GPT 5.6 Sol delivered a new Tester Act where you can change its shape.

Visually Fable 5 still feels cleaner.

But when it came to the onboarding.

GPT did something different:

> a loading screen that looks like it came from a video game
> hover effects
> smooth animations
> a clean transition into the product

It didn’t just look nice.

It felt closer to a game interface than a regular website.

At the end he says the simple thing: if everyone claims vibe coding is dead, just build something no one has seen before.

Use GPT 5.6 Sol, use Fable 5 — doesn’t matter.

Gamify the product, add animations, and make something that breaks the template.

[Abyssal @abyssallD]

@aiseomastery i agree, prompts really save a lot of time 🤩

[Abyssal @abyssallD]

GOOD SITE FOR PEOPLE MAKING VIDEOS WITH AI

It has almost all the main camera movements with visual examples and ready-to-use prompts.

The site is called
https://t.co/s1CocYmjDa

You can just grab the movement you need and drop it straight into your generation.

Especially useful when you want to quickly add a specific camera move:

>push in
>orbit
>dolly
>zoom, etc

Instead of struggling to describe it every time, you copy the right technique and add it to your scene.

Useful both for beginners and for people who want to work with camera more intentionally in AI video.

[Abyssal @abyssallD]

A GUY DECIDED TO PUSH GPT 5.6 SOL TO THE LIMIT

He already had his own Poster Act built with Claude Fable 5.

He asked GPT 5.6 Sol to make a better version.

Then he told it to completely rebuild the onboarding so that instead of a normal landing page the user would feel an instant wow effect.

GPT 5.6 Sol delivered a new Tester Act where you can change its shape.

Visually Fable 5 still feels cleaner.

But when it came to the onboarding.

GPT did something different:

> a loading screen that looks like it came from a video game
> hover effects
> smooth animations
> a clean transition into the product

It didn’t just look nice.

It felt closer to a game interface than a regular website.

At the end he says the simple thing: if everyone claims vibe coding is dead, just build something no one has seen before.

Use GPT 5.6 Sol, use Fable 5 — doesn’t matter.

Gamify the product, add animations, and make something that breaks the template.

[Abyssal @abyssallD]

@MantaRiccardo thx for quoting, and it’s really a very interesting resource

[Abyssal @abyssallD]

@misat0x AI is great at validating demand and speeding up production, but without your own perspective and an understanding of what will actually resonate with the audience, it’s just noise 🤯

[Abyssal @abyssallD]

@misat0x emotions from a single trade overshadow real calibration

[Abyssal @abyssallD]

@misat0x to be honest, I would too 😅

[Abyssal @abyssallD]

the resource is really great for beginners.
for someone who's just starting to make AI videos, it's one of the most useful and beginner-friendly tools right now.
his idea with animation/videography skills + a simple three.js camera viewer is also excellent, especially if you already have some experience. Both approaches are solid!

[Abyssal @abyssallD]

@DomRutledge_ yep, the resource are really very good

[Abyssal @abyssallD]

@madebywavers I highly doubt it, Claude is a very powerful tool, in some ways it's better than ChatGPT, in others it's behind, that's normal

[Abyssal @abyssallD]

CHATGPT 5.6 SOL IS CURRENTLY THE BEST TOOL FOR SHORT VIDEOS

A guy did a side-by-side comparison between ChatGPT 5.6 Sol and Claude Fable 5 using a 3D globe video about politics.

Claude handled the basics — showed the country, used stock footage, and kept things simple.

But it still looked like typical AI output.

ChatGPT 5.6 Sol in the same prompt delivered seven different camera angles.

The globe had much more detail, the graphics looked cleaner, and the overall visual quality was noticeably higher.

It also handled product design better — backgrounds, composition, and overall feel.

The difference was especially obvious on the “AI for Savages” page.

Claude made it look decent.

ChatGPT made it look expensive and polished.

It wasn’t just generation anymore — it was strong design combined with image creation.

Right now ChatGPT 5.6 Sol is clearly ahead when it comes to visual quality and understanding how to make a product actually look good.

[Abyssal @abyssallD]

@LatheesanK cool, thanks for the test! I'm glad the prompts really helped with LTX 2.3