This workflow is really cool to play around with. Makes me want to learn more about workflows, because they seem to be really powerful. Are there any other checkpoints you recommend? or is juggernaut the best?
The Juggernaut XL Lightning version is indeed impressive in terms of speed, though the quality might not always be top-notch. Personally, I’m a fan of using Albedobase combined with a LoRA like "Midjourney Mimic". It’s worth experimenting with different setups to find what works best for your needs!
Thanks so much for the mention in your video! Your V2 Workflow with Lama remover & BiRefNet is very nice 😀😀. I've been playing with my idea of compositing several elements in the image. I tried some methods, Attention Masking by Matteo and with the DenseDiffusion method (Huchenlei). Both methods work with large objects (that occupy a big space of the image) but with small elements they have difficulty. On the other hand, I've been testing Flux for product photography and it's amazing! I think you could easily integrate Flux in the future into your work (maybe when controlnet for flux is more stable). Thank you very much for sharing your work!
Thank you for your feedback! ControlNet for Flux is indeed working great now. I've integrated it into a new workflow for background replacement. You can check it out here: openart.ai/workflows/myaiforce/UdbHePrLFEP9WzdrmmFj. Feel free to give it a try and let me know what you think!
For tilted product images, it’s hard to get them at the right angle. My cake is about to topple over, haha! I kept trying until I realized it might be because there’s no depth map? I'm a beginner; I only know how to use... but I can't make reasonable adjustments to it...
When I see the final output from ComfyUI, it's far from perfect. Not only does it take too much time, but the results are also not as good as Photoshop editing. For example, where are the product shadows?
Thanks for checking out the video! I know the workflow might seem complex at first, but that's because it’s designed to be incredibly powerful and flexible. Once you get the hang of it, you'll see just how much it can do. I’d love to hear your thoughts if you decide to give it a try!
Glad you integrated the lama remover 😎😎 Great job !
Thanks for the suggestion.🙏
This is pretty amazing. Thanks for explaining how to do it and awesome workflow
Glad you like it!
This workflow is really cool to play around with. Makes me want to learn more about workflows, because they seem to be really powerful. Are there any other checkpoints you recommend? or is juggernaut the best?
The Juggernaut XL Lightning version is indeed impressive in terms of speed, though the quality might not always be top-notch. Personally, I’m a fan of using Albedobase combined with a LoRA like "Midjourney Mimic". It’s worth experimenting with different setups to find what works best for your needs!
Thanks so much for the mention in your video! Your V2 Workflow with Lama remover & BiRefNet is very nice 😀😀. I've been playing with my idea of compositing several elements in the image. I tried some methods, Attention Masking by Matteo and with the DenseDiffusion method (Huchenlei). Both methods work with large objects (that occupy a big space of the image) but with small elements they have difficulty. On the other hand, I've been testing Flux for product photography and it's amazing! I think you could easily integrate Flux in the future into your work (maybe when controlnet for flux is more stable). Thank you very much for sharing your work!
Thank you for your feedback! ControlNet for Flux is indeed working great now. I've integrated it into a new workflow for background replacement. You can check it out here: openart.ai/workflows/myaiforce/UdbHePrLFEP9WzdrmmFj. Feel free to give it a try and let me know what you think!
I can't install, keeps failing : ComfyUI-BiRefNet-ZHO, ComfyUI-BiRefNet
Same here. I installed the Hugo package instead, but don't know how to switch from ZHO to Hugo.
For tilted product images, it’s hard to get them at the right angle. My cake is about to topple over, haha! I kept trying until I realized it might be because there’s no depth map? I'm a beginner; I only know how to use... but I can't make reasonable adjustments to it...
Maybe if there were pictures it would be easier for us to understand your situation.
thanks for this video Wei.
My pleasure!
When I see the final output from ComfyUI, it's far from perfect. Not only does it take too much time, but the results are also not as good as Photoshop editing. For example, where are the product shadows?
still cannot download and install zhozho. already install but it still said zho zho biref is missing
. please help
I switched to BiRefNet-Hugo.
I installed birefNet-Hugo instead. Works pertectly !
Well, you lost me at 50 seconds as soon as I saw that insane diagram workflow
Thanks for checking out the video! I know the workflow might seem complex at first, but that's because it’s designed to be incredibly powerful and flexible. Once you get the hang of it, you'll see just how much it can do. I’d love to hear your thoughts if you decide to give it a try!