Spaces:
Running on Zero
Qwen image to image (image edit) model when?
Hi there!. Your space is the absolute best on HF hands down. No one has done like you have.
I was wondering when we'll get more image edit models. You already put flux2 reference edit, which is good. So I'm wondering if we'll soon get Qwen. Or better yet Krea2 reference edit. Having Krea2 would be absolutely sick!.. but let me know if you have it in mind and if you'll add it in the next update. Thanks.
Hi there! Thank you so much for the kind words!
The features you mentioned above will be implemented very soon. Stay tuned for the next update!
You did it!!. Thanks so so much. You're the absolute best.
Do you have a discord channel?
Thank you so much! Glad you like the update!
Currently, I don't have a Discord channel. Feel free to reach out or leave your suggestions right here in HF Discussions anytime!
Thank you!!.. you're a legend
Do you have any future plans for video generation?. Something like your imagegen but instead of lots image models options , it's video models options..
Do you have any plans like this? Is it something you've been thinking about doing?
Thank you so much for the kind words!
Actually, in my Comfy web UI GitHub repository (https://github.com/RioShiina47/comfy-webui), besides the ImageGen module, there are also ImageEdit, VideoGen, and 3DGen modules. The VideoGen module already integrates a large number of open-source video generation models and dynamic chains. Feel free to check it out if you're interested!
As for bringing a VideoGen Space to Hugging Face: video generation requires significantly more compute than image generation. Under HF's limited ZeroGPU duration, implementing it here would require simplifying it into rigid, fixed pipelines. That wouldn't be much different from other existing standalone VideoGen Spaces, so I currently don't have plans to build a VideoGen Space on HF.
ImageGen is different—thanks to the relatively sufficient ZeroGPU duration for image tasks, you can unleash your creativity by freely combining 5 major tasks, 30 model architectures, and various dynamic chains into all kinds of unique workflows. That's what makes it truly fun and special!
Thanks again for your support!