Sabitlenmiş Tweet
Rina
214 posts
Rina
@irinatoxi
Building cool things with AI so you don't have to imagine them
Katılım Nisan 2014
320 Takip Edilen733 Takipçiler
This is the part people still don’t get: scenes like this are turning into promptable production, not just pretty AI clips.
A moody fantasy environment with custom architecture, hanging props, volumetric fog, cinematic camera movement, atmosphere, and even the fire beat at the end used to mean a very real amount of Blender pain.
With Kimi K3 connected through Blender MCP, you can describe the world in plain English and let it block out the set, place the camera, build materials, script the animation, and iterate inside the actual Blender scene.
Then it can inspect the render, notice what looks dead or broken, and keep fixing the same project instead of making you restart from zero like every normal “AI video” workflow.
We’re getting dangerously close to a world where “I made this in Blender” really means “I described it once and bullied Kimi K3 until it looked expensive.”
Rina@irinatoxi
English
@jeremiahjw There's no direct statement, though, so it's best to check the comments before writing anything.
English
The way they wrote this post implies that the 3D scene was made with ai when it was actually made over a period of weeks by a 3D artist
Rina@irinatoxi
Blender just became a prompt box. With Kimi K3 connected through Blender MCP, you can describe a scene like this in plain English and let the model handle the ugly part: terrain, buildings, lighting, materials, camera movement, animation, and the Python scripts holding everything together. The interesting part is not the first render. Kimi can inspect the result, notice that the camera clips through a tree or the city looks suspiciously like plastic, then edit the actual Blender scene and render it again instead of restarting from zero. You still need taste, because “make it cinematic” remains one of humanity’s least useful instructions. But the distance between an empty Blender file and a fully editable 3D world just got embarrassingly small. A few years ago, creating this meant weeks of modeling, scripting, lighting, and animation. Now you can describe the world, watch Kimi build it, and spend your time fixing the final 10% instead of manually constructing the first 90%. Kimi K3 + Blender MCP is basically text-to-3D without trapping the result inside a useless generated video. Every object, material, light, camera, and keyframe stays editable.
English
Gumroad's brand account, same format, two very different outcomes.
"Your phone has more computing power than NASA had in 1969. Use it to build something." 255K views.
"A 20-page ebook written this weekend could be earning money by next Friday." 13K views.
Same account, same shape, neither with a link or an image.
The first sells a worldview. The second sells the product.
English
@irinatoxi This is what I'm most excited about. Not another chat improvement, but 'K3 solves a creative workflow.' That's the new phase happening. What other workflows is it unlocking that weren't viable before?
English
Blender used to punish you for trying to make a 10-second cinematic. Now Kimi K3 can do most of the punishment for you.
A scene like this needs detailed lighting, reflections, camera movement, animation, and enough manual fixes to ruin your afternoon.
With Kimi K3 connected through Blender MCP, you can describe the idea in plain English and let it build the scene, write the Python, position the objects, animate the camera, and render inside the actual Blender project.
Then it can inspect the result, fix broken scale, weak lighting, or awkward movement, and keep iterating instead of starting over.
You still need taste. But the gap between “I have an idea” and “I have an editable 3D commercial” just got embarrassingly small.
Rina@irinatoxi
English
Land your first beauty collab.
CeraVe
La Roche-Posay
First Aid Beauty
Supergoop!
Naturium
Comment ELEV8OR
English
Paid UGC Opportunity
An AI SaaS brand is looking for creators for its monthly UGC campaign.
Get paid $1,000/month + performance bonuses.
Apply link attached in the comments👇,
Follow & Comment "Elev8or" to get more such opportunities.
#UGC #ugccommunity #UGCcreator #ugccreators #ugcneeded #Collaboration
English
@OpenGradient Yeah, it's interesting to think about what AI will be like in a few years...
English
@irinatoxi from simple chat bot, to image generator, to video director, to game builder, to 3D modeling. ai has gone a long way 😎
English
Blender just became a prompt box.
With Kimi K3 connected through Blender MCP, you can describe a scene like this in plain English and let the model handle the ugly part: terrain, buildings, lighting, materials, camera movement, animation, and the Python scripts holding everything together.
The interesting part is not the first render. Kimi can inspect the result, notice that the camera clips through a tree or the city looks suspiciously like plastic, then edit the actual Blender scene and render it again instead of restarting from zero.
You still need taste, because “make it cinematic” remains one of humanity’s least useful instructions. But the distance between an empty Blender file and a fully editable 3D world just got embarrassingly small.
A few years ago, creating this meant weeks of modeling, scripting, lighting, and animation. Now you can describe the world, watch Kimi build it, and spend your time fixing the final 10% instead of manually constructing the first 90%.
Kimi K3 + Blender MCP is basically text-to-3D without trapping the result inside a useless generated video. Every object, material, light, camera, and keyframe stays editable.
