Helpful blog post for understanding what kind of data is needed for these models!
Does this have any application for generating realistic scenes for robotics training?
sangwulee · 5h ago
Hi! I'm lead researcher on Krea-1. FLUX.1 Krea is a 12B rectified flow model distilled from Krea-1, designed to be compatible with FLUX architecture. Happy to answer any technical questions :)
dvrp · 6h ago
Hello everyone.
I’m the Co-founder and CTO of Krea. We’re excited because we wanted to release the weights for our model and share it with the HN community for a long time.
My team and I will try to be online and try to answer any questions you may have throughout the day.
OsrsNeedsf2P · 10m ago
Cool to see an open weight model for this. But what's the business use case? Is it for people who want to put fake faces on their website that don't look AI generated?
dvrp · 3m ago
Thanks!
From a business point of view, there are many use-cases. Here's a list in no particular order:
- You can quickly generate assets that can be used _alongside_ more traditional tools such as Adobe Photoshop, After Effects, or Maya/Blender/3ds Max. I've seen people creating diffuse maps for 3D using a mix of diffusion models and manual tweaking with Photoshop.
- Because this model is compatible with the FLUX architecture, we've also seen people personalizing the model to keep products or characters consistent across shots. This is useful in e-commerce and fashion industry. We allow easy training in our website — we labeled it Krea 1 — to do this, but the idea with this release is to encourage people with local rigs and more powerful GPUs to be able to tweak with LoRAs themselves too.
- Then I've seen fascinating use-cases such as UI/UX designers who prompt the model to create icons, illustrations, and sometimes even whole layouts that then they use as a reference (like Pinterest) to refine their designs on Figma. This reminds me of people who have a raster image and then vectorize it manually using the pen tool in Adobe Illustrator.
We also have seen big companies using it for both internal presentations and external ads across marketing teams and big agencies like Publicis.
yoo i'm also a researcher on the krea 1 project and happy to answer any questions :)
dvrp · 6h ago
For Dang or HN-mods:
I noticed that the URL for this submission is wrong: I tried to submit the correct URL (https://www.krea.ai/blog/flux-krea-open-source-release) but, for some reason, the submission gets flagged as duplicated and then I can only find this item which has a URL to our old blog post.
In the mean time, I'll setup a server-side redirect from the old blog post to our new one, but it would be nice to fix the link and I don't think I can do it on my side.
Our software follows canonical links when it finds them.
I've fixed the link above now (and rolled back the clock on the submission, to make up for lost time) but you might want to fix this for future pages.
dvrp · 22m ago
OMG. Thank you! I had to setup a CDN-level redirect and I was so confused as to why when I asked others to help, their submissions were flagged as [dupe] or [dead]
Thank you so much! I knew that HN software was advanced, but I didn’t know you guys used Canonical URLs like Google does. Smart and thanks for helping us with this slip!!!
Does this have any application for generating realistic scenes for robotics training?
I’m the Co-founder and CTO of Krea. We’re excited because we wanted to release the weights for our model and share it with the HN community for a long time.
My team and I will try to be online and try to answer any questions you may have throughout the day.
From a business point of view, there are many use-cases. Here's a list in no particular order:
- You can quickly generate assets that can be used _alongside_ more traditional tools such as Adobe Photoshop, After Effects, or Maya/Blender/3ds Max. I've seen people creating diffuse maps for 3D using a mix of diffusion models and manual tweaking with Photoshop.
- Because this model is compatible with the FLUX architecture, we've also seen people personalizing the model to keep products or characters consistent across shots. This is useful in e-commerce and fashion industry. We allow easy training in our website — we labeled it Krea 1 — to do this, but the idea with this release is to encourage people with local rigs and more powerful GPUs to be able to tweak with LoRAs themselves too.
- Then I've seen fascinating use-cases such as UI/UX designers who prompt the model to create icons, illustrations, and sometimes even whole layouts that then they use as a reference (like Pinterest) to refine their designs on Figma. This reminds me of people who have a raster image and then vectorize it manually using the pen tool in Adobe Illustrator.
We also have seen big companies using it for both internal presentations and external ads across marketing teams and big agencies like Publicis.
we prepared a blogpost about how we trained FLUX Krea if you're interested in learning more: https://www.krea.ai/blog/flux-krea-open-source-release
I noticed that the URL for this submission is wrong: I tried to submit the correct URL (https://www.krea.ai/blog/flux-krea-open-source-release) but, for some reason, the submission gets flagged as duplicated and then I can only find this item which has a URL to our old blog post.
In the mean time, I'll setup a server-side redirect from the old blog post to our new one, but it would be nice to fix the link and I don't think I can do it on my side.
I've fixed the link above now (and rolled back the clock on the submission, to make up for lost time) but you might want to fix this for future pages.
Thank you so much! I knew that HN software was advanced, but I didn’t know you guys used Canonical URLs like Google does. Smart and thanks for helping us with this slip!!!