Back at it
-
The floor also needed some bounce lighting
'Could somebody with no programming experience recreate Photoshop with an LLM?'
I have absolutely zero Metal and near-zero 3D modeling experience. I know the basics of how to use a scene editor, and the names of rendering terms.
And I effectively vibecoded all of this in less than half a day with Codex 5.3 Medium, based on screenshots of a cool-looking app I've never used (Valence3D) and just a general sense of what a 3D app should do
-
'Could somebody with no programming experience recreate Photoshop with an LLM?'
I have absolutely zero Metal and near-zero 3D modeling experience. I know the basics of how to use a scene editor, and the names of rendering terms.
And I effectively vibecoded all of this in less than half a day with Codex 5.3 Medium, based on screenshots of a cool-looking app I've never used (Valence3D) and just a general sense of what a 3D app should do
Raytracing uses every last ounce of my poor M1 Mac mini's GPU. It has never seen such a workout
-
Raytracing uses every last ounce of my poor M1 Mac mini's GPU. It has never seen such a workout
@stroughtonsmith I don't think hardware raytracing was added until M3 — that M1 GPU must be cooking!
-
Raytracing uses every last ounce of my poor M1 Mac mini's GPU. It has never seen such a workout
@stroughtonsmith Did you build the path tracer yourself or is that a system library?
And what exactly will this tool do? I assume you're not building a full modelling tool.
-
@stroughtonsmith Did you build the path tracer yourself or is that a system library?
And what exactly will this tool do? I assume you're not building a full modelling tool.
@k0bin Codex built it from scratch. I don't know what it will do yet, I'm just experimenting
-
@stroughtonsmith I don't think hardware raytracing was added until M3 — that M1 GPU must be cooking!
@edmn yeah it's not using hardware-accelerated raytracing. I don't know enough about the feature to know whether that would even be useful in multi-minute raytracing of static images rather than optimizing things for realtime
-
Raytracing uses every last ounce of my poor M1 Mac mini's GPU. It has never seen such a workout
Since I am doing this on the GPU, not CPU, finding a way to render complex meshes without blowing the budget is a headache. There's a hard cap on how much work you can do before the Metal driver crashes
-
Raytracing uses every last ounce of my poor M1 Mac mini's GPU. It has never seen such a workout
@stroughtonsmith add in some radiosity to really test it
-
@stroughtonsmith add in some radiosity to really test it
@fatlazycat it's already doing bounce lighting
-
@edmn yeah it's not using hardware-accelerated raytracing. I don't know enough about the feature to know whether that would even be useful in multi-minute raytracing of static images rather than optimizing things for realtime
@stroughtonsmith Sounds like you need to buy an M4 Mac mini!
-
'Could somebody with no programming experience recreate Photoshop with an LLM?'
I have absolutely zero Metal and near-zero 3D modeling experience. I know the basics of how to use a scene editor, and the names of rendering terms.
And I effectively vibecoded all of this in less than half a day with Codex 5.3 Medium, based on screenshots of a cool-looking app I've never used (Valence3D) and just a general sense of what a 3D app should do
@stroughtonsmith are you using Codex GUI app, or are you using codex trough the Xcode 16.3 AI CLI ?
-
@stroughtonsmith are you using Codex GUI app, or are you using codex trough the Xcode 16.3 AI CLI ?
@guybrooker the Codex app. Xcode has too many guardrails in place
-
'Could somebody with no programming experience recreate Photoshop with an LLM?'
I have absolutely zero Metal and near-zero 3D modeling experience. I know the basics of how to use a scene editor, and the names of rendering terms.
And I effectively vibecoded all of this in less than half a day with Codex 5.3 Medium, based on screenshots of a cool-looking app I've never used (Valence3D) and just a general sense of what a 3D app should do
@stroughtonsmith This is what vibe coding should be used for. AI should be used to screw over Autodesk and Adobe
-
Since I am doing this on the GPU, not CPU, finding a way to render complex meshes without blowing the budget is a headache. There's a hard cap on how much work you can do before the Metal driver crashes
I think I should probably add a denoiser to this…
-
I think I should probably add a denoiser to this…
It’s a warm, sunny day, so I moved outside to sit with the dog. My iPad has a much faster GPU anyway, so rendering works great here side by side with my Mac over VNC. And it’s a nanotexture model, so working in direct sunlight is effectively the same as working indoors
-
It’s a warm, sunny day, so I moved outside to sit with the dog. My iPad has a much faster GPU anyway, so rendering works great here side by side with my Mac over VNC. And it’s a nanotexture model, so working in direct sunlight is effectively the same as working indoors
@stroughtonsmith sunny enough to sit outside?! It's 2.8C here. 🥶
-
@stroughtonsmith sunny enough to sit outside?! It's 2.8C here. 🥶
@bigzaphod I’ve just had two months of non-stop rain, let me enjoy it
-
@stroughtonsmith sunny enough to sit outside?! It's 2.8C here. 🥶
@bigzaphod @stroughtonsmith We’re back to rain today here, but we just ended a beautiful sunny week — got up to 17°C yesterday. Still frosty some mornings though.
It’s been an exceptionally mild winter overall for us: we never even bothered to cover the vents into the crawlspace, or drain and cover the outside water taps for the first time I can remember.
Which sounds nice, but it’s going to be rough come summer — there’s not nearly enough snowpack, which we kinda depend on.
-
I think I should probably add a denoiser to this…
Added some targeted calls to the hardware raytracing APIs, and a toggle switch so I can flip between non-accelerated RT and accelerated RT. There are plenty of bottlenecks in my renderer to stop it taking /full/ advantage of acceleration, but holy heck it’s much faster regardless. This is on the M5 iPad Pro
-
Added some targeted calls to the hardware raytracing APIs, and a toggle switch so I can flip between non-accelerated RT and accelerated RT. There are plenty of bottlenecks in my renderer to stop it taking /full/ advantage of acceleration, but holy heck it’s much faster regardless. This is on the M5 iPad Pro
Turns out the M1 supports enough of the hardware accelerated raytracing APIs that I get a massive speedup here, too?
Trying to parse Apple's Metal feature support tables is hurting my brain, so I'll just accept it and move on