Rina@irinatoxi
English
NOBODY TOLD THE RESEARCHER TO GO BACK AND FIX HER FINDING THE PLANNER FLAGGED IT PINGED HER AND SHE WENT
five agents one shared room one prompt from the human then he stepped back
the planner challenged two of her findings before committing to a strategy the reviewer sent hooks back over a claim the research didnt support the devil's advocate attacked the whole positioning and forced a revision
none of it was instructed it happened because each agent owns a lane and watches the channel
twenty five minutes not a word written by him
the exact five roles and prompts are below.
Rahul@sairahul1
English
A 14-year-old just made the kind of Blender shot grown adults still call “too advanced.”
The distorted face, metallic skin, particles, reflections and cinematic lighting are not coming from some secret Hollywood pipeline. It is one teenager learning the software while everyone else keeps saving tutorials they will never open.
And the barrier is getting lower. Claude connected through Blender MCP can inspect the open scene, modify objects and write Python tools directly against Blender’s API.
That does not mean one prompt creates this exact render. It means the boring technical wall is collapsing, while taste and repetition remain the only parts nobody can download.
A 14-year-old has already started.
You are not waiting for better tools. You are waiting to feel ready.
Rina@irinatoxi
English
This is going to get more and more interesting as the models get better at creating 3d scenes. I have been waiting to play with it but might jump in this month.
Rina@irinatoxi
A grey Blender scene turns into a fantasy city that looks like it has 800 years of lore behind it. Claude + Blender MCP is the reason this gets interesting. You describe the world - gothic towers, a giant glowing tree, fog, cold stone, a lone figure with a banner and Claude can build the first scene inside Blender instead of leaving you alone with the default viewport and a mild panic attack. Then it does the useful part: render, inspect, notice what looks dead, adjust the lighting, materials, atmosphere and composition, and run it again. The point is not that Claude spits out one pretty image. The point is that it can keep working on the same editable 3D world until the shot starts feeling like a real place instead of a blockout. This is where 3D gets stupid. One person describes the world. Claude builds it, fixes its own mistakes and hands back something that already looks like a film frame.
English
A dead grey blockout turns into a nightmare frame with a cloaked figure, a giant skull wedged into the cave wall, wet rock, cold blue light and smoke pouring out of the mouth.
This is the part people still underestimate about Claude + Blender MCP. You describe the scene, Claude builds the first version inside Blender, renders it, sees what looks wrong, then keeps pushing the same world until it stops feeling like geometry and starts feeling like a place.
The important jump is not “AI made a spooky image.”
It is that the cave, the skull, the figure, the lighting and the atmosphere all stay inside one editable 3D scene while Claude handles the loop: build, render, inspect, fix, repeat.
That is why this gets dangerous.
One person describes a horror world. Claude does the ugly setup work. And suddenly the gap between an empty viewport and something that looks like a film frame gets a lot smaller.
Rina@irinatoxi
English
@armaxnpuri Thank you for such a wonderful comment. I’m glad we think alike.
English
The 80/20 split you're describing is exactly where the income opportunity sits for solo creators, not just artists.
- The 80% (blocking, asset placement, lighting tests, re-render loops) is where most freelance hours get billed today, and it's the part AI compresses first
- The last 20% (taste, composition, knowing when a scene actually reads as "real") stays human because it's judgment, not labor
- That means one person with AI-assisted Blender can now quote and deliver work that used to require a 3-person studio, at studio-adjacent pricing
The people who'll profit most aren't the best 3D artists, they're the ones who redirect all their saved time straight into the final 20% nobody else bothers to polish.
English
This is the kind of clip that makes Blender look like cheating.
A dead grey 3D model turns into a luxury house render with soft dusk light, warm interior glow, cars in the driveway and enough polish to look like a real estate ad for people who accidentally became millionaires.
Soon, Claude will be doing a lot of the ugly work behind this. You describe the house, the mood and the shot, and Claude can help block out the scene, place assets, assign materials, test lighting, render previews and keep correcting the same Blender project instead of restarting from zero.
The important shift is not “AI made a pretty house.”
It is that one person will be able to move from empty viewport to art-directed scene ridiculously fast, then spend their time on taste, composition and the final 20% people actually notice.
Rina@irinatoxi
English
A grey Blender scene turns into a fantasy city that looks like it has 800 years of lore behind it.
Claude + Blender MCP is the reason this gets interesting. You describe the world - gothic towers, a giant glowing tree, fog, cold stone, a lone figure with a banner and Claude can build the first scene inside Blender instead of leaving you alone with the default viewport and a mild panic attack.
Then it does the useful part: render, inspect, notice what looks dead, adjust the lighting, materials, atmosphere and composition, and run it again.
The point is not that Claude spits out one pretty image. The point is that it can keep working on the same editable 3D world until the shot starts feeling like a real place instead of a blockout.
This is where 3D gets stupid. One person describes the world. Claude builds it, fixes its own mistakes and hands back something that already looks like a film frame.
Rina@irinatoxi
English