thank you for the clear and useful explanations concerning the proposed tool. Thank you for all the topics you suggest to us. serious and interesting work. best wishes.
Upscayl does NOT use any AI when upscaling, like within Foocus, SDWebUI, ComfyUI, Midjourney, or Magnific. It simply uses the same models when choosing to upscale-by-model. There are no samplers (like Euler, DPM++ Karras i.e. SAI diffusers) within Upscayl to be able to.
If you need a new Fooocus video idea, try refiner ( if not already done ). I did not see any use of the refiner, but now with vastly different SDXL models it is really useful. For example, you can start with an anime model, and switch to ultra realistic. Or start with an ultra realistic and switch to a painting model.
Hey, thanks for the awesome video. I've noticed that the RealESRGAN_General_WDN_x4_v3 custom model often gives me better results than the default photo model. I'm not really sure why, though.
You can create a batch background remover in comfy I have a workflow I can share. It's really easy. There are a few tutorials as well on remove background batch images
There is discussion about that and I think it might be something that might happen in the future but not sure. With the new ability to drop an image into the upscale/vary section of Fooocus and have it autogenerate a prompt based on the image I think that would be something that could be adapted to make batch upscaling work.
@@KLEEBZTECH or they could take the metadata embedded in the image to do that, actually there is a fork of fooocus named fooocus MRE, That actually support multi image upscale, but it I had issues running it on colab, and especially when replacing the default models I get a lot of error messages, it has a lot of bugs
@@KLEEBZTECH Bro your videos are always great and well explained, thank you so much, I really hope one day you make a video explaining how to run SD1.5 models and loras/embeddings in fooocus
As for SD 1.5 I might make a quick video covering the basics but I don't use 1.5 very often so as far as getting results comparable to pure 1.5 vs how Fooocus does it I probably can't go into a lot of detail. But I will play around with it and try to put something together.
thank you for the clear and useful explanations concerning the proposed tool.
Thank you for all the topics you suggest to us. serious and interesting work. best wishes.
You're very welcome!
Upscayl does NOT use any AI when upscaling, like within Foocus, SDWebUI, ComfyUI, Midjourney, or Magnific. It simply uses the same models when choosing to upscale-by-model. There are no samplers (like Euler, DPM++ Karras i.e. SAI diffusers) within Upscayl to be able to.
If you need a new Fooocus video idea, try refiner ( if not already done ). I did not see any use of the refiner, but now with vastly different SDXL models it is really useful. For example, you can start with an anime model, and switch to ultra realistic. Or start with an ultra realistic and switch to a painting model.
Great video! First time I've heard of this.
Thanks. It's not as good as some of the paid ones but it is potentially the best free one I've found.
Thank you. This programme works on gtx 750ti.
Hey, thanks for the awesome video. I've noticed that the RealESRGAN_General_WDN_x4_v3 custom model often gives me better results than the default photo model. I'm not really sure why, though.
Thanks for the info!
Because the model has been trained differently.
we need a batch background remover now!
Do you have a favorite background remover?
You can create a batch background remover in comfy I have a workflow I can share. It's really easy. There are a few tutorials as well on remove background batch images
Were you able to solve this problem? When I upgrade multiple images, they are saved in a random folder, which wastes a lot of time.......
I downloaded a few weeks ago for Mac but it doesn’t really sharpen when upscaling but blurs edges and adds a bit of painterly feel in my experience
I have found it to be good for some images and not so good for others and also depends on the model used. I don't use it for major upscaling though.
Thanks! Do you know a free alternative like this to remini ? for low resolution human portraits
Is there any ai wich converts bitmap to vector?
What program do you use to display the comparison of the digital art where you have three images on the screen in separate windows @ 5:14?
Just the Windows Photo app.
There is no update option?
I use version 2.9.5
Thanks
Not that I have seen. Probably need to download the latest.
Hi, What are the best settings for images created with artificial intelligence programs?
I feel it comes down to style of image more than where it comes from. I personally try a few on each image and none seem perfect for everything.
@ 14:20 did you mean 600 MB?
yup.
can you give the drive link of this photo 04:54 its pretty good for desktop wallpaper
drive.google.com/drive/folders/1AzWAsrkLL5kfbVcb6WXPPj5LMKsYIELd?usp=sharing
Hi, it is possible resize an image from 72dpi to 300dpi. Thank you......the best😊
That would not really be resizing the image. You can use many tools such as GIMP to change the DPI.
@@KLEEBZTECH sorry, I meant to change the resolution for printing and the use Upscayl. Sorry again
We need patch upscale in fooocus or using the same upscaler of fooocus but to do batch upscale and preferably on colab
There is discussion about that and I think it might be something that might happen in the future but not sure. With the new ability to drop an image into the upscale/vary section of Fooocus and have it autogenerate a prompt based on the image I think that would be something that could be adapted to make batch upscaling work.
@@KLEEBZTECH or they could take the metadata embedded in the image to do that, actually there is a fork of fooocus named fooocus MRE, That actually support multi image upscale, but it I had issues running it on colab, and especially when replacing the default models I get a lot of error messages, it has a lot of bugs
@@KLEEBZTECH Bro your videos are always great and well explained, thank you so much, I really hope one day you make a video explaining how to run SD1.5 models and loras/embeddings in fooocus
As for SD 1.5 I might make a quick video covering the basics but I don't use 1.5 very often so as far as getting results comparable to pure 1.5 vs how Fooocus does it I probably can't go into a lot of detail. But I will play around with it and try to put something together.
Can you tell me which is better? Topaz Photo Ai or this Upscale?
Topaz is 300 bucks
I couldn't give a direct comparison but Topaz is not cheap.
Hey how to download upscaled image cause I don't find any option for it
If you don't set an output folder the image will be saved to the folder that the original file is located in.
@@KLEEBZTECH yup thx fixed it 👍🏻
taaaaaaaaaaaanks
“Computer enhance”
not good enough models like Magnific AI especially with low quality images upscayl can't do anything
Magnific is certainly better but also very very expensive.
@@KLEEBZTECH i know there is also Supir but need 16 GB vram
Have they gotten to work with just 16GB? Last I had checked it seemed to be more.
But Fooocus upscaling is much better.
Yes. But you can only upscale with that so much. Then I use this.
Harris Amy Johnson Shirley Jones Kimberly