MathiBarragan's profile picture. :)

Mathias

@MathiBarragan

:)

Dit vind je misschien leuk
Vastgezet

Observations on creativity, art, and ai


i just want to comfortable play different mode solos


people are still defiling friend posters in nyc @AviSchiffmann you still paying for them?


Mathias heeft deze post opnieuw geplaatst

In case you missed it last Friday, the recording is now live on YouTube! Using WAN 2.1 VACE + Masking + Pose Estimation to create Hollywood level edits in @ComfyUI youtube.com/watch?v=a17D2R…

gregschoeninger's tweet card. Fine-Tuning Wan 2.1 VACE to Swap Costumes

youtube.com

YouTube

Fine-Tuning Wan 2.1 VACE to Swap Costumes

Learn how Hollywood is using ComfyUI to fix shots in post that they could never dream of before. Tomorrow, we'll be live streaming how to use @ComfyUI plus a fine-tuned LoRA from @oxen_ai to do a high quality, cinematic, outfit swap. We'll be taking a famous scene from Pulp…



Mathias heeft deze post opnieuw geplaatst

Learn how Hollywood is using ComfyUI to fix shots in post that they could never dream of before. Tomorrow, we'll be live streaming how to use @ComfyUI plus a fine-tuned LoRA from @oxen_ai to do a high quality, cinematic, outfit swap. We'll be taking a famous scene from Pulp…


you can tell a lot about a musician by their opening act


Mathias heeft deze post opnieuw geplaatst

On the @ArtificialAnlys Image Editing leaderboard, Riverflow ranks at the top, outperforming every other editing model on most tasks. It doesn’t guess. It designs.

sourceful's tweet image. On the @ArtificialAnlys Image Editing leaderboard, Riverflow ranks at the top, outperforming every other editing model on most tasks.

It doesn’t guess. It designs.

Mathias heeft deze post opnieuw geplaatst

Why do some diffusion model fine-tunes nail composition but fail on details? Or capture perfect details but mess up the overall structure? It all comes down to how you schedule noise during training. In our latest blog post we break down why different noise timestep strategies…

gregschoeninger's tweet image. Why do some diffusion model fine-tunes nail composition but fail on details? Or capture perfect details but mess up the overall structure? It all comes down to how you schedule noise during training.

In our latest blog post we break down why different noise timestep strategies…

Mathias heeft deze post opnieuw geplaatst

Looking for builders with some Spooky AI demos for @AITinkerers on October 20th, in Venice, CA 👻 Feel free to DM me if you are building something cool and want to show it off. la.aitinkerers.org/p/ai-tinkerers…


Mathias heeft deze post opnieuw geplaatst

Open Weights AI vs OpenAI

gregschoeninger's tweet image. Open Weights AI vs OpenAI
gregschoeninger's tweet image. Open Weights AI vs OpenAI

Day 3 of Fine-Tuning: Multi-Character Video Generation Pipeline > masking (@Meta's DinoV3) > inpainting (@Alibaba_Qwen's Qwen-Image-Edit) > image-to-video (@Alibaba_Wan's WAN-2.2-I2V-A14B) Checkout the full process 👇



“dance so fast that the dancer disappears and only the dance remains” - Osho


optimizing for your current workflow != efficiency/ effectiveness


who else has been getting so many ai slop ads on youtube? like 5 in a row…


there’s no such thing as genres in music anymore


United States Trends

Dit vind je misschien leuk

Loading...

Something went wrong.


Something went wrong.