Tracking the rise of AI. Here for the breakthroughs, the chaos, and the future.
United Kingdom
Joined December 2013
- Tweets12.2K
- Following2.5K
- Followers26.5K
- Likes24.3K
OrctonAI retweeted
Meet Pika H3 Flex, debuting today as the #1 open weight model on DesignArena’s t2v board.
It’s built for professional iteration, with exportable audio stems, resolutions up to 4k, and faster-than-realtime generation. It can produce a 15-second clip with soundtrack in as little as 7 seconds.
So, yeah. It’s kind of a flex 💅
New Crissy 4k60 Crissy Music Vid 🎶📽️
Look To Meridian
Made with @Hailuo_AI Open H3 Local
Upscaled with @topazlabs Local
Edited Davinci Resolve
Music Suno
Available on all good streaming services 😉
And now another 2 additions to 4090 H3 tests😅
1. Singularity v1.3 + hyperflow - same prompt native 768p - 9m15s
2. Singularity v2 - same prompt - native 768p - 11m58s
I think v2 may need different prompting, its fast but a bit messy, will experiment.
Another quick addition.
Singularity V1.3, native 768p - cinematic version on 4090 RTX - 12m9s
More H3 tests on RTX 4090. All use same prompt with anime style, 15s duration & 4 image refs: 2 for location, 1 for character & 1 for costume.
1.FastH3, native 768p -4m56s
2.Singularity V1.3, 480p to 768p learned latent upscale + refine -6m26s
3.Singularity V1.3, native 768p -12m22s
Custom ComfyUI workflows with memory optimisations & experimental nodes.
More H3 tests on RTX 4090. All use same prompt with anime style, 15s duration & 4 image refs: 2 for location, 1 for character & 1 for costume.
1.FastH3, native 768p -4m56s
2.Singularity V1.3, 480p to 768p learned latent upscale + refine -6m26s
3.Singularity V1.3, native 768p -12m22s
Custom ComfyUI workflows with memory optimisations & experimental nodes.
OrctonAI retweeted
Thank you to everyone who ran FastH3 Omni on day one 🐥
@DeepForest_Inc @OrctonAI @sep_is_heim @Alone1Moon @AibeginnerR8 @nookwright @mapte_fireball @realrebelai, and @posi_posi8 for the M5 Max runs!
みなさん、ありがとうございます！🇯🇵haoailab.com/blogs/fasth3-om…
OrctonAI retweeted
check it out !
hope to see more of your awesome creations in future 🚀
haoailab.com/blogs/fasth3-om…
OrctonAI retweeted
(6/7) On day one, the community ran it everywhere: RTX 4070 12 GB, 16 GB cards, RTX 3090, 4090 and 5090, mostly in @ComfyUI and many from Japan 📷
Thank you @DeepForest_Inc @OrctonAI @sep_is_heim @realrebelai and everyone sharing what they make!
Links in the blog!
Been testing FastH3 Omni in @comfyui since the weights were released yesterday on my 4090 RTX.
Heres 4 clips using single image reference for character, different seeds, all first gens.
1. 768p 30s in 10m4s
2. 768p 15s in 3m48s
3. 480p 30s in 3m6s
4. 480p 15s in 1m24
Workflow is work in progress with custom nodes, Think theres room for more improvement.
Thanks for the open weights!
@aryan_xv @haoailab @reactorworld
OrctonAI retweeted
Day 0 support for FastH3 Omni 🤩
FastH3 OmniRef by @haoailab's is now supported in ComfyUI.
This distilled version of Minimax H3 allows multi-modal reference generation of videos. It runs locally on your hardware and provide great balance between quality and speed.
Click the links below for open weights and cloud ⬇️
The Meridian Bounty — Part 2: The Brother
A 15-Minute Cyberpunk Sci-Fi Action Film🎬
This took a bit longer than expected 😏
Seedanca 2.5 @dreamina_ai #DreaminaCPP
Upscaled with @topazlabs
🤖 Made with AI
Been making a follow up short to the Meridian bounty for a week and a half … also a new Crissy song for the short .
Learnt so much doing this one…
I think it’s the best yet by a mile…
Releasing tomorrow 3pm ish uk time
Let’s hope the Algo will be kinda than pt 1 😬
