Ugo Capeto 3d

Researcher and software developer: stereo photography, depth map generation, 2d-3d conversion, structure from motion, multi-view stereo, photogrammetry, etc.

[Ugo Capeto 3d @ugocapeto3d]

3D Software (depth map generation, 2d to 3d conversion, structure from motion, lenticular imaging, etc): https://t.co/bM1PXSwetj

[Ugo Capeto 3d @ugocapeto3d]

I have tested runwayml Gen-2 Text to Video AI tool. Not that great imo. https://t.co/II41zSqSI6

[Ugo Capeto 3d @ugocapeto3d]

Using Leonardo . ai (image to image) to create renderings that kinda look like oil paintings https://t.co/YzatNUQjYj

[Ugo Capeto 3d @ugocapeto3d]

@ArtSquirrelBlog Haha, awesome!

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/T063aannAp MiDaS V3.1 is out! I compare it with LeiaPix Converter (Boosting Monocular Depth). All about depth maps from single images using AI.

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/34LG6FWw9U Depth map generation from stereo pair VS depth map generation from single image using AI

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/P9bBkdVwLD
From stereo pair to lenticular printing using Frame Interpolation AI (runway ml) and Grape (interlacing)

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/JkAw2j3838
Image upscaling using AI (using runway ML). From blurry to sharp (and larger) with minimal effort.

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/3F4h1zlvzG
Frame interpolation using AI (runway ML). Useful to get synthetic frames between 2 images (left and right of a stereo pair, for example). Useful for lenticular imaging/printing.

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/xtS5m41aFy
I have tested "3D Photo Inpainting" on huggingface

[Ugo Capeto 3d @ugocapeto3d]

I have tested "Stable Diffusion Inpainting" aka inpainting using AI
https://t.co/5rnTLlc0gb

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/bHRQbHrABt
Using lenticular lens for animations (photicular book)

[Ugo Capeto 3d @ugocapeto3d]

I have tested frame interpolation using AI (for the purpose of getting intermediate views between left and right images of a stereo pair) https://t.co/pbfQfXnx5j

[Ugo Capeto 3d @ugocapeto3d]

I have tested DreamStudio (beta), a stable diffusion image generator from text prompts
https://t.co/6fhBckMw1s

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/lPdtkAMpSI
Dense 3d reconstruction from photos (MVS10)
Now on github!

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/00Urm8YJ4t
Structure from motion (photogrammetry) software to compute camera poses (sfm10)
is now on github

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/s2IXZoSY7W
Depth map improvement using the fast bilateral solver (DMAG9b).
Now on github.

[Ugo Capeto 3d @ugocapeto3d]

DMAG5 is now on github. Depth map from stereo pair using the Guided Image Filter.
https://t.co/CyGFOtkFex

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/lvJgvkHcl0

Given image + depth map, how to generate the other image to get a stereo pair in StereoPhoto Maker

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/it1iaNMJUC
Depth map from stereo pair using local method (window matching) aka DMAG2.
Classic Computer Vision algorithms. Before AI became a thing.
Code is now available on github.

[Ugo Capeto 3d @ugocapeto3d]

https://t.co/eTLPewhvDN My first depth map generator (from stereo pair) using optical flow. The original DMAG. Code now on github.