Cyril Diagne's Avatar

Cyril Diagne

@cyrildiagne.bsky.social

AI indie hacker. Prev: founded Clipdrop (YCW21, acq. Stability AI), resident at Google A&C Lab, Prof. & head of MID at ECAL

761 Followers  |  171 Following  |  42 Posts  |  Joined: 14.11.2024  |  1.9408

Latest posts by cyrildiagne.bsky.social on Bluesky

Video thumbnail

Excited to share our new research at Jasper Research! ๐Ÿš€

LBM: Latent Bridge Matching for Fast Image-to-Image Translation

Try out our @hf.co space for object relighting!

๐Ÿค— @gradio-hf.bsky.social demo: huggingface.co/spaces/jaspe...
๐Ÿ‘‰ Paper: arxiv.org/abs/2503.07535
๐Ÿ’ป Repo: github.com/gojasper/LBM

13.03.2025 16:00 โ€” ๐Ÿ‘ 22    ๐Ÿ” 7    ๐Ÿ’ฌ 2    ๐Ÿ“Œ 4

Amazing!

14.03.2025 08:10 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Wow awesome work!! ๐Ÿคฉ

14.03.2025 08:09 โ€” ๐Ÿ‘ 3    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

Alright whoโ€™s making this

07.03.2025 17:08 โ€” ๐Ÿ‘ 4    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

oh wow looks awesome, somehow missed it

25.01.2025 15:42 โ€” ๐Ÿ‘ 4    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

๐Ÿ˜…

16.01.2025 15:44 โ€” ๐Ÿ‘ 5    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Hopefully they'll remove some of the FP16 nerfs they have on the 4090.. The 2-slots factor is also a nice improvement for multi-gpu builds

16.01.2025 08:52 โ€” ๐Ÿ‘ 6    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

๐Ÿฅฐ

11.12.2024 19:50 โ€” ๐Ÿ‘ 3    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Video thumbnail

How does one keep track? The monodepth/tracking field these days:

06.12.2024 10:05 โ€” ๐Ÿ‘ 13    ๐Ÿ” 2    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Video thumbnail

Align3R: estimates camera poses and consistent depth maps from monocular videos.

Combining it with trackers like Cotracker3 or SAM2 could unlock many fun applications! (cf: VideoDoodles by Yu & al)

Project page (with demo): igl-hkust.github.io/Align3R.gith...
Code: github.com/jiah-cloud/A...

06.12.2024 09:48 โ€” ๐Ÿ‘ 24    ๐Ÿ” 3    ๐Ÿ’ฌ 2    ๐Ÿ“Œ 0
Post image

Am I the only one amazed that this is what 2*4TB (with thermal case) looks like now?

04.12.2024 09:41 โ€” ๐Ÿ‘ 9    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

๐Ÿ˜ฎ - wall street seems to take the news well though

02.12.2024 16:24 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

quite the illustration ๐Ÿ˜…

01.12.2024 12:32 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Preview
The UX of LEGO Interface Panels LEGO interface panels are beautiful, iconic, and great for learning interface design basics. I bought 52 of them from BrickLink to explore the design, layout and organisation of complex interfaces.

The UX of LEGO Interface Panels interactionmagic.com/UX-LEGO-Inte...

28.11.2024 19:51 โ€” ๐Ÿ‘ 120    ๐Ÿ” 33    ๐Ÿ’ฌ 2    ๐Ÿ“Œ 8

Banned from bsky or HF?

28.11.2024 12:48 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

fair enough!

28.11.2024 11:44 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Oh no ๐Ÿ˜‚ Iโ€™m torn between rofl for this troll and a fear to see this little drama escalating

28.11.2024 10:38 โ€” ๐Ÿ‘ 7    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Look Ma, no markers

Why the preference for multiview? Maybe something like github.com/ttxskk/AiOS can be adapted/finetuned with multiple views from a synthetic dataset like microsoft.github.io/SynthMoCap/

27.11.2024 23:26 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

oh nice, bookmarked!

27.11.2024 21:53 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

The past few months have been... intense! There's still quite some work to do before the finish line, but excited to launch in the coming weeks ๐Ÿ’ชโšก๏ธ

27.11.2024 16:42 โ€” ๐Ÿ‘ 13    ๐Ÿ” 0    ๐Ÿ’ฌ 2    ๐Ÿ“Œ 0

And early social media platforms!

22.11.2024 16:35 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Video thumbnail

Resemble Enhance seems pretty good: github.com/resemble-ai/...

22.11.2024 08:39 โ€” ๐Ÿ‘ 6    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Adobe Podcast V2 is a really impressive audio enhancer.
Is there any open-source tech close to it?

bsky.app/profile/pins...

22.11.2024 08:38 โ€” ๐Ÿ‘ 8    ๐Ÿ” 1    ๐Ÿ’ฌ 2    ๐Ÿ“Œ 0
Video thumbnail

SAMURAI: improve the tracking robustness of SAM2 with 2 main contributions:
- adding motion information to the mask selection
- curating the memory bank based on motion cues

Project: yangchris11.github.io/samurai
Code: github.com/yangchris11/...
Paper: arxiv.org/abs/2411.11922

21.11.2024 08:17 โ€” ๐Ÿ‘ 17    ๐Ÿ” 6    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 1
Video thumbnail

Pyramid Flow is quite impressive for img2video, given than it was only trained on public datasets. Clearly not as dynamic and stable as commercial solutions, but the gap seems to be closing github.com/jy0205/Pyram...

20.11.2024 09:35 โ€” ๐Ÿ‘ 4    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

A bit surprised with this data from Clerk on sign-in methods preferences: From a sample of 2.5M sign-in, <2% of users chose to use magic links.

20.11.2024 09:31 โ€” ๐Ÿ‘ 2    ๐Ÿ” 1    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Yessss ๐Ÿ”ฅ๐Ÿ”ฅ

19.11.2024 19:20 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Preview
Realtime diffusion in the cloud In December 2023, I implemented a realtime diffusion toolkit with Daito Manabe and Rhizomatiks. The toolkit is based on SDXL Turbo runningโ€ฆ

"A nicely maintained and over-specโ€™d server just has a smell to it" - great writeup by @kcimc.bsky.social benchmarking various cloud GPU providers for a realtime diffusion installation: kcimc.medium.com/realtime-dif...

19.11.2024 09:18 โ€” ๐Ÿ‘ 12    ๐Ÿ” 2    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

new life goal: be added to that list ๐Ÿ˜…

18.11.2024 21:38 โ€” ๐Ÿ‘ 3    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

These are really impressive

18.11.2024 19:44 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

@cyrildiagne is following 20 prominent accounts