20+ years in software. VP Engineering, visual systems, audio… and I also bike every day.
Brooklyn, NY
- Tweets 186
- Following 16
- Followers 390
- Likes 170
I made a virtual depth of field effect by using the Apple NPU to general real time depth probes and ran it thru the DOF algorithm by Earl Hammon Jr (Infinity Ward)
This is just a regular macbook camera. Absolutely bonkers that you can now simulate the properties of DSLR cameras in real time at a stable framerate
1
1
11
1,391
I made an optical camo filter and hooked it up to NDI and now I can gaslight people on Zoom calls that “I’m not seeing anything weird on my end”
1
11
329
Apple Neural Networks for off axis projection, Claude + Blender MCP to procedurally generate some of the models, FFT analysis + Simplex noise frequency distribution + dual pass glass shader all at a stable 60fps
2
18
1,099
Apple Neural Networks are so fast that you can now build off axis projections that use glass shaders and PBR lighting
7
45
748
59,893
I integrated Apple-native NNs running in sidecar and can now run accurate landmarks at a stable 60fps, freeing up GPU for intensive shader work
53
127
2,760
4,632,094
This effect got me 1M views on reels. Explaining how the viral virtual shadow projection technique works with real-time parallel face tracking+segmentation running in Sidecar at a stable 60fps, even with high-poly 3D models and complex lighting.
1
23
1,265
Realtime depth maps are so hot right now. Running realtime depth estimation (with my fixed relative algo), face tracking and the dual-pass refraction shader (pioneered by @MaximeHeckel ). Comment GitHub for the repo.
1
1
19
913
Can't let Konrad have all the fun. I built a realtime TD network running both depth and landmark tracking natively on the M4 Pro's GPU, stable at 60fps. GitHub repo here: github.com/ojrgb/appletd
Depth-aware light injection in TypeGPU
I got a 448x448 monocular depth model down to ~8 ms on my M4 Pro across ~250 dispatches, which is fast enough to use in realtime :D
Since the inference is written directly in TypeGPU, I can just feed the depth buffer straight into the lighting pass. It never has to leave the GPU or go through any extra synchronization/interop step
Inference, lighting and draw all go through the same command encoder.
4
10
113
10,064
Quick look at the blobtrack effect I made for @MaximeHeckel ‘s blog. Some math involved to get the videos side by side and it uses the custom Dither + digital crunch effect rack for that retro Gameboy Advance feel.
2
12
1,509
I started playing around with visual algos because that’s the only remaining medium that lets me actually code.
2
360
If you made it here from one of those posts calling me a 20 yr old American student who sold his software for $230k, welcome. I’ll be drip feeding some vids so you can see it’s actually me. Not vibe coded, just coded. Peace! ✌️ #touchdesigner #vfx #visuals
5
417
Quick video of the off-axis projection trick.
Disclaimer: I am not a 20 year old American student
3
268
