Cheers for such a helpful tutorial, had tried to do a local install of FLUX on my Mac M1 but could not get it to work. Then I implemented the use of Hyper 8 Lora with FLUX and to my amazement it works rather well with 10 steps! Still takes around 10 minutes per render but quality is great : )
Thanks so much for another wonderful tutorial, especially on the more advanced features of DT. My image generations of XL dropped by half after using a 30 step process, changing out to an 8 step.
to generate 20 images, would it be better to set batch size to 4 and generate 5 sets or set batch size to 1 and generate 20 sets. I've noticed you always keep your batch size on 1.
Hi, is there a way to face swap in drawthings? Like reactor or roop in stable diffusiom, I tried ip adapter plus face and full face, it doesn’t fit in right.
Is this "CosmicDynavision" model one you installed or mixed yourself? I can't find it but it looks great. Thanks for the help. I'm messing with SD3 and the initial output was... blobs... but after watching this and another tutorial the output is MUCH better (albeit not as good as yours yet)
There's the Hyper 8 steps CFG-preserved (~5-8 guidance scale). "Preserved" not as the same, but you can use the negative prompt to some extent.
Cheers for such a helpful tutorial, had tried to do a local install of FLUX on my Mac M1 but could not get it to work. Then I implemented the use of Hyper 8 Lora with FLUX and to my amazement it works rather well with 10 steps! Still takes around 10 minutes per render but quality is great : )
big thanks! ✌
can you do SD video on DT next?
Thanks so much for another wonderful tutorial, especially on the more advanced features of DT. My image generations of XL dropped by half after using a 30 step process, changing out to an 8 step.
Thanks for the great tutorial!!
to generate 20 images, would it be better to set batch size to 4 and generate 5 sets or set batch size to 1 and generate 20 sets. I've noticed you always keep your batch size on 1.
maybe inpainting with the face loaded in mood board and using one of the control thingies that has "face" in the name?😊
Hi, is there a way to face swap in drawthings? Like reactor or roop in stable diffusiom, I tried ip adapter plus face and full face, it doesn’t fit in right.
Is this "CosmicDynavision" model one you installed or mixed yourself? I can't find it but it looks great.
Thanks for the help. I'm messing with SD3 and the initial output was... blobs... but after watching this and another tutorial the output is MUCH better (albeit not as good as yours yet)
It is a mix of Cosmicman and DynaVision... I have a recent video about model mixing in DrawThings.
great video! thank you for sharing
Also how did you get the generated time at the bottom of Draw things? thanks.
that just happens... there is a console that scripts can log to... although that rendertime doesn't stay visible for very long.