Something that I think might be a recent addition to Comfy (that or I'm unobservant and hadn't noticed it before), which was speed this up a tiny bit is to right click the KSampler node and you get the option to add a Hi-res fix. Adds the nodes and connections for everything, bar the output for the new KSampler.
Try the node called "Ksampler Efficient Fooocus" it has a script thing you can attach a high res fix node too. Its not like exactly like you mentioned here ( i too agree that would be sick) but its a really fast and easy way to get the hi res fix on
Really good. I prefer these simpler workflows. A lot of the complex workflows take a day to understand and still might not work. Thanks. Should mention this WF doesn't seem to work with facial images. I'm getting duplicate eyes etc
@@andrewcampbell8938 Not a problem another thing that might help i realized i forgot is maybe try tiling the upscaled part to be within SDXL resolution requirements too
![ฮักเทื่อสุดท้าย - บอย ศิริชัย [OFFICIAL MV]](http://i.ytimg.com/vi/-1PoPoQ-B-k/mqdefault.jpg)
Didn't know we could organize the models and LoRAs into subfolders! Neat trick.
Welcome to TH-cam Colvin!
Thanks for the tutorial.
Something that I think might be a recent addition to Comfy (that or I'm unobservant and hadn't noticed it before), which was speed this up a tiny bit is to right click the KSampler node and you get the option to add a Hi-res fix. Adds the nodes and connections for everything, bar the output for the new KSampler.
Try the node called "Ksampler Efficient Fooocus" it has a script thing you can attach a high res fix node too. Its not like exactly like you mentioned here ( i too agree that would be sick) but its a really fast and easy way to get the hi res fix on
Welcome, great job!
Awesome work
Very good video. How do you make the connections between the nodes in a straight line? It looks more organized. Thank you very much
Really good. I prefer these simpler workflows. A lot of the complex workflows take a day to understand and still might not work. Thanks. Should mention this WF doesn't seem to work with facial images. I'm getting duplicate eyes etc
Turn down noise or add a controlnet or ipadaptor
@@JustArtsCreations turning down noise gives a blocky mess and doesn't work. will try ipadapter. thx
@@andrewcampbell8938 Not a problem another thing that might help i realized i forgot is maybe try tiling the upscaled part to be within SDXL resolution requirements too
Thanks for sharing!
What do I need if I want the image not to change but only to be detailed? noise reduction causes very poor results.
Looking to hire someone to build us a workflow in ComfyUi, Anyone have any suggestions?